Occam’s Razor in Trading: Applying Parsimony in the Development of Trading Strategies

Many of you may have heard about Occam’s Razor  – which is also known as the law of parsimony – which tells us that when confronted with two different choices in any field it is often much wiser to choose the path which contains the least assumptions. In science this principle is widely applied as when we have two different theories which explain the exact same phenomena we would rather choose the theory which requires us to assume the smallest amount of things. Through the following few paragraphs I will talk to you about the law of parsimony in Forex trading and how you can successfully apply it to increase the success and effectiveness of any algorithmic strategies you create. You’ll learn how this principle makes natural sense and why it is so applicable to mechanical trading system development.

Let me first explain what Occam’s Razor is about through an extreme example. Let us suppose you have left your dog inside your kitchen with a cookie on the floor. After you return the cookie is gone and the dog is there. Supposing we couldn’t possibly test the dog to see if it had eaten the cookie we are confronted with a question: What happened to the cookie? Even though the answer in this case seems obvious, the fact that you cannot prove it leads to the fact that an infinite number of possible explanations can be formulated. For example we could say that an alien transported itself to your kitchen and stole the cookie or that the cookie – through an infinitesimally small chance – used quantum tunneling effects to end up somewhere else.

This is exactly where the principle of parsimony comes into play. The first thing we do is try to quantify the assumptions made by the different hypothesis and then we choose the one which makes the least assumptions. In this case it is very easy to see that the alien and quantum tunneling hypothesis require a large set of assumptions (for example that aliens exist, that they can materialize out of thin air and that they for some reason wanted a cookie or that all the atoms of the cookie somehow tunneled at the exact same time to another location) while the dog hypothesis only requires us to assume that the dog was hungry (and we already know dogs can be hungry and eat cookies while we do not know if aliens exist or if they like cookies and we certainly do not know if such large and coherent tunneling effects are ever to be observed). In the end we take the theory with the least assumptions: you left your dog with a cookie and it simply ate it.

Now that we understand better what Occam’s Razor is about we can now apply it to the field of automated trading system design. Whenever you’re confronted with the design of a trading strategy there are many ways in which you can carry out this process and your choices regarding how to develop a system determine the assumptions that need to be true if your system is going to be profitable. For example if you decide to develop a system which is only profitable during the past 3 years of trading data and you use the 5 minute time frame you’re assuming that the future will always be very similar to the past three years and that there is no significant variability or execution problems which will cause your time frame choice to affect your trading.

It is therefore evident that you can make different choices which change the amount of assumptions which are necessary for your strategy to be successful. For example you can just as easily choose to develop a daily trading strategy that simulates reliable using only daily OHLC data and test it during a 15 year period on several different instruments. In the end you have a strategy that makes a much more limited set of assumptions and therefore this strategy makes more sense when applying the law of parsimony. You cannot possibly know which one of the two strategies will be profitable in the future (akin to testing the dog) but choosing a strategy which makes the least assumptions increases your chances of achieving this outcome (since you have a larger probability space where the outcome you desire would become a reality).

The law of parsimony is in effect the law which allows us to develop trading systems with the highest degrees of robustness. When a person develops a strategy without making any but the most basic assumptions there is a much higher inherent chance of the simulated outcomes repeating themselves in the future. For example in the above mentioned cases for the 20 year strategy to be profitable the market must just behave how it did across some of those 20 years with even bad execution while on the 5 minute strategy not only must the broker behave exactly how it did within only the past three years but it must also offer good execution and it must not alter its feed significantly when compared with the data used for simulations. Certainly for anyone thinking about which strategy to choose using Occam’s razor the answer would be absolutely obvious.

Of course we can make more assumptions and attempt to develop more profitable outcomes but you should always be careful about the extent and scope of the assumptions you’re making because the more you’re making the more the chance of you simply fooling yourself with unrealistic simulations. In my mind there is no reason as why a trader would want to make very strong assumptions when you can make much fewer assumptions and gain similar profitability levels. For example why would you spend a lot of time developing a 5 minute strategy with a ton of necessary assumptions when you can as easily create a portfolio of 1H to daily time frame strategies which yield the overall same or even better results ? When building trading systems do not ask what you would want to build or what you believe can be more profitable but look at evidence and assumptions and choose a development path which follows the law of parsimony. Why would you make a ton of assumptions if you can just as easily get there by making half of them ?

Occam’s Razor is definitely a very useful philosophical principle which applies perfectly to the design of trading strategies and the inherent robustness behind their development. When you develop systems always consider the assumptions being made and create strategies in a way which makes the least possible assumptions to reach the outcomes you want. If the outcome you want requires too many assumptions then you have a good reason to believe that there might be something wrong with it in the first place (for example if what you want can only be achieved assuming very good execution, on a very specific set of parameters, only when back-testing across a couple of years, etc).

If you would like to learn more about the development of algorithmic trading strategies and how you too can learn to build systems based on sound principles please consider joining Asirikuy.com, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach towards automated trading in general . I hope you enjoyed this article ! :o)

Print Friendly, PDF & Email
You can leave a response, or trackback from your own site.

One Response to “Occam’s Razor in Trading: Applying Parsimony in the Development of Trading Strategies”

  1. Rimas says:

    Simplicity rules:)

Leave a Reply

Subscribe to RSS Feed Follow me on Twitter!
Show Buttons
Hide Buttons