- Joined
- 21 April 2005
- Posts
- 3,922
- Reactions
- 5
After reading a post in another thread, I thought I would start this one. The title is not sarcasm by the way.
Michael,
I have some questions:
Why do people backtest?
What constitutes a successful backtest leading to a blueprint?
What biases are there to be aware of?
What issues are there with equipment and software deficiencies?
Why can every aspect of human nature be backtested?
What is one thing that corrupts all backtesting, and what same thing is relevant for trading without backtesting?
Answers to these questions would be greatly appreciated. Others please feel free to comment.
Thanks
Snake
MichaelD said:I'm afraid I'll have to steadfastly disagree with this.
I agree on one point - that it is very hard or impossible to backtest or prospectively test discretionary strategies with software.
I vehemently disagree, however, that it is impossible to test these strategies. Indeed, I'd argue even more strongly that such testing is ESSENTIAL for shorter term strategies.
Right now, your strategy is extremely profitable. Fantastic! (No sarcasm or envy intended). But...is this because of skill or because of current market conditions? How will you know the difference? How will it perform in different market conditions?
You know these answers for long term trend following systems, and spent a long time working them out, so why all of a sudden do you not need to know these same answers for your short term system? It's ironic, don't you think, that 3 months ago we were on opposite sides of this argument. What changed?
I'm doing things a bit differently. I essentially have 3 short term plan methodologies, all of which attempt to achieve the same thing. The entry signals are the same. I'm keeping records of how all 3 trade going forwards (systems 2 and 3 have real money - system 1 has been superceded but I'm still tracking it). I can look back at market conditions and how all 3 systems traded these conditions. It's very, very time consuming and tedious, but the knowledge available from this data is well worth the effort, particularly in the ability to go back and try a new idea out on existing data. It also means it is going to be more and more obvious as time goes by when a system is no longer performing as expected.
Is it simply a different road to the same end point? Maybe.
Michael,
I have some questions:
Why do people backtest?
What constitutes a successful backtest leading to a blueprint?
What biases are there to be aware of?
What issues are there with equipment and software deficiencies?
Why can every aspect of human nature be backtested?
What is one thing that corrupts all backtesting, and what same thing is relevant for trading without backtesting?
Answers to these questions would be greatly appreciated. Others please feel free to comment.
Thanks
Snake