Tuned Moving Averages

[CeeVee777]

Hi,
I'm fairly new to trading and I've been looking at various methods to trade.

I've been doing some back-testing on a SqlServer database holding about 7 years worth of data for the London Stock Exchange.

I've found that potential trades show better profitability if I use values for moving averages that are specific for each share.
Having read a lot on these and other sites I haven't found any mention of tuning the MA values for individual shares.

Is this something worth pursuing or is it just a statistical anomaly ?

 

[dcraig1]

I've found that potential trades show better profitability if I use values for moving averages that are specific for each share.
?

Not ARE better but WERE better - past tense. It may or may not hold true in the future. Is there any reason to assume that it should be so ?

 

[trendie]

I second dcraig1s reply.

By saying each share has its own specific moving average is just another way of saying curve-fitting.

Have you "tuned" the MAs for the first 4 years data, and then run those values on the remaining 3 years?
Cant remember what the technical term is, but you cant use the same data to test a hypothesis that you used to derive the hypothesis.

 

[daytradingUK]

Trendie is bang on here, you are making the experiment prove the hypothesis , an unfortunate side effect of trying to best fit systems.

If you think about it the system may work now but be useless two days after your data ran out because market conditions back then were more favourable.

I personally believe you need to do some forward testing before you can 'see' what settings are necessary. You may also find that changes in the stock's cycle will require you to alter these settings again too...

 

[Hoggums]

Hi,
I'm fairly new to trading and I've been looking at various methods to trade.

I've been doing some back-testing on a SqlServer database holding about 7 years worth of data for the London Stock Exchange.

I've found that potential trades show better profitability if I use values for moving averages that are specific for each share.
Having read a lot on these and other sites I haven't found any mention of tuning the MA values for individual shares.

Is this something worth pursuing or is it just a statistical anomaly ?

What you need to do is "tune" your indices to half the data you have - and then see how they perform on the other half.

You may be suprised at how much money you lose!