Articles in mainstream media constantly praise the returns of some hedge funds and attribute them to quantitative trading. But can quantitative models extract billions of dollars worth of alpha from noisy data? Is it possible that these hedge funds have some secrets and any appeal to quantitative trading is just a smokescreen? In this article we provide some plausible answers.
It is important to understand that trading profits do not come from price series or from quantitative models; they come from the accounts of those who lose money. Trading is a zero-sum game in principle but it is actually a negative-sum game due to trading cost. Investing in stocks may or may not be a zero-sum game depending on how one measures performance.
The justification of zero-sum game is actually pretty simple: There is no way for a group of traders to realize profits unless there is another group of traders that realizes losses. Trading does not generate any new money; it only redistributes existing money deposited in existing accounts. While inflows and outflows are dynamic, there is no way for someone to make a profit unless someone else loses.
Despite its zero-sum game nature, trading is an extremely useful activity because it is the only mechanism that allows price discovery in a free market.
Limits of quantitative trading
Some quantitative hedge funds claim that they realize billions in profits by applying advanced data-mining to large volumes of data that even include satellite images of weather, crop monitoring and even road traffic near shopping malls.
This sounds nice but data-mining suggests models from data and bias is high. In addition, reuse of data leads to data-snooping bias. Unless the number of predictors is kept down to a minimum, the probability of Type I error (false discovery) is high. Actually, when unrelated predictors are used, for example weather patterns and stock index performance, the bias, also known as data-mining bias, is so large that the probability of a Type I error is close to 100%. The error can be minimized if the power of the test is high, which translates to a need of sufficient samples. However, size of sufficient sample is very large in finance due to non-stationary price series.
The above paragraph is worse than Mandarin for the average reader of the articles in financial media that attribute high returns of some hedge funds to quantitative trading. But this is where these articles bet: the subject of quantitative trading is so complex and it is impossible for the average reader to know its perils. The result is that this complexity impresses the reader to think that this is how hedge funds make their money.
How some hedge funds actually profit
In early 1990s I was well-known in a closed circle for my work on developing trading strategies, some of which were already semi-automated. During that time, I was approached by someone who worked for some wealthy investors and was asked if I would help them develop a software program that could determine the probability of some classical chart patterns forming and indicator signals occurring. I thought they needed that to “front-run” those signals. But as I found out, they actually wanted to first contribute to the final realization of the signals but then fade them and profit from pulling the rag from under retail traders’ feet because that was the only way to make a lots of money, given that trading is a zero-sum game.
I refused to participate because I thought that developing software to fade retail traders was not compatible with my ethics. Maybe that was a mistake because these individuals possibly retired in a few years. They probably ruined millions of retail traders, known as “weak-hands.” When a head and shoulders pattern would start forming, for example, they would assist in its final formation by making a small investment but when it was confirmed they would start heavy buying to panic the retail crowd. This strategy involves positive feedback because when confirmation is lost, also known as a throwback, there is retail stampede for the exit. The funds were happy to be the counterparty to losers looking for a quick exit.
We are getting there regarding hedge funds’ secrets but before that it is interesting to mention that while the fading of retail traders was happening, scores of gurus, financed mainly by the brokerage industry, were trying to convince retail traders that accepting losses was showing discipline and were also publishing new books that offered random formations and indicators to use without any statistics or analysis but relying on selection bias, i.e., showing a few examples where these random signals worked. Technical analysis was the ultimate wealth redistribution tool.
Therefore, most hedge funds, CTAs and other professionals, profited during the 1990s and 2000s by ruining retail traders. All this talk about quantitative models is a smoke screen. Actually, it is more profitable fading most of these particular models than using them to trade. It could be the case that even the scientists that work for these funds do not know that. Funds employ many quants but only a few trusted traders. Do the quants know what trades are actually placed? I highly doubt that. I was not allowed to see that when I traded for a hedge fund; I received a fee based on the performance of the fund but I never saw a trade ledger. All trades were passed to the head trader and no one knew what actually happened after that.
Passive investing is a hedge fund killer
I could write a book, probably with the title “Fooled By Hedge Funds”, because the subject has many tangents but I think this article offers the general idea. After the mass extermination of retail traders in the last 15 to 20 years, passive investing popularity increased. Some of the new passive investors are the old ruined retail traders who are now trying to breakeven and have their hopes on Fed maintaining a free market put option for them.
Hedge funds’ problems are increasing as passive investing popularity is growing. There are no longer many retail counterparties to profit from. All this talk about quantitative models is hogwash in my opinion. These models cannot provide alpha to the tune of billions of dollars; they are useful only to retail quant traders with small accounts. Most hedge funds cannot match the performance of SDOG (ALPS Sector Dividend Dogs ETF), up 23.6% year-to-date. So they need more investors and they use returns from the past they know they will not be able to ever realize again to attract investors so they can increase their purchasing power. And I would like to end this article with this
High purchasing power is the only true edge in the markets
This is the true edge of some hedge funds. They can move price and profit. It’s that simple. The rest, especially about quantitative trading, mathematics, rocket science, data-mining and machine learning, is hogwash.
If you have any questions or comments, happy to connect on Twitter:@mikeharrisNY