SQLite Database Export PDF

SQLite Database Export 1 | 1312059809029161033 | 1253625758832660541 | edau87 | Eldi is up to you brother. | 2024-12-11 09:48:23 | None 2 | 1305272456449228863 | 1253625758832660541 | edau87 | That's really cool great work | 2024-12-11 09:48:23 | None 3 | 1305272309346603108 | 1253625758832660541 | ethan83_ | yes had to implement genetic optimixation was to many tests to do lol | 2024-12-11 09:48:23 | None 4 | 1305272214773301248 | 1253625758832660541 | edau87 | A workflow generator? | 2024-12-11 09:48:23 | 1305272105449033840 5 | 1305272105449033840 | 1253625758832660541 | ethan83_ | | 2024-12-11 09:48:23 | None 6 | 1305268140787105935 | 1253625758832660541 | coensio | Maybe you overthink what others think too much. In my experience, you can have a meaningful discussion with only about 5% of people at most;) | 2024-12-11 09:48:23 | 1305265347003285545 7 | 1305265347003285545 | 1253625758832660541 | edau87 | But you are more forward than I was lol. I wanted to say that but held back. | 2024-12-11 09:48:23 | None 8 | 1305265208842911816 | 1253625758832660541 | edau87 | The algotime pro chat my bad. | 2024-12-11 09:48:23 | None 9 | 1305263691234676787 | 1253625758832660541 | edau87 | Steve257 in the main chat. | 2024-12-11 09:48:23 | None 10 | 1305255834636521623 | 1253625758832660541 | coensio | do we have traders that try to trade with $1500? | 2024-12-11 09:48:23 | 1305208432554541117 11 | 1305208432554541117 | 1253625758832660541 | edau87 | I'm probably going to insist upon at least $5k to $10k in trading funds on future applications. Trying to make day trading work with $1500 is almost impossible and I feel bad. | 2024-12-11 09:48:23 | None 12 | 1301301777576099850 | 1253625758832660541 | edau87 | If I use those blocks (which I barely do) I do one regime at a time. | 2024-12-11 09:48:23 | None 13 | 1301301664413913150 | 1253625758832660541 | edau87 | Maybe? You have too many degrees of freedom there. You aren't going to take every trade in every one of those regimes. | 2024-12-11 09:48:23 | None 14 | 1301262161737289849 | 1253625758832660541 | ethan83_ | just an idea it should find the best correct? so value of indicator and also choosing the best indicator if using or? | 2024-12-11 09:48:24 | None 15 | 1300864044105863209 | 1253625758832660541 | ethan83_ | first priliminary test to low results getting 234 strategyes with corrolation = 700 trades | 2024-12-11 09:48:46 | None 307 | 1283117362287087648 | 1253625758832660541 | coensio | now it is very easy to get 100% SR in all runs so will be hard to compare which idea is better if they all have 100%, we need to come up with a much more cleaver rating.. | 2024-12-11 09:48:46 | 1283113876497629339 308 | 1283116381297971210 | 1253625758832660541 | coensio | should be possible with custom analysis....but would need some coding | 2024-12-11 09:48:46 | 1283100799903207435 309 | 1283113919547703389 | 1253625758832660541 | ethan83_ | updated | 2024-12-11 09:48:46 | None 310 | 1283113876497629339 | 1253625758832660541 | ethan83_ | | 2024-12-11 09:48:46 | None 311 | 1283108698058653790 | 1253625758832660541 | ethan83_ | Here is the idea | 2024-12-11 09:48:46 | None 312 | 1283101780871479336 | 1253625758832660541 | ethan83_ | well crossing my fingers it works | 2024-12-11 09:48:46 | None 313 | 1283101734381813810 | 1253625758832660541 | ethan83_ | stupid me ofc i just choose to only store 100 and it should do it automatic | 2024-12-11 09:48:46 | None 314 | 1283100799903207435 | 1253625758832660541 | ethan83_ | wish there was away of filtering like selecting 100 best opf that databank regarning this parameters | 2024-12-11 09:48:46 | None 315 | 1283100304791048337 | 1253625758832660541 | ethan83_ | along wth min 40 trades per year | 2024-12-11 09:48:46 | None 316 | 1283100259811196980 | 1253625758832660541 | ethan83_ | im using weighted so im using stagnation , retdd , avgtrade and stability | 2024-12-11 09:48:46 | None 317 | 1283100137262157907 | 1253625758832660541 | ethan83_ | 10 min running building phase 1 | 2024-12-11 09:48:46 | None 318 | 1283099161637425275 | 1253625758832660541 | ethan83_ | testing it now with your new updated version | 2024-12-11 09:48:46 | None 319 | 1283082795161223270 | 1253625758832660541 | coensio | nope | 2024-12-11 09:48:47 | None 320 | 1283074977368637501 | 1253625758832660541 | ethan83_ | what if using filters instead using time like run for 1 hour without filter it does automatic replaces with better strategies if choosing for example ret/dd amyone tried that approach? | 2024-12-11 09:48:47 | None 321 | 1283019571539873873 | 1253625758832660541 | coensio | ok sent him to your page | 2024-12-11 09:48:47 | 1282983447186505783 322 | 1282999380160286791 | 1253625758832660541 | edau87 | Just try things over and over, notating each time like in the examples I shared or otherwise. | 2024-12-11 09:48:47 | 1282988943545073746 323 | 1282990911654985749 | 1253625758832660541 | coensio | trial and error | 2024-12-11 09:48:47 | None 324 | 1282988943545073746 | 1253625758832660541 | ethan83_ | still the issue is conditions not sure how to find god and bad ones and thresholds | 2024-12-11 09:48:47 | None 325 | 1282988791359078401 | 1253625758832660541 | ethan83_ | finally my xml creator works with exit times and limit trade times | 2024-12-11 09:48:47 | None 326 | 1282988711499403274 | 1253625758832660541 | ethan83_ | | 2024-12-11 09:48:47 | None 327 | 1282983447186505783 | 1253625758832660541 | edau87 | Nope! Just one. I spoke with him a bit and screened him. | 2024-12-11 09:48:47 | 1282959786454810697 328 | 1282965063606472807 | 1253625758832660541 | coensio | here's the updated version with updated metrics overview, it also displays NP and DD columns for each single selection criterion/run. Probably best to zoom your browser window to 70% to get everything in one view. | 2024-12-11 09:48:47 | None 329 | 1282959786454810697 | 1253625758832660541 | coensio | sure so there are multiple traders there? | 2024-12-11 09:48:47 | 1282889374375350312 330 | 1282889374375350312 | 1253625758832660541 | edau87 | I have an applicant who is incubating many strategies and watching them trade US stocks on TradeStation. They are interested in futures and have been using SQX for a long time but have mostly just played around and gotten to know the settings and how they work rather than why they work or that some work better than others. They use simple, self-developed excel solutions to track and evaluate workflow and asset performance but want to learn and play with Python as well. Crucially, they know ChatGPT. All in all, they seem like a beginner but not in a beginner's beginner sense. More like they have all the basics down and know how to take strategies live on our platforms but are looking for guidance with how to test their own workflow ideas. I'd support their addition if you agree. | 2024-12-11 09:48:47 | None 331 | 1282774871163666442 | 1253625758832660541 | coensio | soon | 2024-12-11 09:48:47 | None 332 | 1282774786270953603 | 1253625758832660541 | coensio | update coming... | 2024-12-11 09:48:47 | None 333 | 1282261060648173621 | 1253625758832660541 | coensio | now we are still in beta phase just experimenting to see what is possible | 2024-12-11 09:48:47 | None 334 | 1282260971527606293 | 1253625758832660541 | coensio | I think that is exactly what mitch is doing, as he tries to auto-loop through almost everything, but his work is still unfinished...maybe we can later combine ideas and integrate it all to one single solution. | 2024-12-11 09:48:47 | 1282250331081801769 335 | 1282252276815233087 | 1253625758832660541 | ethan83_ | and 1 more thing with the amount of # trades the issue can be finding god strategies with high avg.trade and ret/dd atleast after all my searching i understand sucessrate is high. | 2024-12-11 09:48:47 | None 336 | 1282250399201366017 | 1253625758832660541 | ethan83_ | this course have turned me to something else haha | 2024-12-11 09:48:47 | None 337 | 1282250331081801769 | 1253625758832660541 | ethan83_ | Yes this is easier to see i like it aswell great work. Im using it now with my combined workflow pythin program too see on my verry best filters and combinations and see what i will get:) , Sitting and thinking about if this can be used somehow to get best combinations of filters. And how i can develop it further creating workflows is awsome and my head is already spining hehe | 2024-12-11 09:48:47 | None 338 | 1282249128176390147 | 1253625758832660541 | coensio | yes but this is % based approach without graphical distribution analysis (which is still better imho) | 2024-12-11 09:48:48 | 1282248029163753573 339 | 1282248029163753573 | 1253625758832660541 | ethan83_ | ahh i see , But point is to use this thresholds as selected filter to get highest sucessrate just like in the course correct if so i understand and its awsome | 2024-12-11 09:48:48 | None 340 | 1282246102493691925 | 1253625758832660541 | coensio | so yes in this case you have larger amount of passing strategies but since SR is calculated per TOOS you have worst success rate | 2024-12-11 09:48:48 | None 341 | 1282245932037177410 | 1253625758832660541 | coensio | you filter each metric of each workflow run using >80th percentile of the given metric distribution. So in case of DD, you the selected IS strategies for each run and TOOS passing strategies numbers, for first TOOS you selected 61 and got 57 with NP >-=0 this equals to SR=93%. | 2024-12-11 09:48:48 | None 342 | 1282244074220552223 | 1253625758832660541 | ethan83_ | i have 1080 is strategies if we check row 1 # trades i get 100 pass in oos 89 with run 1 and 11 with run 2 and is it with 1697 trades insample? but with drawdown as filter i get more strategies passing? but not as high sucessrate. So with this i would use 1697 trades as selecting filter ? | 2024-12-11 09:48:48 | None 343 | 1282241690865766420 | 1253625758832660541 | coensio | which part is unclear? | 2024-12-11 09:48:48 | None 344 | 1282096205094060032 | 1253625758832660541 | ethan83_ | results 2 runs not quite understand this atm, http://127.0.0.1:8050/.. | 2024-12-11 09:48:48 | None 345 | 1282095526455672844 | 1253625758832660541 | ethan83_ | | 2024-12-11 09:48:48 | None 346 | 1282076117435678851 | 1253625758832660541 | ethan83_ | So it basicly creates workflows with different tooos we run workflows it save ins and oos and it then loads results and shows what values to use as ranking to get as many as possible to pass toos ? , brilliant | 2024-12-11 09:48:48 | None 347 | 1282069726612295750 | 1253625758832660541 | ethan83_ | and with oos i remeber i had to create multiple workflows with didicated dates so correct some bug | 2024-12-11 09:48:48 | None 348 | 1282068930780991629 | 1253625758832660541 | ethan83_ | cool stuff | 2024-12-11 09:48:48 | None 349 | 1282068912023932961 | 1253625758832660541 | ethan83_ | Perfect im running 2 workflows now and saving data to ins and oos folder | 2024-12-11 09:48:48 | None 350 | 1282054422465220703 | 1253625758832660541 | coensio | try this one as base Workflow | 2024-12-11 09:48:48 | None 351 | 1282051921217065133 | 1253625758832660541 | coensio | but do you seen datarange shift in 2nd workflow generated? | 2024-12-11 09:48:48 | None 352 | 1282051725217366160 | 1253625758832660541 | ethan83_ | yep belive i had that issue aswell | 2024-12-11 09:48:48 | None 353 | 1282051658666479690 | 1253625758832660541 | coensio | no thats a bug, we need to remove OOS from the base workflow for now | 2024-12-11 09:48:48 | 1282050614385447012 354 | 1282051125708853278 | 1253625758832660541 | ethan83_ | this is the base | 2024-12-11 09:48:48 | None 355 | 1282050614385447012 | 1253625758832660541 | ethan83_ | should there be this much oos data? i created 2 workflows this is the second | 2024-12-11 09:48:49 | None 356 | 1282030921284976741 | 1253625758832660541 | edau87 | Yes this sounds like the right move. | 2024-12-11 09:48:49 | 1282023905070022727 357 | 1282024190261723261 | 1253625758832660541 | coensio | I owe you 1 more commission if I count correctly. | 2024-12-11 09:48:49 | None 358 | 1282024092551217173 | 1253625758832660541 | coensio | Price will increase next week, but I need to change few things first. | 2024-12-11 09:48:49 | None 