RIT Case Solutions-1-1 PDF
Document Details
Uploaded by QuickerLight2270
University of Sydney
Tags
Related
Summary
This document is a case study on investments and portfolio management. It covers concepts like market making, limit orders, risk management, and liquidity risk. The case study is relevant to an undergraduate finance course at the University of Sydney
Full Transcript
lOMoARcPSD|24718830 RIT CASE Solutions-1-1 Investments and Portfolio Management (University of Sydney) Scan to open on Studocu Studocu is not sponsored or endorsed by any college or university Downloaded by Bob Dog (byod201...
lOMoARcPSD|24718830 RIT CASE Solutions-1-1 Investments and Portfolio Management (University of Sydney) Scan to open on Studocu Studocu is not sponsored or endorsed by any college or university Downloaded by Bob Dog ([email protected]) lOMoARcPSD|24718830 LT1 Solution Key This case is designed to familiarize students with market making and submitting limit orders instead of market orders. Specifically, students should learn about the profitability that comes from “making the bid-ask spread” instead of “paying the bid-ask spread”. Students will also learn about risk management with regards to their positions and how to manage that risk by widening/cancelling their orders or being more aggressive to close out their positions. Brokers typically “internalize” as much of their retail trade flow as possible with their internal market making desks or other client orders. While the market-making desk will often need to provide price improvement2 over the NBBO (national best bid-offer), we have relaxed that rule for the sake of simplifying the trading case. The following is an evaluation guideline: Profit > $12,000 Very Good Profit > $6,000 Good Profit > $2,000 Average Profit > -$1,000 Needs improvement Profit < -$1,000 Improper execution In the RIT 2.0 Trader reports, page two will show the stock price, the bid and ask, and all of the execu琀椀ons. A successful trader should have many green and red hollow dots (hollow indicates limit orders ge琀 ng 昀椀lled) on their report. LT2 Solution Key This case builds on the agency trading cases by adding the element of risk/liability trading to the student’s experience. In this simula琀椀on, the student will directly generate pro昀椀ts or losses as a result of their decision making (in agency trading, a bad 昀椀ll gets passed on to the clients, and the client will not be sa琀椀s昀椀ed, but no direct losses are accumulated). This simula琀椀on also highlights liquidity risk by forcing students to trade in an illiquid (ANON) market. Students will be heavily mo琀椀vated to use limit orders as the programmed liquidity traders (ANON) trade quite infrequently and in small sizes. A single trader buying or selling 50,000 shares all at once will cause signi昀椀cant price impact. The buy and sell orders in this simula琀椀on are distributed evenly, that is n/2 students will receive buy orders and n/2 students will receive sell orders. In perfect equilibrium, all traders will either buy or sell 50,000 shares at exactly $10.00 and the market will clear (as will all their inventories that need to be disposed of, that is, worked into the market). In reality, some students will a琀琀empt to get be琀琀er prices by asking for more than $10.00 or bidding less than $10.00. In addi琀椀on, some students will be more anxious to close out their posi琀椀on and be willing to use marketable limit orders and pay a small premium or sell at a discount for the assurance that they will get the trade completed. Many students, speci昀椀cally novice students, will not completely unwind their order by the 2-minute mark. The so昀琀ware will provide an incen琀椀ve for them to cover their orders quickly by imposing a 昀椀ne of $2 per share for exceeding a gross posi琀椀on limit and, at close, for any remaining uncovered posi琀椀on within the posi琀椀on limit. Downloaded by Bob Dog ([email protected]) lOMoARcPSD|24718830 Trading (posi琀椀on) limits are set such that they cannot be long or short more than 50,000 shares at any point in 琀椀me. This prevents a trader from accep琀椀ng a 50,000-share posi琀椀on and then adding to it (or shor琀椀ng 50,000 shares then shor琀椀ng more). Traders cannot submit orders to cross their trading limits, but the ins琀椀tu琀椀onal orders (endowments) automa琀椀cally put into their accounts will breach trading limits if they have outstanding posi琀椀ons. If at the 2-minute mark they have not closed their long posi琀椀on, and they receive a subsequent buy order, they will be 昀椀ned for breaking their trading limit. Instead