How to Quant

A machine can beat a human, but combining man and machine is unbeatable.

Hi everyone, this blog is an introduction to the impact of AI, Adaptive Alpha & Automation on Finance, Economics and Basketball. The blog is written from the perspective of a serious size mover and fictional character under the name Adaptive Alph. He has experience from some of the largest and best economies, basketball leagues & managed futures funds, according to both size and track record. In this blog he will talk about how he developed his passions. Later in life he also developed a specific passion for quantitative investments through moving lots of size in the managed futures industry. Other than a brief introduction to Adaptive Alph, this blog will provide some color on the different types of investment processes that falls under managed futures umbrella. The investment process for all managed future programs is highly statistical and ever since Adaptive Alph was a young kid he has loved statistics and how one can use statistics to find advantages both on the court and in the market.

Check this out about Steph: https://www.nytimes.com/interactive/2016/04/16/upshot/stephen-curry-golden-state-warriors-3-pointers.html

Adaptive Alph loves statistics Adaptive Alph’s introduction to statistics came through playing basketball. Basketball is a rapid game with a lot of points and he really liked scoring a lot of points! Adaptive Alph also found it interesting how some shots were worth 1 point, while other shots counted as 2 or 3 points. This is why he always practiced his three point shots because why get 2 points when you can get 3? If anyone is familiar with NBA basketball and fellow size mover, Stephen Curry, you know that NBA might have to expand the 3-point line because shooting 3-pointers at such a high percentage is a form of statistical arbitrage. Adaptive Alph’s passion for basketball guided him to the U.S. when he was 16 years old and he ended up playing high school ball in NJ for two years. As Adaptive Alph grew older, but not taller, he understood that basketball was not the path to make living so he adapted. At university, he studied economics and statistics and that is how he learned about long vol/uncorrelated investment strategies such as managed futures.

What is Managed Futures There are many different types of managed futures programs and each program has a unique investment process used to attack the constantly evolving market beast. In addition to the newest and most exotic managed future strategy, Machine learning, which Alph will talk more about later, the three largest strategies within managed futures is short-term, systematic macro and trend following programs. The latter is by far the largest by AUM size, accounting for roughly 75% of the entire managed futures space. There are plenty of similarities between these three programs. All of them tend to have high turnover and trading cost, which is why they invest in futures.

Futures Contract? A futures contract is a form of financial derivative that is traded on exchanges across the world. The reason for investing in futures is that they are highly cost effective, liquid and regulated, making these high turnover programs more cost effective. We will go into more depth about futures in a future blog.

The Edge of Managed Futures Most MF programs have the ability to take both long and short positions in the market. This is an edge that MF programs have over long only investments as MF programs can make money in both up and down markets through either directional or relative value bets. MF’s also have the ability to take an infinite amount of uncorrelated bets and they have low to zero correlation to a traditional 60-40 fixed income equity portfolio Finally the MF programs have a similar fee structure to hedge funds although there has been a great deal of fee pressure lately.

Data tells quants what to do

Different Types of Managed Futures As mentioned earlier the investment process for MF programs is very different. Short-term trading programs try to capture market anomalies that may exist in the market for just a few seconds. The anomaly can exist because a trader placed an order on one exchange pushing the price up so the short-term mangers try to get in quickly to capture the price increase. A systematic macro program usually takes relative value bets based on macro economic fundamentals. An example would be to short a currency with a low interest rate and invest the proceeds in a currency with a higher interest rate. This would be referred to as a carry trade and is an example of using fundamental factors to capture market anomalies. Finally, there are trend followers trying to capture trends across as many liquid futures markets as possible in order to generate alpha. These trends have been persistent for the past 30 years and they exist because of irrational human behavior. The anomalies also exist because hedgers are willing to pay a premium to insure a certain price level such as an airline trying to hedge oil price.

Super Computer

Adaptive Alpha & AI Finally, there is machine learning and this strategy is different from the other three programs because it uses statistical techniques that are adaptive. By adaptive one means that the algorithm can adapt to changing regimes. An example would be a regime shift in the economy such as QE, which due to flooding the market with liquidity changes relationships between financial instruments. Some of the techniques used have been very successful in making cancer predictions and machine learning algorithms are also used in self-driving cars so why not use them in finance? It is Alph's understanding that successful hedge funds have utilized machine learning both as a stand alone product and in their flagship programs as a diversifying part of the portfolio with huge success for over 10 years. New funds based on adaptive techniques are now constantly being launched. The statistical techniques ranges from dimensionality reduction techniques such as PCA, to self-updating algorithms using Bayesian statistics or neural networks. Each technique has a different edge. Some algorithms may be more successful than others depending on the amount of data available, processing power of the computer used to employ the algorithm and also the problem that needs to be solved. It is hard to define the alpha that machine learning can capture because a lot of the patterns are nonlinear and are therefore conceptually impossible for the human brain to understand. We tend to think in a linear fashion and an example of this is to predict that when interest rates go up bond prices go down. I will go into more depth of machine learning algorithms in the future.

Legendary Size Mover

Future of Managed Futures (LOL) We are entering a super interesting period in managed futures and this despite that some people, such as Warren Buffet, argue that markets have become so efficient that they are unbeatable and that most people would do better by just investing in the S&P 500. Adaptive Alph is not arguing against Mr. Buffet, but a large majority still believes that hedge funds do provide value in a portfolio context, which is why the MF industry still exists. There are basically three ways to beat the market. One can either take a discretionary approach, systematic approach or a hybrid approach. Adaptive Alph believes in the following - A machine can beat a human, but combining man and machine is unbeatable.

Cheers/

Adaptive Alph