359 | 1282023905070022727 | 1253625758832660541 | coensio | BTW: I added this note to the sales page: "Before making a purchase, please ensure you complete our application process first. Failure to do so may result in your application being declined." As I am unable to manage the overwhelming number of basic questions in my DMs :/ | 2024-12-11 09:48:49 | None 360 | 1282022002278011051 | 1253625758832660541 | coensio | 1. See attached zip file, you can unzip it on C:\ 2. Install all required python libs 3. Run it using: python CoensioWorkflowValidationV03.py 4. Dashboard interface is at this address: http://127.0.0.1:8050/ - Example basic workflow which is compatible with this tool is located in BaseWorkflow directory. But you can use your workflows but then you need to meet requirements as defined on tab 1. You can also directly upload your IS/TOOS DatabankExport.csv files into the corresponding directories and skip tab 1. In this tool we are using selection based on percentile of distribution of each separate workflow run. Risk: if you make too many iterations with all workflows included, you will turn your TOOS into IS. So the best is to test/define your selection criterion idea using 1 or 2 workflow runs and then add another 3 or 2 to test this selection idea. Please share your findings. | 2024-12-11 09:48:49 | None 361 | 1282021638510088273 | 1253625758832660541 | edau87 | I am interested in beta'ing the tool but I'm still in Italy | 2024-12-11 09:48:49 | None 362 | 1280860752130867262 | 1253625758832660541 | coensio | yes we will do some collabs to cover this topic as I get many similar requests...soon... | 2024-12-11 09:48:49 | 1280843787408507031 363 | 1280843787408507031 | 1253625758832660541 | ethan83_ | Yep atm its crazy much todo bought a new house selling the old, home with kids to January when i start working again. Still thinking about new ideas to test and find new trading ideas. still thinking about testing your way aswell with templates so on my todo list. and cant wait for that colibration with java. And still using seq optimization after simple optimization only thing i want is to get stable zones dont care if it get better ot worse just trying to avoid over omptimization :). | 2024-12-11 09:48:49 | None 364 | 1280809558704394271 | 1253625758832660541 | coensio | Will do very soon | 2024-12-11 09:48:49 | None 365 | 1280804793136189441 | 1253625758832660541 | edau87 | Haha oh really? Raise the price! Stop the bleeding! | 2024-12-11 09:48:49 | None 366 | 1280791491890909194 | 1253625758832660541 | coensio | you got 2 new guys in September already, 1 got refunded because of B2B transaction. | 2024-12-11 09:48:49 | None 367 | 1280502226615533620 | 1253625758832660541 | coensio | I am very busy guy so no hurry at all with this... | 2024-12-11 09:48:49 | None 368 | 1280501923555966999 | 1253625758832660541 | edau87 | I can take a look if you are okay with waiting! | 2024-12-11 09:48:49 | None 369 | 1280501178399981619 | 1253625758832660541 | coensio | "I can't handle being inundated with questions like, "Can I use templates?" "Why doesn't my mt5 match any thoughts?" "Why can't I use ninjatrader?" "why can't I use TradeStation data to trade dukascopy on mt5?" that's why we need to do some selection I can let you make a page with 'optimizepress' tool I work with it is as simple as as dragging some blocks and text into some places onscreen...Or you just prepare some content I will copy and paste it into a website. | 2024-12-11 09:48:49 | 1280500309868810262 370 | 1280500614400442471 | 1253625758832660541 | edau87 | Once I return, that is. | 2024-12-11 09:48:50 | None 371 | 1280500587934519358 | 1253625758832660541 | edau87 | I am about to leave for Italy for a few weeks, I bet ChatGPT and I can come up with a landing page or something. | 2024-12-11 09:48:50 | None 372 | 1280500309868810262 | 1253625758832660541 | edau87 | There isn't a need to sell high ticket courses, you're right. The price range you propose would hopefully filter out some of what we are seeing recently (people trying to trade the only $5,000 USD they have to their name...a recipe for disaster ). I can't handle being inundated with questions like, "Can I use templates?" "Why doesn't my mt5 match any thoughts?" "Why can't I use ninjatrader?" "why can't I use TradeStation data to trade dukascopy on mt5?" | 2024-12-11 09:48:50 | 1280496045792690297 373 | 1280496812096229398 | 1253625758832660541 | coensio | What I propose for referrals is to make some kind of a blog page or introduction website with some contact info attached to it. It can be an external page or it can be also hosted on my web-server coensio.com. So people who are interested could first reach out, tell us about their goals and we can check if there is a good fit. | 2024-12-11 09:48:50 | None 374 | 1280496201413955596 | 1253625758832660541 | coensio | Now they rush to the purchase page, and I get emails like: "I paid for the course where is my access?" | 2024-12-11 09:48:50 | None 375 | 1280496045792690297 | 1253625758832660541 | coensio | Yes, as I see it, the most recent influx of traders can be categorized as 'tourists' because we kept the entry price extremely low. That's one of the reasons people don't feel obligated to put in any effort. If they had paid $3,000 for the course, they would at least feel 'accountable' and obligated to do some work, such as testing the workflows and our methods. However, I promised myself that I would never sell 'high-ticket' courses, as I don't believe any course is worth that kind of money. Even $2,000 is too high, and $1,000 would be too much of a 'risk' for many. So, I think a fair price would be between $497 and $997. This would allow me and all referrals to provide more support, but only for dedicated traders...like a part-time job or so. Moreover I was thinking about making discord support based a recurring fee. Small monthly fee to keep tourists away, and keep only motivated guys in our group. But here is a catch, as you already proposed we need to have some kind of selection process. | 2024-12-11 09:48:50 | 1280264050617487402 376 | 1280264050617487402 | 1253625758832660541 | edau87 | I notice that the main purchase page has changed to serious applicants only. I think I'll pull the referral link from my profile and maybe add something about asking me about the course so I can pre-screen individuals. | 2024-12-11 09:48:50 | None 377 | 1280225266064560210 | 1253625758832660541 | coensio | no this doesnt work people are too lazy and this means even more work for moderators | 2024-12-11 09:48:50 | None 378 | 1280191730578227263 | 1253625758832660541 | edau87 | I honestly thought it was a great idea lol. | 2024-12-11 09:48:50 | None 379 | 1280191670880702524 | 1253625758832660541 | edau87 | Go back to giving out homework to unlock subsequent lessons? | 2024-12-11 09:48:51 | 1280182870085996665 380 | 1280182870085996665 | 1253625758832660541 | coensio | we need to rethink it | 2024-12-11 09:48:51 | None 381 | 1280182566896533568 | 1253625758832660541 | coensio | I do not see how we can support 100 traders at this kind of level in the current formula, this becomes a part time job now. | 2024-12-11 09:48:51 | None 382 | 1280178444961058910 | 1253625758832660541 | edau87 | so I feel like this is a whole new level of basic | 2024-12-11 09:48:51 | None 383 | 1280178405668552829 | 1253625758832660541 | edau87 | I think you're right. The Stat Oasis community can get pretty bad at times but right now I'm trying to explain why you can't use TradeStation data to build strategies on the Mt5 engine and trade them on NinjaTrader. | 2024-12-11 09:48:51 | 1280178189473415260 384 | 1280178220259475596 | 1253625758832660541 | coensio | | 2024-12-11 09:48:51 | None 385 | 1280178189473415260 | 1253625758832660541 | coensio | maybe thats why other courses ask $2000 | 2024-12-11 09:48:51 | None 386 | 1280177727525355592 | 1253625758832660541 | coensio | lol I think I told you before this is gonna look like that;) | 2024-12-11 09:48:51 | None 387 | 1280176587727175805 | 1253625758832660541 | edau87 | The moderator channel is going to turn in to the vent channel. I am not cut out for the trader education industry lol. | 2024-12-11 09:48:51 | None 388 | 1279549267115114538 | 1253625758832660541 | coensio | thanks, not 100% yet, but I will get there | 2024-12-11 09:48:51 | 1279547820046684293 389 | 1279547820046684293 | 1253625758832660541 | edau87 | Have a great time! Glad you're feeling better. | 2024-12-11 09:48:51 | None 390 | 1279547294068379699 | 1253625758832660541 | coensio | Okay, I need to get some sleep now. Tomorrow, I'm going to the mountains to clear my head a little bit after my last covid | 2024-12-11 09:48:51 | None 391 | 1279546632303542292 | 1253625758832660541 | edau87 | You sure did over deliver. My trading has never been better. I've got so many plans now. My expectation was to learn and develop myself. | 2024-12-11 09:48:51 | None 392 | 1279546552813224080 | 1253625758832660541 | coensio | but yeah people have always crazy expectations...but it's not my problem... | 2024-12-11 09:48:51 | None 393 | 1279546332524318742 | 1253625758832660541 | coensio | You should not attach any emotions to this....I sleep like a baby as I know I 10x over delivered this course and there is no other course like that. More over education always costs money...if someone cannot afford $297 he should not be here. | 2024-12-11 09:48:51 | 1279545364634140673 394 | 1279545523598262418 | 1253625758832660541 | edau87 | But I'm quickly losing my capacity to pity when people talk to me the way they do. I can't imagine the DMs you must get. | 2024-12-11 09:48:51 | None 395 | 1279545364634140673 | 1253625758832660541 | edau87 | Yes. It makes me feel bad because many can't afford to lose the money they will end up costing themselves while looking for shortcuts. | 2024-12-11 09:48:51 | None 396 | 1279545093254156358 | 1253625758832660541 | coensio | I told you most of them come here only for 'free' workflows, they just want to 'copy and paste' and make millions... | 2024-12-11 09:48:51 | 1279544562250944615 397 | 1279544562250944615 | 1253625758832660541 | edau87 | Looks like a very small account trader who wants a lot of answers. I hope he does the lessons his first question to me was why couldn't he access the workflows | 2024-12-11 09:48:51 | None 398 | 1279543902512087111 | 1253625758832660541 | coensio | | 2024-12-11 09:48:51 | None 399 | 1279543567387459706 | 1253625758832660541 | edau87 | I know. He's blowing me up lol. | 2024-12-11 09:48:52 | None 400 | 1279543520037699776 | 1253625758832660541 | coensio | new referral 'rozzi rossi' | 2024-12-11 09:48:52 | None 401 | 1278621457471115288 | 1253625758832660541 | coensio | can you tell your typical performance increase w/wo SPO? | 2024-12-11 09:48:52 | 1278597563674525717 402 | 1278597563674525717 | 1253625758832660541 | ethan83_ | I only use it to validate that its not over optimized, so i have a stable zone for indicators. | 2024-12-11 09:48:52 | None 403 | 1278368246168289331 | 1253625758832660541 | edau87 | Has the sequential optimization step improved true OOS at all? Remember the overall metrics of an IS test are basically meaningless. | 2024-12-11 09:48:52 | None 404 | 1278340039658569860 | 1253625758832660541 | ethan83_ | thi by ret/dd | 2024-12-11 09:48:52 | None 405 | 1278339858880008263 | 1253625758832660541 | ethan83_ | little special edition im adding the seq optimizatuion after simple omptimixation so this all are passing and still quite nice results. this sorted by Avg trade | 2024-12-11 09:48:52 | None 406 | 1277669563772965055 | 1253625758832660541 | edau87 | No worries at all I am sorry to hear you're still sick! | 2024-12-11 09:48:52 | None 407 | 1277646830041301076 | 1253625758832660541 | coensio | I will settle this, this week, I'm still pretty sick btw | 2024-12-11 09:48:52 | None 408 | 1277644399433744518 | 1253625758832660541 | edau87 | Excellent thank you! | 2024-12-11 09:48:52 | None 409 | 1277642765068533790 | 1253625758832660541 | coensio | otherwise it is generating to much API data and it hangs | 2024-12-11 09:48:52 | None 410 | 1277642481084928135 | 1253625758832660541 | coensio | two new referrals, the API link is changed