of being measured against VWAP, traders in this case should be measured based on their accumulated Pro昀椀ts & Losses (P&L). The following is an evalua琀椀on guideline: Pro昀椀t > $12,000 Very Good Pro昀椀t > $6,000 Good Pro昀椀t > $2,000 Average Pro昀椀t > -$1,000 Needs improvement Pro昀椀t < -$1,000 Improper execu琀椀on Further Context on Liability Trading Desks Liability trading desks o昀琀en lose money. Their role on the trading desk is seen as a loss-leader to generate business (i.e. if you’re willing to help the buy side out by facilita琀椀ng ‘hard to trade, illiquid stocks’, they will reward you by sending you more orders for ‘easy to trade, liquid stocks’). In order to mo琀椀vate students, we make the liability trades pro昀椀table by o昀昀ering a rela琀椀vely large liquidity premium. A typical situa琀椀on for a liability trade is a fund that wants to sell a large posi琀椀on (say 1 million shares) of a stock that typically has an average daily volume of 50,000 shares. The fund (through agency traders) has already shopped around to see if anyone wants to buy the 1 million share block but no other buy-side par琀椀cipants are interested in it. So, they sell the block to the liability trading desk at a discount to its current market price1. The liability desk will then unwind the posi琀椀on over the next month (or hold it and hedge their exposure using related securi琀椀es etc.) The LT1 case is a stylized version that mo琀椀vates the same types of risks, albeit over a shorter 琀椀meframe with a 琀椀ghter market. NB: “Liability” trading is also referred to as “Principal” trading, since the trader is ac琀椀ng as a principal instead of an agent. 1 Regula琀椀ons in some countries prevent a block of shares from trading outside the NBBO. The trading desk will buy/sell shares on the open market; as a result the price will move. If there is enough liquidity, the order will be traded at a be琀琀er price compared to the one speci昀椀ed in the block; if not, the trader will cross the order at the speci昀椀ed price in the block. The regulatory rules of how trades like these are handled change frequently. AT1 Solu琀椀on Key Downloaded by Bob Dog ([email protected]) lOMoARcPSD|24718830 This case is designed to introduce students to the RIT trading pla琀昀orm with a rela琀椀vely simple trading task. Purchase 100,000 shares of a stock over 5 minutes. They can execute this trade in any way they wish, and typically, they will try some type of ‘opportunis琀椀c 琀椀ming’ using the price chart as a guide (they believe they can predict a random walk) or simply trying to purchase shares as fast as possible to “get the shares before the rest of the class”. These two strategies will not be successful given that a random walk cannot be predicted, and because the computerized traders provide more than enough liquidity that the class-based trading 昀氀ow will have a minimal e昀昀ect on the market price. Through class discussion or their own observa琀椀on, they should come to the conclusion that the op琀椀mal method to match the market VWAP is to purchase shares intermi琀琀ently through the trading session, and closely match their 昀椀ll por琀椀on to the volume that have traded throughout the day versus the average total volume. While they can’t forecast the movements of the market price, they can use historical volume data to es琀椀mate how many shares will trade throughout the day. Using market orders, they should buy shares so that their cost is a representa琀椀ve sample of all of the trades that occur throughout the day. The data provided shows a total of 12.36 Million shares traded. Showing each segment as a % of daily volume, we observe the following: In 琀椀me slice one (9:30 to 9:40), 6.7% of the daily volume traded. That implies that they should purchase 6.7%, or 6700 shares of their 100,000 requirements, in the 昀椀rst 10 minutes of the day. Since the simula琀椀on treats 5 minutes of trading 琀椀me as a full-day, that interval would be 300 seconds/39 = ~7.5 seconds. Alterna琀椀vely, they could add two 琀椀me slices together and purchase (6.7% + 5.4% = 12.1%) of their posi琀椀on in the 昀椀rst ~15 seconds. The more granular their trades, the more accurately their 昀椀ll will re昀氀ect the VWAP (assuming the actual volume pa琀琀ern in the market matches the daily average). The following is an evalua琀椀on guideline: Within 2 cents of VWAP Very Good Downloaded by Bob Dog ([email protected]) lOMoARcPSD|24718830 Within 4 cents of VWAP Good Within 6 cents of VWAP Average Within 10 cents of VWAP Needs improvement More than 10 cents from VWAP Improper execu琀椀on The pro昀椀t/loss generated in the