What is the Fractal Market Hypothesis

The Fractal Market Hypothesis (FMH) is a theory about how heightened market uncertainty can lead to sudden market crises and crashes. FMH argues that market prices exhibit fractal properties over time, which can be disrupted when the information sets and time horizons of investors change.

FMH, developed by Ed Peters in his 1994 book, Fractal Market Analysis: Applying Chaos Theory to Investment and Economics, is an extension of the widely utilized efficient market hypothesis (EMH). The most glaring problem with quantifying and utilizing the FMH is deciding the length of time that the “fractal” pattern should be repeated in trying to project market direction.


How Does the FMH Work?

The fractal markets hypothesis states that financial markets are similar to fractals in that they display self-similarity across different time scales. In other words, market behavior at any given moment can be seen as a reflection of past behavior. This self-similarity implies that market prices follow a repeating pattern over time.

As per FMH, the stock market prices do not resemble a random walk described by Efficient Market Hypothesis (EMH).

The fractal pattern in stock market prices distinguishes between long-term investors who may focus on market fundamentals and short-term investors who may focus more on technical analysis. Trades by long-term investors help to stabilize the market by providing liquidity to short-term investors and vice-versa.

The most important implication of the FMH is that sudden changes in market behavior can lead to sharp price declines, known as market crises or crashes. These sudden changes usually occur when the information sets and time horizons of investors change. For example, a shift from long-term to short-term thinking by investors can lead to a sell-off in assets and a decline in prices.

The FMH has been used to explain several historical events, such as the 1987 stock market crash,  the 1998 Russian financial crisis, and the 2008 subprime mortgage crisis. However, the most glaring problem with quantifying and utilizing the FMH is deciding the length of time that the “fractal” pattern should be repeated in trying to project market direction.

How to detect Fractal Patterns in data?

While many retail traders try to detect patterns visually by going through hundreds and thousands of stock charts manually, it is a laborious task and often prone to cognitive errors. The fast, scalable, accurate, and systematic approach to detecting patterns is to use timeseries-based pattern recognition algorithms.

Pattern recognition is a branch of machine learning that focuses on the identification and classification of patterns. In the financial world, pattern recognition can be used to identify trends in stock prices, detect unusual activity in trading data, and predict future movements in asset prices.

Timeseries pattern recognition is a type of pattern recognition that is specifically designed to work with data that is arranged in a series of timestamped data like the OHLC price series.

Classification is the process of assigning a label to a pattern. For example, micro trends might be classified as 1, 0, -1 for up, sideways, or down trends. Now the algorithms can be used to learn the repeating patterns of number sequences that are seen during types of crisis events or elevated volatility periods to identify similarities or differences. The classification process is important because it allows traders and investors to quickly identify what type of pattern they are dealing with. It also allows them to make informed decisions about how to act on that information.

Anomalies are something that does not fit the normal pattern. Anomalies can be caused by news events, changes in market conditions, or errors in the data. Identifying anomalies early can help traders and investors detect shifts in market behavior and take action early to avoid potential losses or gain from it.


Example of similar pattern detection between 2008 and 2022 in S&P 500

*****

A lot of emphases is given in the retail trading world to training the eye when it comes to building trading expertise. While that is good theoretical advice to build domain knowledge, it is a bit dated in algorithmic trading, machine trading, and systematic trading, which is what most trading happens in exchanges today. Institutions and hedge funds use algorithms, machine learning, and such advanced techniques to win over retail traders in this zero-sum game of trading. It is important for traders, especially the next generation, to familiarize themselves with the systematic approach to play this game by the new rules.

Researchfin.ai makes it extremely easy for anyone to learn, research, test, and deploy trading systems powered by AI and Machine Learning, irrespective of their trading and coding experience. We will be launching our no-code Trading Systems platform soon for everyone.

If you want to learn the fundamentals of systematic trading, this foundational book by Ruban Phukan, our co-founder and CEO, is a great starting point. It has been a #1 bestseller on Amazon with 5-star reviews!