to display referrals by month so for this month link is https://www.coensio.com/scripts/paypal/GetReferralsV02.php?RefCode=EDAU87&month=8 | 2024-12-11 09:48:52 | None 411 | 1275868573499396209 | 1253625758832660541 | edau87 | Thank you! | 2024-12-11 09:48:52 | None 412 | 1275835585499500601 | 1253625758832660541 | coensio | You have a new referral: zizzissimo | 2024-12-11 09:48:52 | None 413 | 1275466932614725703 | 1253625758832660541 | edau87 | Trading is not a very nice world at times. | 2024-12-11 09:48:52 | None 414 | 1275435038544101447 | 1253625758832660541 | coensio | But the issue was that access was completely free in exchange for performing certain tasks. However, scammers, scouts, and affiliates from other courses joined, turning it into one big mess. I ended up kicking many people out of the group. | 2024-12-11 09:48:52 | None 415 | 1275432308056522892 | 1253625758832660541 | coensio | Indeed, we had homework assignments with completion levels. We also had a few SQX team 'gurus' here, but they left, likely because they didnt want to do the homework. From like 50 traders only few of them completed all 7 levels, it was fun, and an enormous amount of work has been done, but you cannot build a community with 3 people. | 2024-12-11 09:48:52 | 1275421631275073536 416 | 1275421631275073536 | 1253625758832660541 | edau87 | I hear you used to give out homework by the way lol. | 2024-12-11 09:48:53 | None 417 | 1275421587587469345 | 1253625758832660541 | edau87 | Very interesting! | 2024-12-11 09:48:53 | None 418 | 1275404720583999523 | 1253625758832660541 | coensio | Yeah I remember him now, he joined one of my first SQX collaboration groups back in 2018, in general a good and very enthusiastic guy, no worries here | 2024-12-11 09:48:53 | None 419 | 1275394488776261694 | 1253625758832660541 | ethan83_ | Had been a value having him in discord | 2024-12-11 09:48:53 | None 420 | 1275394417888591974 | 1253625758832660541 | ethan83_ | Ahh hes a smart guy | 2024-12-11 09:48:53 | None 421 | 1275376593375723550 | 1253625758832660541 | coensio | The SQX world is smaller than we think | 2024-12-11 09:48:53 | None 422 | 1273740332315447347 | 1253625758832660541 | coensio | it's all about time and time is the most precious asset you have... | 2024-12-11 09:48:53 | 1273740053151092778 423 | 1273740053151092778 | 1253625758832660541 | ethan83_ | Im ready when you are my friend:) , Wish you had some more spare time for all this case studys i really wanna proceed my learning and knowledge | 2024-12-11 09:48:53 | None 424 | 1273739722174496851 | 1253625758832660541 | coensio | maybe this will come later as a group exercise...or case study | 2024-12-11 09:48:53 | 1273739510576054272 425 | 1273739510576054272 | 1253625758832660541 | ethan83_ | Any idea how in the heck to learn this? cant in my head even know where to begin sounds awsome then i hear Java | 2024-12-11 09:48:53 | 1272649563693383842 426 | 1273565856361742438 | 1253625758832660541 | edau87 | | 2024-12-11 09:48:53 | None 427 | 1273565639331811369 | 1253625758832660541 | coensio | He claims he did not, I need to check server logs.... | 2024-12-11 09:48:53 | 1273565530061541482 428 | 1273565530061541482 | 1253625758832660541 | edau87 | Woooow did he get access to the workflows? | 2024-12-11 09:48:53 | None 429 | 1273565403347681300 | 1253625758832660541 | coensio | I sent you commission for your first 'ghost' guy, the 2nd one opened a paypal ticket case against me, so this needs to be clarified. | 2024-12-11 09:48:53 | None 430 | 1272965805822181458 | 1253625758832660541 | coensio | exactly I going away from my PC till EOD | 2024-12-11 09:48:53 | 1272964888645468190 431 | 1272965321497772186 | 1253625758832660541 | ethan83_ | 18 + 18 + 2,5 small account | 2024-12-11 09:48:53 | None 432 | 1272964888645468190 | 1253625758832660541 | ethan83_ | time to not watch | 2024-12-11 09:48:53 | None 433 | 1272964831342891103 | 1253625758832660541 | ethan83_ | holy smokes + 1550 so far | 2024-12-11 09:48:53 | None 434 | 1272964232828424222 | 1253625758832660541 | edau87 | 30's is wonderful though. | 2024-12-11 09:48:53 | None 435 | 1272964183813521408 | 1253625758832660541 | edau87 | Funny you say that because I'm starting to see unacceptable OOS degradation as I get this high. | 2024-12-11 09:48:53 | 1272964069837766697 436 | 1272964069837766697 | 1253625758832660541 | coensio | what was the story about Icarus again? | 2024-12-11 09:48:53 | 1272963742271017023 437 | 1272963742271017023 | 1253625758832660541 | edau87 | I do! lol But yes, I'm into the 40's on Ret:DD and it can be addictive. | 2024-12-11 09:48:53 | 1272961463895720022 438 | 1272962553118457877 | 1253625758832660541 | coensio | no idea, this is something new | 2024-12-11 09:48:53 | 1272962365943582820 439 | 1272962427834732701 | 1253625758832660541 | coensio | CL is much harder than GC, and from my experience and only short side | 2024-12-11 09:48:53 | None 440 | 1272962365943582820 | 1253625758832660541 | ethan83_ | only issue is my adding 2 optimizations and wionder if i should use your model of seq optimization | 2024-12-11 09:48:54 | None 441 | 1272962252080943209 | 1253625758832660541 | ethan83_ | im exploring atm CL gone through MES and MNQ im starting from the back | 2024-12-11 09:48:54 | None 442 | 1272962047147249827 | 1253625758832660541 | coensio | like for kids | 2024-12-11 09:48:54 | None 443 | 1272961993023684740 | 1253625758832660541 | coensio | exactly this level of magic! | 2024-12-11 09:48:54 | None 444 | 1272961982395580499 | 1253625758832660541 | ethan83_ | just basic workflow with some modifications | 2024-12-11 09:48:54 | None 445 | 1272961904511418541 | 1253625758832660541 | ethan83_ | haha here we go again just basic workflow and bit of coensio | 2024-12-11 09:48:54 | None 446 | 1272961891395965100 | 1253625758832660541 | ethan83_ | https://tenor.com/view/aeons-aeonsroleplay-aeons-of-sand-robin-kaecilius-gif-26034715 | 2024-12-11 09:48:54 | None 447 | 1272961793962016831 | 1253625758832660541 | coensio | in a portfolio | 2024-12-11 09:48:54 | None 448 | 1272961765461721090 | 1253625758832660541 | coensio | while everything around Rdd of 10 after costs is tradable | 2024-12-11 09:48:54 | None 449 | 1272961619592216577 | 1253625758832660541 | coensio | because it hard to find Rdd30 outside indices and everybody is blinded by rdd and avg trade | 2024-12-11 09:48:54 | None 450 | 1272961463895720022 | 1253625758832660541 | coensio | but nobody takes the effort to look into other markets... | 2024-12-11 09:48:54 | None 451 | 1272961405221470238 | 1253625758832660541 | coensio | this is with only slightly modified basic workflow | 2024-12-11 09:48:54 | None 452 | 1272961288628211732 | 1253625758832660541 | ethan83_ | equty curve is epix c i must say | 2024-12-11 09:48:54 | None 453 | 1272961145564696597 | 1253625758832660541 | coensio | but I trade it on MGC | 2024-12-11 09:48:54 | None 454 | 1272961107094536243 | 1253625758832660541 | coensio | I think I reserved $20 for that | 2024-12-11 09:48:54 | None 455 | 1272960917503741953 | 1253625758832660541 | ethan83_ | is the gc with or without slippage and comissions? | 2024-12-11 09:48:54 | None 456 | 1272960777317253312 | 1253625758832660541 | coensio | flat | 2024-12-11 09:48:54 | None 457 | 1272960225300971541 | 1253625758832660541 | coensio | hard to see that way I need to load SQX version | 2024-12-11 09:48:54 | None 458 | 1272960024846798963 | 1253625758832660541 | coensio | wait | 2024-12-11 09:48:54 | None 459 | 1272959978159997010 | 1253625758832660541 | edau87 | Can I see what that did in September of 2023? | 2024-12-11 09:48:54 | None 460 | 1272959874355040296 | 1253625758832660541 | coensio | one of my best strats: | 2024-12-11 09:48:54 | None 461 | 1272959826242310219 | 1253625758832660541 | edau87 | I only need a quick impulse to build off of. Hopefully the patterns stay good through 2022-2024 since they are my OOS periods | 2024-12-11 09:48:54 | None 462 | 1272959752753774642 | 1253625758832660541 | coensio | there are tons of possibilities in GC | 2024-12-11 09:48:54 | None 463 | 1272959655349588018 | 1253625758832660541 | edau87 | I've just started exploring GC daily patterns which lead to good 2-4 day swings in hopes of finding good H2/H4 strategies. | 2024-12-11 09:48:55 | 1272945104985784383 464 | 1272959553960673444 | 1253625758832660541 | coensio | domain is blocked you cannot send him an email | 2024-12-11 09:48:55 | None 465 | 1272959477708230819 | 1253625758832660541 | coensio | I have no means to verify that..... | 2024-12-11 09:48:55 | None 466 | 1272959459664072755 | 1253625758832660541 | edau87 | Fake email addresses | 2024-12-11 09:48:55 | None 467 | 1272959410469081139 | 1253625758832660541 | edau87 | Only if they turn out to be legit!! | 2024-12-11 09:48:55 | None 468 | 1272959331813429341 | 1253625758832660541 | coensio | another referral from you with your code but this time with fake email address on a qq.com domain. I owe you 2x, will settle this tomorrow. | 2024-12-11 09:48:55 | None 469 | 1272945104985784383 | 1253625758832660541 | coensio | Additional I think diversification it's the best when done on multiple markets, have you looked eg at GC/MGC strategies? these are exploding now, and are very easy to find. | 2024-12-11 09:48:55 | None 470 | 1272943904970244097 | 1253625758832660541 | coensio | MNQ and MES is fun at Rdd30 and high Avg.Trade, when the market is aligned but if it's not then you get DD, and moreover your portfolio will be 95% of the time in DD, get used to it, make sure your max DD is not painful so you can survive bad markets. I can easily handle -30%, without crying at night but my portfolio goes mostly to -10% and back | 2024-12-11 09:48:55 | 1272838023138246706 471 | 1272897807283654720 | 1253625758832660541 | edau87 | But if you can handle 15% it's all good I suppose. | 2024-12-11 09:48:55 | None 472 | 1272897465548279848 | 1253625758832660541 | edau87 | I think you might still be oversized. I topped out at 11% unrealized with daily swings entered near the very start of the down move and held all the way down. | 2024-12-11 09:48:55 | 1272835000575721514 473 | 1272838089538015242 | 1253625758832660541 | ethan83_ | so atm my head is spinning | 2024-12-11 09:48:55 | None 474 | 1272838023138246706 | 1253625758832660541 | ethan83_ | Finding the part hard is atm building portfolios and it chooses strategys that can be in a bigger drawdown how to handle that ? , Just dont think about it and watch so its not in its buggest alltime drawdown?, Can be bit tricky cause on one side it can be great getting in on a drawdown on the other hand it can keep tanking down. And also making me think about my workflows am i overoptimizing going to har finding strategies with stagnation 60% correlation) even across multiple runs. | 2024-12-11 09:49:21 | None 859 | 1314681749757759539 | 1285270256981377077 | coensio | at the very end I will do some correlation checks | 2024-12-11 09:49:22 | None 860 | 1314681670150127616 | 1285270256981377077 | coensio | yes only the name, strategies are different | 2024-12-11 09:49:22 | None 861 | 1314681553628041316 | 1285270256981377077 | goodday20 | cool. for blocks, you just track the indicator name and not how exacty it is used by the strategy - right? | 2024-12-11 09:49:22 | None 862 | 1314681377362542723 | 1285270256981377077 | coensio | yes I increased block pool to 6 now | 2024-12-11 09:49:22 | 1314681231329329253 863 | 1314681231329329253 | 1285270256981377077 | goodday20 | yep. so this is with more blocks and still RETDD > 95%? it seems very very good and also chances are these selected strategies are not all very correlated | 2024-12-11 09:49:22 | None 864 | 1314680900763517000 | 1285270256981377077 | coensio | yes...getting better and better | 2024-12-11 09:49:22 | 1314678098532499576 865 | 1314678671625293826 | 1285270256981377077 | goodday20 | at 95% SR, for e.g. if you pick 2 randomly from selected pool, the chances that both fail is 0.25% i.e. almost 100% that at least one of them will be successful. | 2024-12-11 09:49:22 | None 866 | 1314678098532499576 | 1285270256981377077 | goodday20 | ok. well i think 100% is not really needed. anything about 95-98 shd be good enough. the 100% can be ruined by just 1 failing