How long of a period before using a system?

[billygoat1234]

hello
I have a system for the dow but im not sure how long i should wait.. its been about 2 months and put up pretty consistant reslults each month wins only about 55-60% of the time but with 2 or 3 big winners in that mix..

 

[Mathemagician]

The last 2-3 months have been pretty homogeneous. You're best off waiting until the system has seen a few different market environments out-of-sample.

 

[GammaJammer]

Plus I recommend plotting your expectancy. That's a good raw measure. Once the volatility in that dies down, if it's still healthy you're looking good.

 

[Stewenson133]

daily?

 

[NetTecture]

::each month wins only about 55-60% of the time but with 2 or 3 big winners in that mix

Carefull here. This is statistically weak. Note that if you have 2-3 winners a month - what about the rest? If the rest is negative.... then a fluctuation of the winners may put the whole system down. In this case I would make sure I have enough test data to cover 20-30 of those, to see how "stable" those are. If the rest is positive, and stable, then - it is only a question of "smaller return or big return", not of "win or loss".

 

[mrsoul]

hello
I have a system for the dow but im not sure how long i should wait.. its been about 2 months and put up pretty consistant reslults each month wins only about 55-60% of the time but with 2 or 3 big winners in that mix..

You're never really going to know if your method makes money until you test it going forward with real money.

If you demo the system, you will never know if your sample size is accurate.

Anytime I get an idea that looks good using past data, I trade it live using the smallest possible size.

What I am saying is you will never really KNOW by demo testing; you must trade live to find out.

Your results look encouraging, so don't be a chicken.
Have confidence and test it live.
Just use small size.
This way you will find out for sure if it works.
In the worst case you will lose a little but it is worth the risk.

 

[NetTecture]

Well, seriously - you dont need real money. Programs like Ninja Trader have very good trade simulators when tied to a real level 2 feed. Unless one trades high clip sizes they are as good as the exchange. Includes little items like delaying order execution (transfer speed to exchange), as well as keeping track of ones order position in the order book (limit order execution when the price only is hit for some time). I found no difference between simulator and real here.

But forwrd testing is critical. I suggest a minitmum of 75 trades.

 

[goonior]

Goonior2
anita920