My first fully automated system: are these statistics sound enough for you?

ronblack · May 24, 2009

My comments:

(A) Your success rate is below 50%. To me - I emphasize this is IMO - robust high frequency trading should have a success rate higher than 70%.

(B) Your avg. win / avg. loss ratio is too low. Add slippage and a few missed trades and it can quickly get much lower.

(C) Sharpe ratio is not acceptable.

(D) Profit factor is too low and nowhere close to a value professionals would even spend 1 minute looking at this system statistics in more detail

(A) - (D) combined may indicate one or a combination of the following:

(1) The results reflect pure chance

(2) System is fitted

(3) System is based on unrealistic assumptions about trade execution

(4) Several filters have been applied to massage the system behavior

My 2 cents, as always

TRY

Profit factor > 3

Sharpe ratio > 2

Avg win//avg loss > 2

Sucess rate > 70%

No filters

No optimization

Add slippage

2 years of foward testing

ourson · May 25, 2009

Wow, Thanks a lot to all of you. It really helps. I will definitely try to forward test it.

Thank you ronblack for your detailed comments on the statistics. One thing is for sure: there is absolutely no filter nor optimization in my strategy. What clearly affects the win:loss and sharpe ratio is the built-in money management rules which cut the cumulated profit by 2/3 with a max drawdown of still 12% but without such rules daily fluctuations are too erratic to my taste. Consistency is key for me. Since I will start small anyway (I certainly do not target to make a killing overnight), my wish is to make steady gains month after month and increase my confidence in my own system. When I do, I will progressively increase the size of my trading. Some guys are arguing I should start bigger and I do get their point but since it will be the first time I put my own money on the table, I want to make sure I can trade without worrying about what is at stake. I will certainly not teach you how psychology is of tremendous importance and therefore while I feel confident, I cannot fully assess how it will turn out and how I will manage my actions in the process.

ronblack · May 25, 2009

Quote from ourson:

Thank you ronblack for your detailed comments on the statistics. One thing is for sure: there is absolutely no filter nor optimization in my strategy.

No problem...Be careful there you may be trading noise. Over the years i have tested hundreds, maybe close to a thousand systems. I found out the hard way that one of the most important parameters is the profit factor. The higher the better but it must be as high as 2 at least. Otherwise you are risking trading noise. The problem is it may take you 2 or 3 years to realize it since your returns will fluctuate about 0% all the time.

My suggestion is to try to develop a system with a high profit factor, > 2. This variable is related to the success rate but many don't know it. Here is a paper with good info about trading system development and variables like profit factor, success rate, etc.