strategy | 2024-12-11 09:49:22 | 1314677450046967888 867 | 1314677450046967888 | 1285270256981377077 | coensio | this is with selection at 95%, the only way to get SR of 100% across all runs, but I am trying to find the largest possible logic blocks pool | 2024-12-11 09:49:22 | 1314656006491344910 868 | 1314656006491344910 | 1285270256981377077 | goodday20 | so these selected # is from the pool of 50k that you generated? right.. i.e. less than 1% in most cases | 2024-12-11 09:49:22 | 1314622801751052368 869 | 1314655787712512030 | 1285270256981377077 | goodday20 | wow. 100% SR with good # of strategies | 2024-12-11 09:49:22 | 1314622801751052368 870 | 1314622904050389073 | 1285270256981377077 | coensio | working now on much wider selection let's see what can we get | 2024-12-11 09:49:22 | None 871 | 1314622801751052368 | 1285270256981377077 | coensio | | 2024-12-11 09:49:22 | None 872 | 1314587400025935953 | 1285270256981377077 | goodday20 | got it. The selection criteria seems quite strict in that case. Also to get multiple uncorrelated strategies , we probably need more variations in the selected pool | 2024-12-11 09:49:22 | 1314586201243521036 873 | 1314586976451563620 | 1285270256981377077 | coensio | I think it will take me few days to process all data...it's huge. | 2024-12-11 09:49:22 | None 874 | 1314586201243521036 | 1285270256981377077 | coensio | yes indeed some runs produced 0 selectable strategies, but I'm working on it I see this can be improved | 2024-12-11 09:49:22 | 1314585913468125185 875 | 1314585913468125185 | 1285270256981377077 | goodday20 | Wow 100pct. Does this show that in run 6, you ended up selecting 1 strategy out of 50k? And used pivots and ret/dd greater than 95 pct as the criteria? Just want yo understand what this is showing | 2024-12-11 09:49:23 | 1314583365139169331 876 | 1314583365139169331 | 1285270256981377077 | coensio | Pivots and RDD>95% | 2024-12-11 09:49:23 | None 877 | 1314518260225675304 | 1285270256981377077 | coensio | Total TOOS distributions | 2024-12-11 09:49:23 | None 878 | 1314517937461133376 | 1285270256981377077 | coensio | ok so I did not change anything to my previously found selection model and this time I shifted workflow each quarter 13x50k=650k strategies built from scratch. Resulting SR is not that bad at all 98.1% over 13 different dataranges, while previously optimized with 3. | 2024-12-11 09:49:23 | None 879 | 1312923936492621897 | 1285270256981377077 | bla0018 | Youre essentially doing the same thing when using templates or selecting good indicators. The key difference is that templates allow you to pre-validate each indicator independently. Without this validation, its impossible to know which indicators perform poorly in out-of-sample tests. By validating each indicator one by one, I can build a library of templates that are proven to work across multiple OOS tests. This approach allows me to quickly swap templates into strategies while being confident in their reliability. I wouldnt say templates are inherently better than running the workflow without them. However, keep in mind that without isolating and testing each indicator, you wont have a clear understanding of its success rate on a specific instrument and timeframe. Templates eliminate this uncertainty, giving you a validated foundation to build strategies on. | 2024-12-11 09:49:23 | 1312904508392607866 880 | 1312908881378410496 | 1285270256981377077 | edau87 | The correlation within a run is meaningless. You only select one strategy. Templates decrease correlation across runs. | 2024-12-11 09:49:23 | 1312904767982534676 881 | 1312904767982534676 | 1285270256981377077 | goodday20 | what i have seen is that when the SR is high, the strategies also tend to be more correlated. Our ultimate goal is to develop strategies which hae low correlation and high TOOS SR | 2024-12-11 09:49:23 | None 882 | 1312904508392607866 | 1285270256981377077 | goodday20 | what is the benefit of using templates over just selecting "good" indicators in the builder. I know template provides even more control on exact conditions etc but the generated strategies are likely more correlated with each other. | 2024-12-11 09:49:23 | 1312881783888679004 883 | 1312881783888679004 | 1285270256981377077 | bla0018 | Yeah I came to the same conclusion. After spending time validating the workflow I will suddenly get a bad run where the genetic algo will select poor price indicators resulting to a low SR for that TOOS even though I was achieving good success rate 9/10 times. I too am now only using templates to restrict the genetic algo from having to much freedom | 2024-12-11 09:49:23 | 1312804910655471648 884 | 1312820531384287370 | 1285270256981377077 | goodday20 | ok. Thanks. | 2024-12-11 09:49:23 | None 885 | 1312804910655471648 | 1285270256981377077 | coensio | This is with all blocks without predefined SQX signals, then SQX will use a combination of indicators to generate it's own signals. | 2024-12-11 09:49:23 | 1312799907395932250 886 | 1312800130080051231 | 1285270256981377077 | goodday20 | I have also noticed the same. For mnq, it seems very basic config with basic custom filters give 90+ SR. | 2024-12-11 09:49:23 | None 887 | 1312799907395932250 | 1285270256981377077 | goodday20 | , thanks for sharing your conclusions. Just to clarify, when you say building blocks, you mean indicators (on building blocks tab) or signals or both. Also if the workflow doesn't allow entry signals (i.e. always true as in basic mnq workflow) then indicators/signals dont matter.. just stop/limit order settings matter.. | 2024-12-11 09:49:23 | None 888 | 1312792697601523783 | 1285270256981377077 | coensio | yes this is similar, and will probably also work. | 2024-12-11 09:49:24 | 1312792193202913332 889 | 1312792193202913332 | 1285270256981377077 | humble.trading | I've come to very similar conclusions. What I am doing now is creating templates that limit the block selection through random groups with the groups broken out between intraday (price, filter and exits) the same for higher timeframe. Then it is a matter of filling out those groups with positive eratio conditions and prices and creating various templates that utilize them. | 2024-12-11 09:49:24 | 1312790711422160946 890 | 1312790711422160946 | 1285270256981377077 | coensio | No the main question was: what has the biggest impact on the resulting SR numbers. | 2024-12-11 09:49:24 | 1312790525811888148 891 | 1312790525811888148 | 1285270256981377077 | humble.trading | So is trying to figure out block selection? | 2024-12-11 09:49:24 | 1312729996464488468 892 | 1312790422585868358 | 1285270256981377077 | coensio | I do not know if I like this, but this is what the data is telling me after running this exercise. | 2024-12-11 09:49:24 | None 893 | 1312789198973177876 | 1285270256981377077 | coensio | Basically, it all comes down to this: **find a "good" set of logic indicators **that will build profitable strategies and pass some/any of **basic validation tests** on a given market, **no matter what the data range is.** The more data rages you cover the more chances you have it will keep working in the "future", Then, use **a simple strategy selection method** to increase your chances of selecting a good live strategy candidate (for next TOOS period). | 2024-12-11 09:49:24 | None 894 | 1312789095201771600 | 1285270256981377077 | coensio | **Conclusions:** 1. Without any selection (raw workflow output), there is very little difference between all workflow configurations. I tend to state that more complex (longer) workflows show only a slight improvement over the simpler, shorter workflows with fewer validation tests. When building with all building blocks, the workflows alone are NOT sufficient. There is no validation tests that really makes a huge difference! There are only "good" and "bad" years (= data ranges) with SRs between 80+ to 90+ %. The differences between the workflows is all within few percent! You can boost this performance by building using weighted fitness functions, but this is all not enough. 2. When using workflows based on a builder with all indicators enabled, the GA will find/build different types of strategies for each workflow data range (TOOS shift). This means it is hard to compare results across different data ranges as strategies/logic are somewhat different. Additionally, it is somewhat harder to find common good strategies across all data ranges (TOOS shifts). Running builders with too many logic blocks enabled seem to be waste of time, at least if you are not doing this with automatic post-optimization in mind. 3. Once you find common logics across all different runs, the picture becomes very clear: Strategy logic/building blocks (especially price indicators) are the number one factor determining the resulting success rate. Once you determine "good" building blocks and cross-validate them over different TOOS shifts (the more, the better), then all other things like: workflow configurations, strategy selection parameters, and all other factors make a only small difference on the resulting success rate, giving you the last few % of improvement. 4. By doing some large scale optimizations, in all cases, I was able to find a good set of blocks + simple selection method resulting in SR's close to 100%. | 2024-12-11 09:49:24 | None 895 | 1312768728231837698 | 1285270256981377077 | edau87 | Just for curiosity's sake. | 2024-12-11 09:49:24 | None 896 | 1312767380723728465 | 1285270256981377077 | edau87 | I wish they'd implement the MC SPP variant | 2024-12-11 09:49:24 | None 897 | 1312767282895781918 | 1285270256981377077 | coensio | I will later write all of my conclusions in a list | 2024-12-11 09:49:24 | None 898 | 1312766471671255173 | 1285270256981377077 | edau87 | Did you notice anything interesting when using SPP? | 2024-12-11 09:49:24 | 1312729996464488468 899 | 1312729996464488468 | 1285270256981377077 | coensio | It's hard to show it in a figure, but I did something like this. Still processing the data. | 2024-12-11 09:49:24 | None 900 | 1312422472708001884 | 1285270256981377077 | bla0018 | Yep you got it. there was a mismatch with my draw down settings noob mistake thanks Chris | 2024-12-11 09:49:24 | 1312421814592606238 901 | 1312421814592606238 | 1285270256981377077 | coensio | are you sure your PASSED/FAILED criteria in SQX are the same as programmed in the tool? | 2024-12-11 09:49:24 | 1312420398687715461 902 | 1312420398687715461 | 1285270256981377077 | bla0018 | I just completed testing the new V8 version (previously using V6). I understand that the second percentage represents the baseline workflow success rate (Net Profit > X and Drawdown < X for all candidates). However, this percentage doesnt align with my databank. Ive attached my databank for reference. According to the databank, the success rate (SR) for TOOS 1 is 87%, but the screenshot shows it as 75%. Could you help clarify the discrepancy? All of them are off | 2024-12-11 09:49:25 | None 903 | 1312052271764475904 | 1285270256981377077 | edau87 | SQX sometimes won't register workflow changes you make. I always close the whole thing out and check every step 2-3x when I'm making changes. | 2024-12-11 09:49:25 | None 904 | 1312033617290989669 | 1285270256981377077 | tonim2802 | Found the error, workflow wasn't complete and it was different data between old run and new run | 2024-12-11 09:49:25 | None 905 | 1311347635868794900 | 1285270256981377077 | greggarnet | I'm trying to get the Workflow Validator Dashboard v08 working. I've installed missing python dependencies, set the SQX and workflow paths in the AutoRun.py file, I set the inputs (last sqx date, toos days, is days, num workflows, etc.), but when I presss the button "Generate Workflows" nothing happens. There is an error in the dashboard saying "A callback is missing Inputs", but I'm not sure that is related to the generation of the workflows. Any ideas? | 2024-12-11 09:49:25 | None 906 | 1309888554570219641 | 1285270256981377077 | eliasmi | This all sounds like a conversation I am having with chatgpt on the subject. I don't understand that either | 2024-12-11 09:49:25 | 1309824196901339207 907 | 1309829540872323072 | 1285270256981377077 | edau87 | I just meant the MC portfolio