Backtesting Explained: Why It’s the Key to Trading Success

Backtesting is a key term in trading. It is one that is often mentioned but it means different things to different people. Some would class paper trading as backtesting whereas others will want to go back many seasons to gain confidence in how their strategy performs. This article will explain all about backtesting and explain why it is key to your trading success. We will use examples to increase your understand of how to backtest effectively.

What do you backtest?

Typically you will have created a strategy using defined criteria before needing to backtest. This is not always true however. You may want to backtest all result over a few seasons in a particular league and see what looks profitable. This can then be the starting point for a new strategy. There is no right or wrong way to backtest. It is simply a means to tell you how games have performed in the past. Whilst there is no guarantee that future games will exhibit the same traits, it gives you a certain amount of confidence.

Be careful not to backfit

Backfitting is a method of positively removing losing criteria in order to falsely inflate the success rate of a backtested strategy. For example, if you were looking at the backtesting results of an over 2.5 goal market and saw that Arsenal never had over 2.5 goals when they were at home in November, this would be classed as backfitting if you discounted all games in November. Your are forcing a criteria which is too specific and not natural. By backfitting, you are penalising yourself as it inflates the effectiveness of the given strategy which will not give the desired results when used on future games.

How to test a strategy

We are going to use a fictitious strategy created in CGMBet. In this strategy we want to find games where both teams to score is false. To avoid conflicting leagues, we are just going to focus on the English Premier League. We will look at data from 21/22 up until the current date.

We have decided on some criteria to select our matches. We can see that out of the original 1620 games, we now have 92 that meet our criteria. Not many, probably not enough but it gives us something to backtest. We will take these games into the CGMBet AGS module for further analysis.

A quick look at the data tells us straight away our strategy is poor. It has a 45.65% strike rate, which in some strategies will give you profit, but not for this one, the yield is negative 11.37%. In fact, using the criteria we entered for both teams not to score, we have succeeded in turning a profit on both teams to score.

Although this isn’t what we were looking for, it is always worth checking other markets which may be profitable with your current criteria. As for our strategy, we go back to the Upcoming module and we’ll try to improve it.

It turns out that the Premier League is hard to find a definitive set of criteria to be profitable for both teams not to score. We turn to another CGMBet module, Reverse Engineering. This module came to the rescue and we are able to turn a profit in backtesting on this market.

We now how more than twice as many games as the first criteria we used. We can see that the strike rate has only slightly improved to 49.59% of games. The new selection criteria though is picking games that are a better fit for this strategy. We now have a positive 5.69% yield. You don’t always need the high strike rate to be profitable.

Key takeaways

Backtesting with software is essential if you really want to improve your strategies and turn them into winners. Keeping records in a spreadsheet does not give you the flexibility and depth to do the kind of analysis we have done in this example. In fact, CGMBet is the only tool that would have got you the profit we did. I was struggling to manually find any profit in the both teams not to score market. I had to turn to Reverse Engineering which gave me one piece of criteria to turn this around. That is why I constantly recommend CGMBet to viewers on Youtube. There are so many modules that work together to give you the answers you require.

The video below talks us through another backtesting example using CGMBet. Please feel free to comment here or on the video itself.