Financial markets are extremely complex, composed of several layers of reflexive activity behind the scenes. With the rise of alternative data sets and new quantitative methods in machine learning, research can become daunting with so much optionality when accompanied with an endless pool of testable ideas. To better navigate the research process, this webinar will propose a systematic framework to explain market anomalies that occur from the irrational behavior of market participants.
Behavioral Arbitrage
Taking advantage of mistakes that derive from behavioral biases. In general, there are two types:
- Cognitive Bias (e.g. Recency Bias, Confirmation Bias)
Biases that are difficult to notice, often due to misinterpretation or false assumptions. - Emotional Bias (e.g. Loss Aversion)
Biases that are difficult to control, inherently due to genetics during incompatible scenarios.
Timing Market Mistakes
Timing strategies are generally mean-reverting, relying on "corrective" behavior that takes place after the market realizes mistakes in the current asset prices. Event studies are helpful to assess the profitability of "corrective" behavior since it reduces noise from other signals.
- What does the Market think?
What is the current sentiment of the market towards this asset? Is it bullish, bearish or going unnoticed? - Why is the Market wrong?
Based on current information, is the market sentiment correct? Too extreme or just wrong? - How wrong is the market?
What conditions can further explain the difference between sentiment and reality?
What does the Market think?
There are many ways to interpret how the market thinks, but the most simplistic version is to use recent prices. Current prices are based on the consensus of the market participants, whose decisions are affected by human behavior. These behaviors are always subject to biases, and large biases lead to profitable opportunities.
Current Price -> Market’s View
Bid-Ask Transactions -> Current Price
Human Decisions -> Bid-Ask Transactions
Human Behavior -> Human Decisions
Information (Ex. News) -> Human Behavior
Why is the Market wrong?
To uncover why the market is wrong, we need to decide on how to model human behavior, and appropriately structure a hypothesis. Using "Information" as the input, and "Human Decisions" as the output, we can reasonably structure any behavioral driven hypothesis.
Input -> Model -> Output
Information -> Human Behavior -> Human Decisions
The key in any behavioral hypothesis is interpretation - with the same information, better interpretation will lead to stronger models and better predictions.
How wrong is the Market?
As new information is created, the market reacts by adjusting prices. In turn, this price adjustment becomes new information. This reflexive process is the root cause for large gaps between market prices and intrinsic value.
To maximize profits, the goal is to find conditions opportunities where the gaps are wide. The size of the gap depends on the difference between the old behavior and new behavior.
Old behavior involves accumulated bias over time. The gap is largest with continuous reinforcement of the bias over a period of time.
New behavior is completely depends on the sensitivity and dispersion of the new information being released. Sensitive information that travels fast will violently close the gap instead of gradually. The speed of dispersion also helps for finding exit signals.
A Real Example
The notebook source code plus backtest are located: https://www.quantopian.com/posts/behavioral-arbitrage-webinar-notebook-plus-backtest