trader portfolio report. | 2024-12-11 09:49:25 | 1309824196901339207 908 | 1309826213354799156 | 1285270256981377077 | humble.trading | All my passed strategies are already in portfolio trader. I handle rebalancing by importing in easylanguage a file that identified which strategies to make active for the next cycle. That gets saved as a new portfolio version and run live. The old cycle in portfolio trader gets a flag set to not take any new trades but manages the exits of existing trades. | 2024-12-11 09:49:25 | None 909 | 1309824196901339207 | 1285270256981377077 | humble.trading | I don't use quantanalyzer/sqx for portfolio generation. The greedy method above is all in java and doesn't require any optimization just a series of logical filters. I also calculate the monte carlo drawdown for the portfolio in python as I save the equity curve and trade details for every strategy in a database and calculate my correlations that way. I also have a sizing step that looks at long term volatility sizing (126 day) to normalize positions across all the products. I've moved away from return / drawdown as a measure to optimize to as I think it can give a false sense of security and under estimate loss. | 2024-12-11 09:49:25 | 1309669980887060655 910 | 1309771648219742218 | 1285270256981377077 | eliasmi | What time period are you judging correlation over ? On a daily or weekly? I wanted to do a 20 day rolling correlation but I don't know how to do that yet. So for now i use the weekly in sqx and the results are pretty good | 2024-12-11 09:49:25 | 1309671081333883033 911 | 1309671081333883033 | 1285270256981377077 | edau87 | I'm way too correlated to the short side. QA just doesn't do a good job. | 2024-12-11 09:49:25 | None 912 | 1309669980887060655 | 1285270256981377077 | edau87 | Rpt | 2024-12-11 09:49:25 | None 913 | 1309669955159068734 | 1285270256981377077 | edau87 | 120?! Let's see that perf rot | 2024-12-11 09:49:25 | 1309654049162788925 914 | 1309668267677192284 | 1285270256981377077 | humble.trading | Correlation wise I split it between 10Y and 3Y | 2024-12-11 09:49:25 | None 915 | 1309667570852302919 | 1285270256981377077 | humble.trading | I'm up to around 120 algos at this point on (mes, mnq, rty, mym, cl, ng). Some are hurting performance but I'm including many for diversity sake. I'm rebalancing the portfolio monthly so the composition changes monthly. | 2024-12-11 09:49:25 | 1309642289232216124 916 | 1309654049162788925 | 1285270256981377077 | edau87 | This sounds a whole lot better than quant analyzer lol. | 2024-12-11 09:49:25 | 1309633762941997057 917 | 1309653156799447093 | 1285270256981377077 | eliasmi | So for your custom analysis you don't need to set the granularity or is the portfolio all one time period eg 15 min and you are effectively looking at correlation every 15 minutes? | 2024-12-11 09:49:25 | 1309633762941997057 918 | 1309644822675460208 | 1285270256981377077 | ar3s7 | finally got it working, you were correct. i had to make some slight changes, but up and running now | 2024-12-11 09:49:26 | 1309419556393193503 919 | 1309642289232216124 | 1285270256981377077 | edau87 | What does your overall portfolio performance report look like these days? | 2024-12-11 09:49:26 | 1309631296590577665 920 | 1309633762941997057 | 1285270256981377077 | humble.trading | I calculate the correlations in a custom analysis between every strategy. It's using the underlying trades/equity curve. | 2024-12-11 09:49:26 | 1309610108535308408 921 | 1309631296590577665 | 1285270256981377077 | humble.trading | My swing strategies on rty are broken. My day trading strats are still working quite well. | 2024-12-11 09:49:26 | 1309591567840841869 922 | 1309631072144850954 | 1285270256981377077 | humble.trading | If I did pursue the single stock approach i would still look to trade a basket of such strategies that in theory were in the market at different times. I use to run the algo cloud type stuff utilizing wealthlab for execution. The problem i ran into is that the strategies tend to trigger all at the same time and you have to manage which of the 30+ limit/stop orders you want to execute and in what priority (which is tough and also very path dependent). What I've seen work well is rotating the composition of a basket of 5-20 stocks every day that is always in the market. The work never ends !!! | 2024-12-11 09:49:26 | 1309565793671057459 923 | 1309610108535308408 | 1285270256981377077 | eliasmi | What are you basing correlation on ? Timeframe | 2024-12-11 09:49:26 | 1309525689724305418 924 | 1309591567840841869 | 1285270256981377077 | edau87 | RTY has been utterly broken for almost five years. I don't know what the future of strategy development on it looks like. | 2024-12-11 09:49:26 | 1309538303791468555 925 | 1309565793671057459 | 1285270256981377077 | goodday20 | i think with stocks you will have to go with the 2nd approach here. running single stock strategies combined into portfolio will not utilize cash well (unless the strategies time in market is large - at least 30+ percentage of time). with stocks, one way to run portfolio is to rank stocks based on some return metric and then go long the top ones. fitting this approach into SQX is hard. i think their algo cloud does something like this | 2024-12-11 09:49:26 | 1309555940034613348 926 | 1309555940034613348 | 1285270256981377077 | humble.trading | Thanks for the tips. I don't run as much leverage with my futures and been meaning to implement some stock methodologies to utilize my cash better. What do you recommend pursuing first single stock strategies combined into a portfolio or a few strategies that are run across a portfolio of stocks (i.e. stockpicker type) ? | 2024-12-11 09:49:26 | 1309554830259388456 927 | 1309554830259388456 | 1285270256981377077 | goodday20 | if you go with stock route , one thing to note is that slippage relative to notional value is much higher. Sometimes 5-10 times more. In options it is even more. So your avg tarde has to be much larger. Or low trade freq and low ret/dd on individual strategies but portfolio with 50 of those will give good portfolio statistics. | 2024-12-11 09:49:26 | 1309549510086754375 928 | 1309549510086754375 | 1285270256981377077 | humble.trading | I've been thinking about instead of other futures markets looking at applying the methodology to fang stocks. Obviously very correlated to NQ but different enough to be uncorrelated. Probably need to execute through options if you need the additional leverage. | 2024-12-11 09:49:27 | None 929 | 1309548241108336680 | 1285270256981377077 | humble.trading | I find it's hard to determine if a good result is due to selection method or good market environment as it all gets lumped together in the TOOS. | 2024-12-11 09:49:27 | 1309541373715943424 930 | 1309547564554780672 | 1285270256981377077 | humble.trading | hmmm... let me think on that. Maybe I should be adding another layer of decision making in the selection process. | 2024-12-11 09:49:27 | 1309538862342606910 931 | 1309547077260546098 | 1285270256981377077 | humble.trading | TOOS results are the best with this approach but the ride can be bumpy... I do like though that you can easily create portfolio's with as many strategies as you want. Thanks for your thoughts! | 2024-12-11 09:49:27 | 1309537439278305440 932 | 1309542403039825920 | 1285270256981377077 | goodday20 | slowly and surely you will get there.. | 2024-12-11 09:49:27 | 1309542046314397829 933 | 1309542046314397829 | 1285270256981377077 | coensio | that is why I said "if I could afford it" | 2024-12-11 09:49:27 | 1309541750452391966 934 | 1309541750452391966 | 1285270256981377077 | goodday20 | yes. there is a correlation chart. many of the other markets dont have micro contract. also trading less so i guess slippage will be higher. but yes overall there are 30-40 futures markets (that typicall trend following CTA funds trade including lean hogs, orange juice etc). | 2024-12-11 09:49:27 | 1309540411962032160 935 | 1309541373715943424 | 1285270256981377077 | goodday20 | exactly. for the tougher periods, if it picks few strategies that is fine. you either runs smaller portfolio in those period or increase lot size. in ideal world the selection method will have low false negative and high true positive. but in practice, selection method with high true positive is still very good. even though it may miss some which will do better in TOOS | 2024-12-11 09:49:27 | 1309539883244847135 936 | 1309540411962032160 | 1285270256981377077 | coensio | GC is highly correlated with MNQ, but think about markets from different sectors..like energies, grains, bonds, etc...these are mostly uncorrelated...somebody here shared a nice correlation table in the past... | 2024-12-11 09:49:27 | 1309540109384945714 937 | 1309540109384945714 | 1285270256981377077 | goodday20 | yes. certainly adding new markets is a better option. i think mnq, mes, gc, cl are a good combo. | 2024-12-11 09:49:27 | 1309539253889400893 938 | 1309539883244847135 | 1285270256981377077 | coensio | yes my model has different functions, for example I do penalize unbalanced distributions across the runs, but it is hard to find selection model with high number of selectees in each single run as some years are very hard to build on. | 2024-12-11 09:49:28 | 1309539337741795350 939 | 1309539337741795350 | 1285270256981377077 | goodday20 | got it. so this selection method picks vastly different #s, some rounds it is over 200, some only 6. that seems quite a variation but also says the method is picking the "gold" in all cases. | 2024-12-11 09:49:28 | 1309538099101171732 940 | 1309539253889400893 | 1285270256981377077 | coensio | yeah if I could afford it, I would trade 20 different uncorrelated markets...if you target different markets then strategies are by default mostly uncorrelated. | 2024-12-11 09:49:28 | 1309538862342606910 941 | 1309538862342606910 | 1285270256981377077 | goodday20 | yes. this is one reasonable way. also a bit depends on your objective. if you want lower stagnation in live trading, you can be greedy in picking strategies which do better when rest of the already selected ones dont.. this way you are basically plugging holes (flat areas/slight dd) in the equity curve. | 2024-12-11 09:49:28 | 1309537439278305440 942 | 1309538700178358282 | 1285270256981377077 | coensio | exactly this is my main conclusion = simpler = better, entry indicators are not relevant, price indicators is everything you need. | 2024-12-11 09:49:28 | 1309538441502920744 943 | 1309538441502920744 | 1285270256981377077 | goodday20 | wow. that is really good in that case. simpler strategies, i.e. keeping fewer conditions means higher robustness by design. i recently started ading an entry condition and wondering if i should spend much time pursuing that path. without entry conditions, it is always a price level + mutiplier * some range as a stop order. very simple construction | 2024-12-11 09:49:28 | 1309537775791509555 944 | 1309538303791468555 | 1285270256981377077 | coensio | I wish I could find robust and profitable strategies on RTY...never did. This market sucks. | 2024-12-11 09:49:28 | 1309529603832021023 945 | 1309538099101171732 | 1285270256981377077 | coensio | I have 1 selection model which for each workflow run selects X strategies and validates how many of them are PASS, 6/6 means all of selected strategies from this run are PASS | 2024-12-11 09:49:28 | 1309529319428849726 946 | 1309537775791509555 | 1285270256981377077 | coensio | no not really, this is still the same as in basic workflow, I only adjusted some thresholds here and there | 2024-12-11 09:49:28 | 1309527000574070804 947 | 1309537439278305440 | 1285270256981377077 | coensio | "greedy methodology" I would do this one as ideally you want to have strong AND uncorrelated strategies in your portfolio, especially when you have poor market diversification. | 2024-12-11 09:49:28 | 1309525689724305418 948 | 1309529603832021023 | 1285270256981377077 | humble.trading | I would be curious on how it performs on a market with a lower success rate like RTY. Have you had a chance to test that yet? | 2024-12-11 09:49:28 | 1309427792861073439 949 | 1309529319428849726 | 1285270256981377077 | goodday20 | what are the