Monday, 30 August 2010

Knowing What to Expect

Public (called "Bank" here in the U.K.) holiday here in the U.K. it's 8am when I started writing this and I've been scalping the 6E but won't be doing a video as I'm out with the family most of today and will just hit and run when I have time.

I wanted to re-visit the subject of back testing as it comes up during the 1 to ones I'm doing with the guys and gals I am mentoring.

Back testing has two purposes. The first is the obvious one - testing what works and what metrics to use for stops and targets. This type of back testing is used to find a so called edge as well as to optimise metrics. Typically, people place their stop loss points way too close and get stopped out of trades that otherwise would be profitable. I don't exit on stops, using a drop dead stop far away from the action in case of connectivity failure. I stop myself out based on what I see. 

The issue for me is that a generic stop is too expensive. The correct stop level depends upon context and is not a mechanical thing for me. The same goes for targets. I can see the benefit of using a mechanical first scale out point that has been shown over, say a year, to capture a very high percentage of first targets but the subsequent scales, for me, are based on what is happening - the context.

The second and equally important reason for back testing is to know what to expect. this is hugely important in order to be able to maintain discipline. Usually fear in trading is because of the unknown and proper back testing reduces or helps mitigate fear.  Also, if you know you can get three losses in a row, when it happens you won't lose your decipline (I hope) and fail to take the next trade.

For back testing, it's important to eyeball the trades. It's all well and good to use a computer to gather and sort the information, but I go back over the trades and look at them on the chart. I want to see the bars and how they traded to give the result.

In summary, knowing what to expect is extremely important.

1 comment:

  1. Hi EL, is there a reason why you use much larger range bars for CL (0.2R) vs 6E (0.0005R)? Given their respective ATR, I believe that 0.2 for CL would be similar to 0.0015R for 6E. Thanks. -J