IT in Financial Markets

Overview of IS4228 Week 9: Information in Financial Markets

• This week focuses on information technology in financial markets, covering topics, including CAPM, insider trading, and AI/ML applications.

Information Technology and Finance

• Machine learning and social networking are information technologies used in financial markets.
• Information technologies support financial infrastructure, including banks, stock markets, payment systems, Defi and Cefi.

Singapore Fintech Landscape

• Blockchain and cryptocurrency companies make up 20% of Singapore's fintech sector.
• Payments account for 17%, investment/wealthtech for 13%, and regtech for 10%.

Singapore Fintech Landscape - Virtual Banking

• Virtual banking provides access to financial services without requiring physical visits to a bank.
• SMEs are rapidly adopting virtual banking, indicated by growth rates between 2-4x.

Singapore Fintech Landscape - Digital (Cashless) Payments

• E-wallets are projected to surpass credit card use, accounting for 27% of online purchases by 2024.
• Singapore's PayNow and Thailand's PromptPay enable cross-border payments, followed by Malaysia's DuitNow, with India's UPI integration planned for July 2022.

Singapore Fintech Landscape - Buy Now, Pay Later (BNPL)

• BNPL allows customers to pay in installments.
• The use of BNPL has surged since the pandemic.
• BNPL is particularly attractive for younger consumers.

The Capital Asset Pricing Model (CAPM)

• Investors should select a portfolio on the capital market line.
• A portfolio should include a mix of risk-free securities and the market portfolio.

CAPM Assumptions

• Investors can trade securities at competitive market prices without incurring taxes or transaction costs, and can borrow/lend at the risk-free interest rate.
• Investors hold efficient portfolios, maximizing expected return for a given risk level.
• Investors share similar expectations on volatilities, correlations, and expected returns of securities (homogeneous expectations).

Supply, Demand, and Market Portfolio Efficiency

• With homogeneous expectations, investors identify the same portfolio as having the highest Sharpe ratio, and demand the same efficient portfolio of risky securities.
• The aggregate of all risky securities held by investors equals the tangent portfolio, which also is the portfolio of all risky securities in the market (market portfolio).
• The efficient, tangent portfolios of risky securities equal the market portfolio.

The Security Market Line (SML)

• A linear relationship exists between a stock's beta and its expected return.

Equity Cost of Capital

• Investments with similar beta values share similar risk levels.
• The cost of capital equals the expected return of investments with the same beta.

Implications of CAPM

• The efficient, tangent portfolio of risky securities equals the market portfolio, which is a passive portfolio.
• Under CAPM, trading will not occur unless risk preferences change.

The Grain Trader Example

• A grain trader must decide whether to speculate on grain futures based on a received private report.
• The trader also needs to decide whether to assume future prices already reflect available information, rendering their own information valueless.
• It raises the fundamental question of how traders with rational expectations respond to new, private information.

Common Knowledge and Trade

• Prices are a potentially important source of information.
• Each agent infers information based on market and non-market signals.
• Each agent assumes all other agents fully use available information.
• Any trade occurring is acceptable to all parties because they are acting voluntarily.
• Every agent knows all other agents are rational, thus any trade is acceptable.
• Market conditions are common knowledge, if a trade happens, then its feasible and acceptable.

Assumptions for No-Trade Theorem

• Traders are strictly risk-averse.
• The initial allocation is Pareto-optimal.
• Traders' beliefs are concordant.
• The terms of potential trades are common knowledge.
• All agents are rational.

The No-Trade Theorem

• No trade will occur when all traders are weakly risk-averse, even with private information.
• In this case, the only possible trades are the ones that will make everyone indiffierent from zero trade.
• In this case, the only possible trade is the zero trade.

Numerical Example

• Considers two agents and two possible states, with five possible values for the information partition. – Examines prior, the information partition, and how risk neutrality affects betting decisions.

Naive Behavior vs. First-Order Sophistication vs. Rational Agents

• Models outcomes based on the decisions that all agents will accept the bet.
• The first assumption is P(θ = 2|P₁) > 1/2 and both agents could expect the agent to accept the bet.
• If agents refuse the bet, then it provides information.