numbers in the paranthesis (e g. R6(6/6) - is that the number of strategies selected? That varies a lot between different rounds. Also how many were the total strategies from which you selected? | 2024-12-11 09:49:28 | 1309427792861073439 950 | 1309527000574070804 | 1285270256981377077 | goodday20 | i assume you had to add entryConditions in simple workflow to get these metrics? | 2024-12-11 09:49:28 | 1309497346215776256 951 | 1309525689724305418 | 1285270256981377077 | humble.trading | First of all, let me share some know how with the portfolio builder. It's not documented but you can create your own fitness functions to solve for. This helps tremendously and also can help you create uncorrelated portfolios by using different fitness functions. Question for you guys... I've been working hard on portfolio generation. I wanted to get your thoughts and ideas at the best way to automate the process. I have hundreds of runs now with profitable, stable stage 2's. One approach I've experimented with is a greedy methodology which is to take all the stage 2 strategies and sort them by fitness, take the top 1, then the next one with a correlation < [x] and repeat until I have the desired amount of strategies. This ultimately produces portfolios with the highest individual strategy fitness with correlation below a given threshold. For subsequent portfolios you eliminate the strategies from the prior selection and repeat the process. This will allow you to create an almost unlimited amount of portfolios with a correlation < [x]. The other approach I've taken is to solve for portfolios using portfolio builder and varying the fitness to get portfolios that aren't correlated because they have different objective functions. Can you think of any other approaches? | 2024-12-11 09:49:29 | None 952 | 1309499729997987880 | 1285270256981377077 | coensio | yeah automation really makes things easier | 2024-12-11 09:49:29 | 1309499152086077452 953 | 1309499152086077452 | 1285270256981377077 | edau87 | I'm still stuck in 2023 when it comes to this. Manually notating changes lol. | 2024-12-11 09:49:29 | 1309498529743376426 954 | 1309498656101109780 | 1285270256981377077 | coensio | I will share my findings if I get any... | 2024-12-11 09:49:29 | None 955 | 1309498529743376426 | 1285270256981377077 | coensio | I'm running now some other tests where you can easily see how the workflow improves or worsens by doing some simple workflow changes...adding removing steps | 2024-12-11 09:49:29 | None 956 | 1309498020974297158 | 1285270256981377077 | edau87 | $2,000 net profit average across 2023 and 2024 is no small feat. | 2024-12-11 09:49:29 | None 957 | 1309497911767208038 | 1285270256981377077 | coensio | yes but with 1 year TOOS for each, workflow shifted each 6months | 2024-12-11 09:49:29 | 1309497587769937990 958 | 1309497608107982969 | 1285270256981377077 | edau87 | Those results are stellar | 2024-12-11 09:49:29 | None 959 | 1309497587769937990 | 1285270256981377077 | edau87 | SO, m15, MNQ, ~$2,000 average TOOS NP on ~$1,000 DD with a ~$20-$30 average trade on six month TOOS, walked forward. | 2024-12-11 09:49:29 | None 960 | 1309497455359950941 | 1285270256981377077 | coensio | costs included | 2024-12-11 09:49:29 | None 961 | 1309497346215776256 | 1285270256981377077 | coensio | yes basic workflow with very little modifications... | 2024-12-11 09:49:29 | None 962 | 1309497246282289273 | 1285270256981377077 | edau87 | ooohhh okay, and this is an m15 MNQ project? | 2024-12-11 09:49:29 | None 963 | 1309497187020967960 | 1285270256981377077 | coensio | all 7x TOOS combined | 2024-12-11 09:49:30 | None 964 | 1309497113343819796 | 1285270256981377077 | edau87 | For reference/context, what TOOS year are we looking at here? | 2024-12-11 09:49:30 | None 965 | 1309496675164749865 | 1285270256981377077 | coensio | selected distribution looks pretty nice, with an extremely thin tail < $0 | 2024-12-11 09:49:30 | None 966 | 1309495991614836817 | 1285270256981377077 | edau87 | Yeah that's what I thought. | 2024-12-11 09:49:30 | None 967 | 1309495809745752074 | 1285270256981377077 | coensio | yes only to get DD MC numbers | 2024-12-11 09:49:30 | 1309495504178253964 968 | 1309495743190663219 | 1285270256981377077 | coensio | that's why my primary goal is to see the continuation of past performance, that's more realistic way of evaluating strategies. | 2024-12-11 09:49:30 | 1309495318353678458 969 | 1309495504178253964 | 1285270256981377077 | edau87 | But, wait, are you running another MC on your TOOS bank? | 2024-12-11 09:49:30 | None 970 | 1309495318353678458 | 1285270256981377077 | edau87 | Excellent work. Sometimes I think I focus too heavily on the EOY profit number. Often, some poor EOY results have really amazing months at times my great EOY profit strategies struggled. | 2024-12-11 09:49:30 | 1309493031178604686 971 | 1309493182861672458 | 1285270256981377077 | coensio | but the results with 'Net profit (OOS)' >= 0 & 'Drawdown (OOS)' 75 Then Sell next bar at open | 2024-12-11 09:49:31 | 1308867231043682417 988 | 1308867231043682417 | 1285270256981377077 | goodday20 | it would be amazing if SQX could split out python code.. you can experiment with generated strategies much more easily. | 2024-12-11 09:49:31 | 1308861556963938336 989 | 1308861556963938336 | 1285270256981377077 | edau87 | Right. That's what this will be. A version of MC that can execute Python strategies. | 2024-12-11 09:49:31 | 1308861170547032094 990 | 1308861170547032094 | 1285270256981377077 | humble.trading | hmmm.... i didn't see any details on what they were planning. I have the.net version which allows you more control over a lot of things. I just don't use it because sqx only spits out easy language. | 2024-12-11 09:49:31 | 1308860607180701821 991 | 1308860607180701821 | 1285270256981377077 | edau87 | I saw this, but it's useless unless you hand code python strategies. | 2024-12-11 09:49:31 | 1308854607992062013 992 | 1308854607992062013 | 1285270256981377077 | humble.trading | https://www.multicharts.com/features/python/ | 2024-12-11 09:49:32 | None 993 | 1308845521246355597 | 1285270256981377077 | ethan83_ | this is yes NP >= 500 But it was just for testing now i know it all works, So now im creating with genetic optimization 10k different workflows all with different parameters. It creates fodlers and copies python script and all. I also set it up so i can use multiple computers to run at the same time which is handy :).I then use a viewer python program pre set that uses scoring to pick 10 best workflows of the 10k save it to a preset folder. Then i open up this program that automatic creates new workflows with preset for the workflow we use in course. It then start running 10 x 4 workflows ST1 to ST4 for all workflows it automatic saves all the results to correct folders. After that i change the tab to this and i can then check all individual and select best workflows with best results. This is my firs go and i start with MNQ.D. KInda hard to explain it all i feel haha:) | 2024-12-11 09:49:32 | None 994 | 1308835122836934696 | 1285270256981377077 | goodday20 | what is your success criteria here btw. (NP > -500 or something more strict)? | 2024-12-11 09:49:32 | 1308806185633448047 995 | 1308808425223880786 | 1285270256981377077 | ethan83_ | God point thanks, me being to strict again i guess. | 2024-12-11 09:49:32 | 1308549532350419077 996 | 1308806292244004957 | 1285270256981377077 | ethan83_ | sleep for a few days now:) | 2024-12-11 09:49:32 | None 997 | 1308806261965459559 | 1285270256981377077 | ethan83_ | totaly redone graphical part and made all mre automated | 2024-12-11 09:49:32 | None 998 | 1308806185633448047 | 1285270256981377077 | ethan83_ | fianlly all macthing | 2024-12-11 09:49:32 | None 999 | 1308549532350419077 | 1285270256981377077 | goodday20 | why annoying? isn't this just saying the # of trades is not a good selection metric for TOOS SR.. all other metrics are doing a good job in getting 90+% SR | 2024-12-11 09:49:32 | 1308545003408326727 1000 | 1308545003408326727 | 1285270256981377077 | ethan83_ | bit anoying | 2024-12-11 09:49:32 | None 1001 | 1308100683152490588 | 1285270256981377077 | edau87 | I agree with this. TNickel's work in the data science channel in the Main SQX discord underscore my feelings on super complex ML methods. They are buzzwords for the moment. | 2024-12-11 09:49:32 | 1308096223898632203 1002 | 1308096223898632203 | 1285270256981377077 | humble.trading | I think it's just whether you use some of the more complex AI methods like tree's and what not. Composite scores in general I found work quite well, particularly if the fitness function and composite score are aligned. | 2024-12-11 09:49:32 | 1308091570993758239 1003 | 1308093174434103356 | 1285270256981377077 | edau87 | You're only talking about a sample of a couple hundred workflow runs, though. | 2024-12-11 09:49:32 | None 1004 | 1308091570993758239 | 1285270256981377077 | edau87 | Yeah. How are we defining "complex" versus "simple" for final selectee methodology? Is my method of SCORE-ing strategies simple or complex? Random selections always succeed, but far more often than not my selectee outperforms the random selectee as well as the average and median values for the selection databank as a whole. | 2024-12-11 09:49:32 | 1308063347169693777 1005 | 1308063347169693777 | 1285270256981377077 | humble.trading | I'm on it.... as to b) above the only caveat I have is depends on the scope of data. Tree models are great if it's too computationally intense to do it any other way which is why I use it depending on the problem. | 2024-12-11 09:49:32 | 1308062904825675826 1006 | 1308062904825675826 | 1285270256981377077 | coensio | it will be, if someone can somehow reproduce it...or try something similar | 2024-12-11 09:49:32 | 1308062793953705994 1007 | 1308062793953705994 | 1285270256981377077 | humble.trading | Thanks man. Great stuff. | 2024-12-11 09:49:33 | 1308062635824254977 1008 | 1308062635824254977 | 1285270256981377077 | coensio | a) yes, price indicator blocks seem to be the most important factor b) precisely c) e.g.: model optimization done using workflow 2011...2021 and TOOS:2021, cross validation on shifted workflow using data 2013...2023 and TOOS:2023. But of course we have many workflow shifts so it can be done using "pairwise iteration" for each of them. | 2024-12-11 09:49:33 | 1308051635863289906 1009 | 1308051635863289906 | 1285270256981377077 | humble.trading | Great work and thanks for sharing. Correct me if I am wrong, what you are saying from your testing is that: a) block selection (particularly price blocks) are critical to success. b) complex approaches to selection don't add much value as simple approaches work just as well c) a selection process that utilizes 1 workflow that optimizes the selection criteria using OOS and cross validating that selection criteria on other OOS seems to have merit | 2024-12-11 09:49:33 | None 1010 | 1307826891893510196 | 1285270256981377077 | goodday20 | Thanks for explaining. Good work!! | 2024-12-11 09:49:33 | None 1011 | 1307819716718366831 | 1285270256981377077 | coensio | but I agree the small step to overfitting is easily made with this kind of large scale optimization | 2024-12-11 09:49:33 | None 1012 | 1307819416343543822 | 1285270256981377077 | coensio | if this doesnt work it goes to trash... | 2024-12-11 09:49:33 | None 1013 | 1307819175108149308 | 1285270256981377077 | coensio | no there is only 1 model trained on 1 workflow, and CVed on other shifted workflows | 2024-12-11 09:49:33 | 1307818612765233183 1014 | 1307818720088817815 | 1285270256981377077 | goodday20 | this is the magic of overfitting | 2024-12-11 09:49:33 | 1307818421559365704 1015 | 1307818612765233183 | 1285270256981377077 | goodday20 | ok. so you would create the optimizaed model for selection for each workflow. and if it fails on the 2nd TOOS, the you will throw this workflow and start again afresh. I was thinking the optimized selection model is same across all workflows you want to test | 2024-12-11 09:49:33 | 1307816915464486952 1016 | 1307818517453738027 | 1285270256981377077 | coensio | to be clear training is still done only using 1 run per model | 2024-12-11 09:49:33 | None 