Thoughts on Trading System Performance

I was recently alerted to some commentary on trading system design and backtesting. That grabbed my attention for reasons you will find out about fairly soon. In the meantime, though, I wanted to chime in on a couple of points that were made in a couple of different places.

Firstly, the guys at trade-ideas software posted in their blog that a good trading system has a win rate of 55% or better (55% or more of trades are profitable) and a ratio of average gainer to average loser of 1.5 or better (the average winning trade is 1.5 times the size of the average losing trade). These would be quite good numbers, but I am not prone to agree that these should be the benchmark by which a trading system is evaluated.

My biggest argument is with the 55% win rate figure. The fact of the matter is that different styles of trading systems will have wildly different win rates. For example, many trend trading methods have quite low win rates, like 30%. Despite that, they can be quite profitable. That’s because they have very high winner/loser ratios, like 5 to 1 in some cases.

The point I want to make is that to assess a trading system’s performance you need to consider both sides of the equation – win rate and winner/loser ratio. The expected performance of a system can be defined by the following formula:

Expected Trade Performance = (Win % x Average Winner) – (Loss % x Average Loser)

If we take the 55% win rate and the 1.5 winner/loser ratio suggested above, then our expectation would looke like this:

(.55 x 1.5) – (.45 x 1) = 0.375

That means for each trade we would expect to make 0.375 points.

Now let’s consider the trend system figures I tossed out.

(.30 x 5) – (.8 x 1) = 0.700

That’s almost twice as good even though the win rate is only about half as strong. The trend system is better on a per trade basis. Before you start looking for a trend trading methodology, though, there is something to keep in mind. That’s opportunity.

Trading systems produce different numbers of trading opportunities depending on a lot of different variables. In order to properly compare systems, we have to look at expected results over a given period of time, not just for a single trade.

Using our two systems above, let’s assume that the trend system produces 10 trades per month, and the other system does 20. We then multiply those trade numbers through by our per trade figures to get 7 and 7.5 respectively (.7 x 10 and .375 x 20). That means we would generate more profits each month with the 55% system.

So, to recap, assessing a trading system properly requires (among other things), looking at the win rate and winner/loser ratio together in conjunction with the frequency of trades. To look at any one or two factors by themselves leaves you with an incomplete picture.

More on trading systems in my next post.

Struggling with support & resistance and knowing what the key market levels are? Check out the Price Distribution Analysis methods I use.

Bookmark to:

Is This a Good Day Trading System?
Trading System Rules I Disagree With
Taking a closer look at trading system win rate
The PEG
Which trading systems should I use?