Numerical Example: Risk-Averse Agents

• Both agents would prefer not to bet when x equals 3 when they are slightly risk-averse.
• There is "No trade!" in this scenario.

What Prices Reveal

• New private information does not incentivize trades, given certain conditions.
• The primary theorem outlines that the change in market prices sufficiently indicates about all traders' new private information.

Conclusions on Private Information

• Private information alone is socially valueless but it is still valuable to individual who is able to obtain it.
• Any attempt to use new information to speculate would lead to information impounded in prices.
• The change in relative prices is what ultimately reveals all information.

Discussion Points

• Why do traders gather information if they cannot profit from it?
• What if the assumptions do not hold in reality?
• Assumptions to consider include traders belief's, if all information is common knowledge, and if all traders are rational.

Insider Trading (Kyle 1985)

• Inquires how noise trading affects price volatility.
• Explores how private information about stock value becomes incoroporated into market prices.
• Also studies the of value of private information and what determines liquidity in a speculative market.

Intuitions About the Model of Insider Trading

• One informed trader and many noise traders exist in the market.
• The informed trader takes price impact into account and future oppurtunites.
• Noise traders provide camouflage for the insider, who can make profits at their expense.

Results of the Model

• The informed trader integrates private information gradually into the price.
• Quantities traded by noise traders induce a doubling of quantities traded.
• Quantities doubles traded has no effect on prices, thus this action doubles the profits of the insider.

Understanding Market Manipulation

• Stock-price manipulation is classified by basis.
• The basis can be actions that change perceived value of the asset, releasing misleading information, or purely based on trade.
• The person maniupulating the market could be informed or uninformed.

Manipulation with Uninformed Agents

• Price manipulation can be explained by uninformed agents capitalizing on noise trading.
• Uninformed insider may also pretend to be informed.
• In doing so, uninformed exploits the unability of the maket to distinguish uninformed agent trades from insider trades with privated information.

Contrarian Behavior Explained

• Trading against one's information can sometimes be optimal.
• It can be induced with with manditory disclosure of trades of corporate insiders giving them incentives to manipulate.
• The insider can trade against their own information, which will unwind their position in a second period.

Contrarian Behavior Further Explained

• In a duopoly, better informed agents will try and minimize the learning of less informed ones.
• Market manipulation involves informed traders trading with the intention on revealing information.
• Insiders may also trade with contrarian behavior when there is uncertainty about their presence and there is a large number of periods before information is revealed.

Key Takeaways on Trading and Information

• Trades will not occour if all traders are rational and the information structure is common knowledge, even with insider information.
• However, noise trades do exist which can can be used to benifit from trading.
• An insider faces a tradeoff between innediate gain and revelation of information which can affect future gain, resulting in the insider trading carefully.

Statistical Models

• Autoregressive Model: This model uses past values of a time series to predict future values such as stock prices, macro indicators, and sentiment.
• Future values of a time series depend linearly on past values, most recent values have the highest impact.
• Auto Regressive Moving Average Model.

Support vector machine (SVM)

• Support Vector Machine(SVM)is a technique of Machine Learning for data.
• A separation hyperplane is need to classify data with maximum margin.
• This can then be used to select stocks that, exceed the percentage return of the market.
• Addresses the strong non-linear trend of stock data.

Machine Learning Models

• Neural Networks work as well.
• Deep Learning(DL): Model trains data to make predictions.
• Deep Learning can perform difficult task without much human involvement.
• Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), etc. are techniques used.

Statistical Estimates

• Beta Estimation is used to estimate betas.

Types of Machine Leaning

• Deep Learning (DL) uses data to train a model to make predictions from new data
• To perform difficult task without much human involvement.
• selecting a stock that is likely to perform well in future.
• predict market trend.
• Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), etc. can used.

Machine Learning Applications

• Reinforcement Learning (RL): Involves an agent to maximize a numerical reward by interacting with an unknown environment.
• Applications are used for Portfolio construction
• and it is a prediction for integration steps
• Constraints such as liquidity and transactions cost can be included.
• Natural Language processing (NLP): is utilized for financial reports, social media
• NLP provides additional information to prediction algorithms