1017 | 1307818421559365704 | 1285270256981377077 | coensio | just for fun I did some yolo-cross-validation on all available data at once with recursive optimization, to see what would I get best case, crazy numbers...see PDF report. | 2024-12-11 09:49:33 | None 1018 | 1307816915464486952 | 1285270256981377077 | coensio | basically I would throw this idea away...manually or automatically (which is already happening in optimization model, CV step) | 2024-12-11 09:49:33 | 1307814626074824845 1019 | 1307816449670254623 | 1285270256981377077 | coensio | splitting could create unbalanced groups where one has mostly one type of strategy, this is unwanted...and moreover when taking 2nd TOOS, not only TOOS is shifted, but also the whole development process to a brand new market conditions, so we CVing the whole process on different data. | 2024-12-11 09:49:33 | 1307814333652402367 1020 | 1307814626074824845 | 1285270256981377077 | goodday20 | Also what do you do if the optimized selection doesn't work as well on CV on second TOOS (ref: "If the optimized selection model performs well during cross-validation on the second TOOS period,...") | 2024-12-11 09:49:33 | None 1021 | 1307814333652402367 | 1285270256981377077 | goodday20 | one comment/question. why use one TOOS for parameter estimation and another for for CV.. we can split strategies from one TOOS period (or more if we like) into k-fold and do parameter and CV this way. Any benefit in the way you are doing? | 2024-12-11 09:49:33 | None 1022 | 1307812136202014810 | 1285270256981377077 | goodday20 | very nice results. This process is becoming better every day!! | 2024-12-11 09:49:33 | None 1023 | 1307748046917865552 | 1285270256981377077 | coensio | BTW: this has been tested with: normal optimization, random forest and manual method. Worked every time with very similar outcomes. | 2024-12-11 09:49:33 | None 1024 | 1307746707710410843 | 1285270256981377077 | coensio | | 2024-12-11 09:49:34 | 1307746120180699276 1025 | 1307746120180699276 | 1285270256981377077 | coensio | ok guys, strategy selection model training using cross-validation methods, which I previously discussed, appears to be working extremely well so far. This approach utilizes two workflow runs and two TOOS periods to 'design and validate' the selection model: - One TOOS/OOS period for selection parameters optimization - Another run (TOOS period preferably under different market conditions) for cross-validation of the chosen selection approach. If the optimized selection model performs well during cross-validation on the second TOOS period, there is a very high probability that this selection method will perform effectively across all other TOOS runs (hopefully including live trading period). The process doesn't necessarily require complex implementations like large parameter optimizations or random forest modeling. You can achieve good results by visually analyzing which types of strategies perform consistently well across both selected runs. And here, besides of standard selection parameters, targeting "good" price indicators is the key. No wonder why builders with minimized block selection work so well. See attached screenshots. Notes: in this analysis each workflow runs collected >5k unique strategies (it took up to 70 full loops per workflow as I collected data in batches of 500 final strategies, after each batch IP was deleted). Moreover, the SR is based on "continuation of IS performance" where TOOS PASS/FAIL is analyzed per each single strategy separately (much harder to pass). | 2024-12-11 09:49:34 | None 1026 | 1307620783660601425 | 1285270256981377077 | ar3s7 | Copy. Ill need to check my settings to see what I might have missed | 2024-12-11 09:49:34 | None 1027 | 1307605597868265532 | 1285270256981377077 | coensio | yes you should see detailed progress | 2024-12-11 09:49:34 | 1307590622831579240 1028 | 1307590622831579240 | 1285270256981377077 | ar3s7 | random question, with running workflows, does the cmd prompt hang here while the WFs are running or do we see some kind of progress? | 2024-12-11 09:49:34 | None 1029 | 1302903977104375808 | 1285270256981377077 | coensio | the order is y,x | 2024-12-11 09:49:34 | 1302903730936746014 1030 | 1302903730936746014 | 1285270256981377077 | eliasmi | in the Metric Finder there is what i think is the SR after filters and the percentiles. what is the order of the percentiles? eg below for Sharpe Ratio and Annual % return is it 95th percentile Sharpe Ratio and 80th Annual return or the other way around? | 2024-12-11 09:49:34 | None 1031 | 1302233458286002217 | 1285270256981377077 | energyburns | Oh like comparing for research what is best to use? | 2024-12-11 09:49:35 | 1302227450109104209 1032 | 1302227450109104209 | 1285270256981377077 | eliasmi | The SR tabs | 2024-12-11 09:49:35 | 1302046195467817081 1033 | 1302046195467817081 | 1285270256981377077 | energyburns | Is there work on other tabs? | 2024-12-11 09:49:35 | 1301633107992248361 1034 | 1302028709578018816 | 1285270256981377077 | coensio | yes you can work on other tabs | 2024-12-11 09:49:35 | 1301633107992248361 1035 | 1301633107992248361 | 1285270256981377077 | eliasmi | If i use process automation to run the workflows can I still work on the other tabs once a few of the runs are done or do I need to wait until the whole process finishes? | 2024-12-11 09:49:35 | None 1036 | 1300043695621541969 | 1285270256981377077 | bearish4ever | okay | 2024-12-11 09:49:35 | None 1037 | 1300039900027949057 | 1285270256981377077 | coensio | no I do not see any need for a video about this right now, if there is any maybe someone else will do it this time | 2024-12-11 09:49:35 | 1300032856113811527 1038 | 1300032856113811527 | 1285270256981377077 | bearish4ever | Regarding this task because of family matters Have been away for 6 months will you release a video of this new feature into the course? would be great to be in action again | 2024-12-11 09:49:36 | 1293099656997503078 1039 | 1294270042799603722 | 1285270256981377077 | coensio | | 2024-12-11 09:49:36 | 1294270005940326543 1040 | 1294270005940326543 | 1285270256981377077 | coensio | Do not forget to enable this feature: | 2024-12-11 09:49:36 | None 1041 | 1294206674391011359 | 1285270256981377077 | ar3s7 | yes, that worked. thanks guys | 2024-12-11 09:49:36 | None 1042 | 1294178641340796959 | 1285270256981377077 | coensio | yes try to put some IS/TOOS data from previous versions or generate them by yourself then restart the app | 2024-12-11 09:49:36 | 1294057275576684644 1043 | 1294058283904139264 | 1285270256981377077 | ar3s7 | ah, that makes sense. thanks mate | 2024-12-11 09:49:36 | None 1044 | 1294057973684895796 | 1285270256981377077 | tonim2802 | Like when you clear the csv from the folders and the workflow didn't end. | 2024-12-11 09:49:36 | None 1045 | 1294057743950417971 | 1285270256981377077 | tonim2802 | Didn't run V8 but I have this error when there aren't any csv files for the runs | 2024-12-11 09:49:36 | None 1046 | 1294057275576684644 | 1285270256981377077 | ar3s7 | hey guys i am testing V8 and getting the following error when running the workflow validation script. any ideas? | 2024-12-11 09:49:36 | None 1047 | 1293813404359725096 | 1285270256981377077 | coensio | | 2024-12-11 09:49:36 | 1293813369240817715 1048 | 1293813369240817715 | 1285270256981377077 | coensio | Latest version V08. What's new?: 1. Added Run Workflows button, so you can run the generated workflows directly from the dashboard using SQX CLI. | 2024-12-11 09:49:36 | None 1049 | 1293481061476728913 | 1285270256981377077 | coensio | yes this will work for everything, I will try to release it today | 2024-12-11 09:49:36 | 1293448443523760205 1050 | 1293448443523760205 | 1285270256981377077 | ethan83_ | tried reading alot , but this most also work for our study runs + 10 min save databank or have you tried that aswell ? | 2024-12-11 09:49:36 | None 1051 | 1293186892442898515 | 1285270256981377077 | ethan83_ | ever told you are great | 2024-12-11 09:49:37 | None 1052 | 1293100827644268574 | 1285270256981377077 | coensio | The ultimate goal, at least mine, is to also embed some kind of easy base workflow configuration interface so we can almost completely skip SQX GUI and use it only as a processing engine. This will allow you to re-configure our base workflow in many different ways, and test it along with strategy selection in one place. | 2024-12-11 09:49:37 | None 1053 | 1293099656997503078 | 1285270256981377077 | coensio | At least during the DD weeks I become very productive So in the next V08 version we will be able to run all generated workflows directly from our dashboard. This is possible using SQX CLI without its buggy GUI. | 2024-12-11 09:49:37 | None 1054 | 1292396149235257395 | 1285270256981377077 | coensio | It's a numbers game. If you do it like that you just shifting the left tail border more to the left, so decreasing your chances of making money from selected strategy group. Look at the avg and med values, as this tells you more how bad it is. If the left tail is 'thin' then you still have big chances to make some $. We should probably also add 1st 2nd 3rd sigma calculations, but you can also see it in the histograms. | 2024-12-11 09:49:37 | 1292387484407435328 1055 | 1292387772509720647 | 1285270256981377077 | ethan83_ | or is it cheating | 2024-12-11 09:49:37 | None 1056 | 1292387654750703668 | 1285270256981377077 | ethan83_ | this is MNQ.D | 2024-12-11 09:49:37 | None 1057 | 1292387627013771264 | 1285270256981377077 | ethan83_ | | 2024-12-11 09:49:37 | None 1058 | 1292387484407435328 | 1285270256981377077 | ethan83_ | just a question as all now its bit of trouble finding like market and limit orders startegies at like MNQ.D intraday atleas for me. Is it correct to lower expectations and raising the worst year to like 1200 to get run 1 to pass ? , Run 2 i had all green but it feels hard using same thresholds as stop orders ? | 2024-12-11 09:49:37 | None 1059 | 1291669499086508052 | 1285270256981377077 | _kaizen1_ | Thanks | 2024-12-11 09:49:37 | 1291275165908340776 1060 | 1291309058577535007 | 1285270256981377077 | circler. | exactly.. green line = median and blue line = average | 2024-12-11 09:49:37 | 1291181738780790889 1061 | 1291275165908340776 | 1285270256981377077 | coensio | type="line", x0=results[workflow_number]['avg_toos_net_profit'], x1=results[workflow_number]['avg_toos_net_profit'], y0=0, y1=1, xref="x", yref="paper", line=dict(color="blue", width=2, dash="dot") # Blue dotted line for avg | 2024-12-11 09:49:37 | 1291157956783505579 1062 | 1291157956783505579 | 1285270256981377077 | _kaizen1_ | What are the blue and green dash line? | 2024-12-11 09:49:37 | 1289931780400353291 1063 | 1291065054824763477 | 1285270256981377077 | coensio | | 2024-12-11 09:49:37 | 1291065010919047303 1064 | 1291065010919047303 | 1285270256981377077 | coensio | Version V07. What's new?: **1.** I introduced a simple "Selection Score" column that shows the gain in SR % for the chosen selection method. If it's negative, the selection performed worse than the base workflow. **2.** Added Tab 4: Metrics Finder, which helps you discover interesting combined selection metrics for a selected "Run." This allows you to validate without curve fitting on all other "Runs." **3.