Some Thoughts on Back Testing

One of the purposes of this blog is to allow me to share ideas and also to get into contact with other trend following traders.  I also occasionally email successful trend following traders in order to glean precious insight from them.  About 2 months ago, I emailed Canadian trend following trader, Jason Russell with a few questions.   Jason Russell, who was a student of Ed Seykota and profiled in Michael Covel's book Trend Following, graciously provided me with a bit of advice:

It sounds like you are on the right track with the exit and breakout ideas. I believe that I would be doing you a disservice in offering anything further other than suggesting that you learn how to define an idea so precisely that you can code it up and test it.

Since I have never been a mechanical trader (I am a rules based discretionary trader), I have never attempted to back test any strategies.  This email, however, provided me the impetus needed to delve into the world of back testing.  

I found a great program for back testing called Trading Blox, which is so simple to use, even I, with no programming skills,  was able to figure it out.   You can download a free trial of the program here.

My intention is not to alter my stock trading system (I am satisfied with that), but to create a Forex trading system to trade in addition to trading stocks.  I have decided to trade a system based on Donchian Channel breakouts, which is a system somewhat similar to what the Turtles traded as outlined in Michael Covel's second book

There is a lot I could discuss about this system, but I would like to focus on just one key variable, which is how much should be risked on each trade. You can't control how much money a trade will make,  but you can control how much you are willing to lose.   Below are the results of my system using different bet sizes:

In the table above, we see that in the first row, risking 0.25% on each trade using my system results in a 10.34% annualized rate of growth, with a maximum drawdown of 26.9%.   The back test was for all liquid futures and used data from 1996 until today.

The row highlighted in blue represents the results of using a 1% risk per trade, a figure often used by veteran traders.

Below is a graph showing risk per trade versus return:

So, basically, if you risk much more than 2% per trade, results become worse (and drawdowns begin to get pretty horrendous).

In summary, the results of my back test confirm what I had already suspected to be true:  don't risk more than 2% per trade.

We approach markets backwards. The first thing we ask is not what can we make, but how much can we lose. We play a defensive game.

-Larry Hite

Never risk more than 1% of total account equity on any one trade.  By risking 1%, I am indifferent to any individual trade.  Keeping your risk small and constant is absolutely critical. 

-Larry Hite

There are old traders and there are bold traders, but there are very few old, bold traders.

-Ed Seykota