Why Quantitative Trading Blows-Up Eventually

Many quantitative-algorithmic trading strategies work well for years before making huge losses in a single day. This is not random and must be expected.

The boom and bust cycle for normal assets like stocks is well known by now. I will argue that a very similar thing occurs for algorithmic trading strategies (also known as quantitative trading or black-box).

Stocks go up and a few people make money. Stocks go up some more and people who are not invested start thinking that they might miss out on something. As time goes by, these people also start investing and drive prices up. There are more and more people invested in stocks. And more and more money gets invested into the stock market at higher and higher prices. It is a self-fulfilling prophecy.

The prices of the stocks can only go up as long as more and more people invest into stocks and are willing to pay higher and higher prices. At some point in time, there are not enough additional new people investing and the market starts to dip, sometimes quite fast and abruptly.

There is a very similar boom and bust cycle at work for many algorithmic trading strategies.

Algorithmic trading strategies are based on some pattern occurring in the markets that they take advantage of.

For the sake of simplicity, let’s make a simple example of a pattern: when prices of a stock have gone up 2 days straight, then they will also trend upward during day 3. The stock price will be higher on noon than when it started trading in the morning, and higher at closing than at noon. The strategy based on this pattern is then to buy during day 3 and sell later on in day 3.

Such simple strategies where not uncommon and can be found in investing books prior to the 1980s, when computer trading was not yet that prevalent and the markets where functioning more simple in terms of patterns. Many simple patterns are discussed in books on “technical analysis”.

Opening a position in an algorithmic trading strategy reinforces the underlying pattern that the strategy is designed to take advantage of. If an investors buys in the morning trade of day 3 a stock, then this pushes prices up slightly. And this is exactly the pattern that the investor is using for his trading strategy. By him opening a position, the pattern happened because of himself. The stock is trending higher on day 3.

Closing the position reinforces the opposite of that pattern. Let’s call it the “opposite-pattern”. By selling (closing) his position, the investor pushes the prices down. This is contrary to the underlying pattern that the investor is betting on. If enough of the investors close their positions in the afternoon of day 3, then the prices do not trend upwards any more in the afternoon of day 3, but down.

This makes it possible that the same thing happens with algorithmic trading strategies as discussed above for general stock market investing.

A few people make money following an algorithmic trading strategy based on a pattern that occurs due to some reason. Some other people discover the same trading strategy and also start investing. More and more people invest according to the same trading strategy.

And because investing according to the trading strategy reinforces the underlying pattern, the trading strategy continues to work and may even work better and better.

This works as long as more and more people open positions according to this trading strategy.As soon as more people are closing their positions in the trading strategy than are opening new positions, the underlying pattern stops to exist. And when the pattern does not exist anymore, then all the other investors will take their money out very fast. This can happen within 20 minutes to half a day.

As the closing of positions creates the opposite of the underlying pattern, all but the first to take out their money will make losses. As many investors close their positions, the opposite-pattern can be strong and the losses huge. And it all happens so fast.

This is why there is an inherent blow-up risk to quantitative strategies, algorithmic trading, black-box trading, or whatever you call it.

 

This process can take years, as for example, one quantitative hedge fund after another starts opening positions according to some pattern until after years there is no additional hedge fund opening new positions. Maybe the money flow to the hedge fund industry has dried up generally and that is why there are no new open positions being created. And then the pattern does not exist anymore like it has existed for all these years and one fund starts closing his positions. And then it cascades and blows-up.

 

What can be done about it? An investor must be one of the first to get out of the market. He must get out, before the trading strategy stops working, because then it is too late. This means, stopping to trade according to the trading strategy while it is still working.

How to know when to get out? An approach could be to measure the amount of additional money being invested into the trading strategy by all investors taken together (new opening of positions vs closing). When that amount does not grow as fast any more as it used to, then it is a red sign. And how do you measure the amount invested? I do not know.


Posted on
Last updated on



Leave a Comment
Name (use whatever you like)
E-Mail (optional and not required), will not be published

Refresh
Numbers and letters from picture above (required)