** Thanks to we now have the option to display distribution histograms using the "Show Histograms" button. | 2024-12-11 09:49:37 | None 1065 | 1290260153400168573 | 1285270256981377077 | tonim2802 | I don't select randomly, best 100 by my criteria, portfolio correlation check, rinse and repeat | 2024-12-11 09:49:38 | None 1066 | 1290255094771748884 | 1285270256981377077 | edau87 | lol so is everybody but me just randomly selecting? | 2024-12-11 09:49:38 | 1290148960924598323 1067 | 1290216193487274027 | 1285270256981377077 | coensio | Thanks! I will merge this idea with the next upcoming release. | 2024-12-11 09:49:38 | 1290214146096369705 1068 | 1290214146096369705 | 1285270256981377077 | circler. | Sure. The script adds Net Profit/DD TOOS histograms for each Run. | 2024-12-11 09:49:38 | 1289953272341336174 1069 | 1290148960924598323 | 1285270256981377077 | eliasmi | I am not sure you would get any statistical significance even if you run something 500 times unless you can adjust for the changes in the underlying market. The easiest way to overcome this is using shorter data windows so the results are more relevant to current conditions. Lately I have been considering that it may be best to confine the data window to medium cycles 3-5 years as the markets are getting faster | 2024-12-11 09:49:38 | None 1070 | 1290145553015115830 | 1285270256981377077 | eliasmi | Probably because you are using RR either directly or indirectly or the highest ret/dd to select the strategy. | 2024-12-11 09:49:38 | 1290019185061265600 1071 | 1290020183879974952 | 1285270256981377077 | goodday20 | however if over 100s of builds your selection method is beating average more than say 90% of the time, there is good chance it has some value add | 2024-12-11 09:49:38 | None 1072 | 1290020068490481766 | 1285270256981377077 | goodday20 | by randomness, i meant that random selection did better in 21 and 23 while worse in 22 and 24 (compared to your selection method) is really just showing random result. if you repeat this process 1000s of times, you see random will be converging to average and yours will obv stay the same. i think if someone has money, they can select all those with correlation < threshold (say 0.6) and trade all of them. | 2024-12-11 09:49:38 | 1290019185061265600 1073 | 1290019185061265600 | 1285270256981377077 | edau87 | I wish I understood why my selection does better in more volatile markets. | 2024-12-11 09:49:38 | 1290017685245001808 1074 | 1290017685245001808 | 1285270256981377077 | goodday20 | yeah this is just randomness. if you run the random selection enough times, it will converge to average. you dont have to run the full workflow everytime.. just to random selection part and see. | 2024-12-11 09:49:38 | 1289944328441172050 1075 | 1289994764766416906 | 1285270256981377077 | coensio | indeed, as our final goal is to make some $ in TOOS I have now few new ideas to implement, to make it even better or at least easier to use. | 2024-12-11 09:49:38 | 1289981241633804362 1076 | 1289981241633804362 | 1285270256981377077 | humble.trading | simplifies things to a degree as well | 2024-12-11 09:49:38 | 1289960033735671848 1077 | 1289961315065921626 | 1285270256981377077 | coensio | In this case it will be very hard to rate or compare any given selection metric, I guess this will be very different for weaker workflows or workflows on other markets, more tests are required. | 2024-12-11 09:49:39 | None 1078 | 1289960033735671848 | 1285270256981377077 | coensio | And this is exactly what I thought....the bottom rating is SR from IS DB (without selection) | 2024-12-11 09:49:39 | 1289912494462337024 1079 | 1289953858524549184 | 1285270256981377077 | edau87 | Stellar example of adapting the process and method to what you consider to be acceptable. Great work. | 2024-12-11 09:49:39 | 1289931780400353291 1080 | 1289953272341336174 | 1285270256981377077 | coensio | BTW would you be willing to share this implementation so we can take this for the next release? | 2024-12-11 09:49:39 | 1289931780400353291 1081 | 1289947944224555069 | 1285270256981377077 | humble.trading | yes and diversify your regime methodology | 2024-12-11 09:49:39 | 1289947802855538739 1082 | 1289947802855538739 | 1285270256981377077 | edau87 | I'd use as many regimes as possible. | 2024-12-11 09:49:39 | None 1083 | 1289947720475480177 | 1285270256981377077 | edau87 | Interesting!!! | 2024-12-11 09:49:39 | 1289947313350901791 1084 | 1289947313350901791 | 1285270256981377077 | humble.trading | that has helped me hugely... I'm now trying to diversify my regime methodologies to get more diversity. Looking at using this methodology that is moving average based to define the regimes. https://www.amazon.com/TRENDadvisor-Guide-Breakthrough-Profits-Financial/dp/0471751472/ref=sr_1_1?crid=2N7K M6AZVP4LK&dib=eyJ2IjoiMSJ9.TN3ZycXzUyARTFTSnzsHHA.kNKyfKZemvwsSrlbpJZE9lE331pbRIYpWDs-hHS3lyE& dib_tag=se&keywords=trendadvisor&qid=1727617788&sprefix=trendadvisor%2Caps%2C115&sr=8-1 | 2024-12-11 09:49:39 | 1289946558271328276 1085 | 1289947112339148810 | 1285270256981377077 | edau87 | I'll probably be paring down my portfolio in December. Some regimes are just trash to trade. | 2024-12-11 09:49:39 | None 1086 | 1289946669022052405 | 1285270256981377077 | edau87 | Otherwise, two similar strategies at 0.60 loser correlation beats one strategy with two contracts. That's a perfect 1.0. | 2024-12-11 09:49:40 | None 1087 | 1289946558271328276 | 1285270256981377077 | edau87 | Regime and directional templates are where you diversify. | 2024-12-11 09:49:40 | None 1088 | 1289946453401141309 | 1285270256981377077 | humble.trading | i hear ya... really digging into portfolio construction is my next hurdle... | 2024-12-11 09:49:40 | 1289945769570471977 1089 | 1289946154959765576 | 1285270256981377077 | edau87 | That statement needs context though. We had a huge reversal day in September. I told you guys it was the worst trading day I've ever had. I also said it was 1/6th my biggest trading day ever. Overall, it was a little less than my average good day. | 2024-12-11 09:49:40 | None 1090 | 1289945769570471977 | 1285270256981377077 | edau87 | I'm clustered as hell on winning days...but losing days not so much. Big reversal days are painful though. | 2024-12-11 09:49:40 | None 1091 | 1289945619493818438 | 1285270256981377077 | edau87 | I try to keep correlation of losers below 0.50 or 0.60 on a weekly basis if I take multiple | 2024-12-11 09:49:40 | 1289945458784993341 1092 | 1289945458784993341 | 1285270256981377077 | humble.trading | I'm leaning towards taking 3-5 if one has enough capital | 2024-12-11 09:49:40 | 1289945105716744233 1093 | 1289945354036318220 | 1285270256981377077 | edau87 | I'm starting to think 2022 is like 2020: very volatile and, therefore, very easy to trade. | 2024-12-11 09:49:40 | None 1094 | 1289945105716744233 | 1285270256981377077 | edau87 | The easy answer is to take two strategies from each run and not worry about it. | 2024-12-11 09:49:40 | None 1095 | 1289944434401742898 | 1285270256981377077 | edau87 | That's only across 10 runs of random selection on each year though. | 2024-12-11 09:49:40 | None 1096 | 1289944328441172050 | 1285270256981377077 | edau87 | Random strategy selection still outperforms my average and median values. Interestingly, my selection method vastly outperforms random in 2022 and 2024 TOOS....but random selection vastly outperforms my selection method in 2021 and 2023 TOOS. | 2024-12-11 09:49:40 | None 1097 | 1289939667017404509 | 1285270256981377077 | humble.trading | That's a great way to think about it. You guys have definitely helped me get pointed in the right direction that's for sure. | 2024-12-11 09:49:40 | 1289938647847534738 1098 | 1289939352675160119 | 1285270256981377077 | edau87 | Plus, trading isn't one size fits all. What is acceptable to me is acceptable to me and that's it. I don't want people following me because they don't have the experience trading the way I do which has made me okay with how I do things. | 2024-12-11 09:49:41 | None 1099 | 1289938647847534738 | 1285270256981377077 | edau87 | It's a balancing act. I'm coming up on a year working under this method. I still get the urge to blow my process up and do everything completely different depending on a random idea I'll have. I try to steer my help towards inspiring that kind of curiosity because that is how you really build something special. | 2024-12-11 09:49:41 | 1289931940115517517 1100 | 1289937864557334630 | 1285270256981377077 | coensio | great idea | 2024-12-11 09:49:41 | 1289931780400353291 1101 | 1289931940115517517 | 1285270256981377077 | humble.trading | I totally agree with both you guys. Also some people contribute more than others. I would give you and any help you need as you guys have given so much to the community. | 2024-12-11 09:49:41 | 1289931425314770944 1102 | 1289931780400353291 | 1285270256981377077 | circler. | i also played with the code yesterday and added new cols in order to add net profit TOOS histograms. The answer about allowing some DD in TOOS and be fine with net_profit > "-500" made me think and realise that all that matters is truly the distribution which can be visualized. and its fine to have some strategies being on the left side of the curve as long as most data points lie in the right positive area | 2024-12-11 09:49:41 | None 1103 | 1289931535964831744 | 1285270256981377077 | coensio | And moreover it is your work | 2024-12-11 09:49:41 | None 1104 | 1289931425314770944 | 1285270256981377077 | coensio | That's is also my mindset, but what is implying this is still introductory course priced at $297 not at $10000 | 2024-12-11 09:49:41 | 1289930934501642261 1105 | 1289930934501642261 | 1285270256981377077 | humble.trading | I'm happy to help others who have a specific need and are stuck. | 2024-12-11 09:49:41 | 1289930254621610094 1106 | 1289930254621610094 | 1285270256981377077 | edau87 | As collaborative as things are around here, stuff like this isn't always the best to share. I don't share any of my custom Python solutions because ChatGPT and I made them based on how my mind works. Past a certain point, your effort deserves to be put on improving your trading. | 2024-12-11 09:49:41 | 1289927541808697496 1107 | 1289929057282293883 | 1285270256981377077 | coensio | ok I must go now, so many new ideas...back to work | 2024-12-11 09:49:41 | None 1108 | 1289928991242719242 | 1285270256981377077 | coensio | it's a real success formula | 2024-12-11 09:49:41 | 1289928716012748840 1109 | 1289928716012748840 | 1285270256981377077 | humble.trading | true... my bugs piled on top of the sqx bugs.... lol | 2024-12-11 09:49:41 | 1289928582344478720 1110 | 1289928582344478720 | 1285270256981377077 | coensio | the more complex it gets the more bugs you will get | 2024-12-11 09:49:41 | None 1111 | 1289928446344167517 | 1285270256981377077 | coensio | you can start with something simple then upgrade if there is need for it. | 2024-12-11 09:49:41 | None 1112 | 1289928248741990481 | 1285270256981377077 | humble.trading | there is a free version | 2024-12-11 09:49:42 | None 1113 | 1289928207608315907 | 1285270256981377077 | coensio | looks professional | 2024-12-11 09:49:42 | 1289927997264232518 1114 | 1289928067262840895 | 1285270256981377077 | humble.trading | https://www.microsoft.com/en-us/power-platform/products/power-bi | 2024-12-11 09:49:42 | None 1115 | 1289927997264232518 | 1285270256981377077 | humble.trading | That's why i was looking at Power BI. It lets you create and publish reports that are interactive. You can drill down into the detail of the report. | 2024-12-11 09:49:42 | None 1116 | 1289927778019577916 | 1285270256981377077 | coensio | sure, it's your work and you decide with who you want to share it... | 2024-12-11 09:49:42 | 1289927541808697496 1117 | 1289927541808697496 | 1285270256981377077 | humble.trading | I have the version of the db working. I'm still testing it though. I'm happy to share pieces of the automation process I created. I just don't want to get into supporting the code for everybody. It's over 5k lines of code for the whole process. | 2024-12-11 09:49:42 | 1289926756035198976 1118 | 1289927017562898514 | 1285270256981377077 | humble.trading | exactly... but what is the best way to select the basket? I've been looking at correlations between the strats, ensuring a diversity of indicators, etc | 2024-12-11 09:49:42 | 1289926489248235601 1119 | 1289926756035198976 | 1285270256981377077 | coensio | yes exactly preferably in a remote DB | 2024-12-11 09:49:42 | 1289926572329140346 1120 | 1289926572329140346 | 1285270256981377077 | humble.trading | that's why I think it's important to start looking at aggregate results across builds. If a selection approach works across hundreds of builds then it would imply that it's quite robust. | 2024-12-11 09:49:42 | 1289925531319668736 1121 | 1289926489248235601 | 1285270256981377077 | edau87 | I would do this before adding contracts all day long. | 2024-12-11 09:49:42 | 1289922387000819722 1122 | 1289925531319668736 | 1285270256981377077 | coensio | and that

SQLite Database Export PDF

Document Details

Tags

Related

Summary

Full Transcript

Upgrade to continue