Following on from last week’s discussion on back-testing.
How much testing is required for a trader to be comfortable to trade the strategy in the live market with real money? This most likely varies greatly depending on the trader’s personality. Quickly sharing some thoughts on where I am at in that regard.
Most likely this will result in me taking a couple of steps backwards, walking back down the mountain, prior to making a fresh attempt at the summit.
Lance Beggs, who runs a great trading educational website at www.yourtradingcoach.com, shared some of his thoughts of how to go about developing a trading strategy/trading business. You can view his article here. I think it is an easy-to-understand illustration of what work is required.
I have reproduced a portion of his diagram in the image below and added in some components that I feel I personally need in order to create confidence in my own trading strategy. Here is an edited version of the diagram:
The addition is the step of large-scale backtesting. I have found it difficult to build confidence from reviewing smaller batches of executed trades (20, 50 or 100 at a time), and refining my trading parameters on the basis of those reviews. I have been doing that for extended periods of time – I am talking months and years here. That hasn’t (yet) turned me into a profitable trader. It has certainly developed my skills and understanding of the market but it has not resulted in me making trading profits. There is still a gap.
Additionally, I completed many hundreds of trades in a simulated environment, using ForexTester. Again this helped in my progression and development of a trader. However, the downside there was that it was very time-consuming (weeks and months of full-time work) and it was difficult to complete what-if scenarios other than retest the entire sample with fresh parameters.
For some strategies, particularly ones with discretionary elements in the setup or trade management phases, testing in a simulated environment is about as good as it gets. It is not feasible to test the strategies programmatically. My colleague, a discretionary trader at a hedge fund, tests his strategy with exactly this approach.
I am coming to realise that I am not comfortable enough testing in that manner. I need bigger sample sizes of data, want to capture bigger testing periods and don’t want to spend months and months doing so. I also want to be able to analyse data quickly and efficiently, including what-if scenarios.
If I can complete some “large-scale back-testing” and show a profit, only then is it really worthwhile to move into simulated trading in ForexTester, and then onto live trading. It seems I just continue going round in circles otherwise. Programming skills are not something that I possess, thus I hope that my decent MS Excel knowledge will be sufficient for completing the testing.
Again, this brings home the point that the path to becoming a successful trader seems to be always longer than one expects. I am sure some traders will read this and say “Geez George! You are only just realizing the stuff – this is a given!”
On a good note, I have collected a lot of technical knowledge about the markets and understanding of technical analysis principles. This nicely feeds into making sure that the strategies I am testing are capable of actually being deployed and tradeable in a live setting. And I am not going to be strategy hopping in that sense – I will simply be trying to refine and simply the trend-following approach that I have been using the past couple of years.
Right, ready to get into testing next week.