traders account should be ignored. In a typical market, all of the shares will be transferred directly into the clients account at the prices that they 昀椀lled at, and no real pro昀椀ts/losses will accrue to the trader. (RIT does not simulate this transfer, so P/L will be calculated and displayed). In the real world, all of the shares will be transferred directly into the clients account at the prices at which they are 昀椀lled. Some students will be tempted to buy and sell during the simula琀椀on but this is a behavior that would not be tolerated by clients. If a broker (the student) buys and sells when instructed to only buy, he/she will lose reputa琀椀on since clients will see these transac琀椀ons and realize that the broker was not execu琀椀ng the order as instructed. The actual number of shares purchased should be 100,000 shares. If not, their client will not be very sa琀椀s昀椀ed with the 昀椀ll, regardless of the price. In addi琀椀on, the student’s average cost should be close to the market VWAP for the day. If they are too far from the market VWAP, it likely indicates that the resul琀椀ng ‘be琀琀er 昀椀ll’ was due to chance -- given that the path of market prices over the day is not predictable in this case. Trying to 琀椀me the market in this situa琀椀on implies that the agency trader took on addi琀椀onal risk which will harm their reputa琀椀on as an agency trader. When evalua琀椀ng students on this case, it is important to remind them about the di昀昀erence between the student’s VWAP for the outstanding shares and the student's VWAP for all trades over the day. These two di昀昀erent ways of calcula琀椀ng the student’s VWAP yield the same result if the student was only buying shares, as instructed. However, they will be di昀昀erent if the student also sold shares. The RIT por琀昀olio module shows only the student’s VWAP (cost) for the outstanding shares (for example, the shares that the students have in their por琀昀olio at the end of the trading day). However, the agency trader should be evaluated based on their VWAP for all trades for their client over the day. An example will help us explain this concept more clearly. Suppose a student made the following purchases: His/her VWAP for all trades over the day would be $11.00 [=(50,000x$10+50,000x$12)/(50,000+ 50,000)]. It is the same as the VWAP for the outstanding shares. Consider a di昀昀erent student who had the following transac琀椀ons: Downloaded by Bob Dog ([email protected]) lOMoARcPSD|24718830 His/her VWAP for all trades over the day would be $12.00 [=(50,000x$10-50,000x$12+ 100,000x$13)/(50,000-50,000+100,000)]. However, the VWAP for the outstanding shares would be di昀昀erent and it will be equal to $13 as shown in the table below: At 琀椀me 30, the student has 50,000 shares in the por琀昀olio and he/she paid $10/share. When he/she sells 50,000 shares, his /her posi琀椀on goes to zero therefore his/her VWAP for the outstanding shares is zero. When he/she buys 100,000 shares for $13/share, then the VWAP of the outstanding shares becomes $13. Buy-side clients typically want their trade execu琀椀on to be as close to the VWAP as possible. The value-added that comes from the trading desks is their execu琀椀on op琀椀miza琀椀ons, i.e. being able to source liquidity and trade shares opportunis琀椀cally. When the traders are forced to use limit orders, and only limit orders, they should observe the following: (1) Ge琀 ng 昀椀lled on orders is more di昀케cult and purchasing exactly 100,000 shares is considerably more challenging. (2) Their average cost, rela琀椀ve to VWAP, will be lower if they are just using limit orders. The reason for #1 is because the only 琀椀me their order gets 昀椀lled is when they are at (or near) the top of the order book and a market-sell order comes into the market. If the market is rising, they will have to constantly submit new limit orders to stay near the top of the order book. Their average 昀椀lled price will be be琀琀er because they’re always trading at the bid. Generally, 50% of trades occur at the bid price, and 50% of the trades occur at the ask price. If all of a trader’s buy orders are always 昀椀lled at the bid, they are ‘saving’ the bid-ask spread, and the result is that their average price will be consistently lower (2). Note: the case is designed so that the computerized traders will trade, on average, 12.36 Million shares over the 5-minute case. The individual traders volume will be addi琀椀ve, i.e. if you have 20 students trading, the observed market volume will be closer to 12.36 + (100,000 x 20 traders) = 14.36 million. Downloaded by Bob Dog ([email protected])