When you have not enough sampling size..

[HolyGrailSeeker]

Has anyone tried to get a larger sampling size by say tweaking a few parameters of your trading strategy so that you can have a larger sampling size? Does this count or will it defeat the purpose of having a sampling size?

For e.g using a 1 period MA and a 5 period MA and also a 2 period MA with a 5 period. Entry is when the price bar closes above both MA with the shorter MA on top of the longer MA.

If price is above 1/5 MA it is also above 2/5 MA so we can use total both results and samples? Thereby effectively almost doubling the sample size. Or does it defeat the purpose of having a sampling size by duplicating almost similar results in your opinion?

 

[zwaen123]

What you are actually measering is robustness of your parameters. In general, a method is more reliable when 1) only a small set of variables are used and 2) the parameters of your variables do work at a range of parameters.

Whats more important for sample size is coverage of all market conditions

Also, i would not be interested in a method which generates 5 signals in the past 10 years. Why is your set so small?

 

[murray t turtle]

Has anyone tried to get a larger sampling size by say tweaking a few parameters of your trading strategy so that you can have a larger sampling size? Does this count or will it defeat the purpose of having a sampling size?

For e.g using a 1 period MA and a 5 period MA and also a 2 period MA with a 5 period. Entry is when the price bar closes above both MA with the shorter MA on top of the longer MA.

If price is above 1/5 MA it is also above 2/5 MA so we can use total both results and samples? Thereby effectively almost doubling the sample size. Or does it defeat the purpose of having a sampling size by duplicating almost similar results in your opinion?

%%
IF your signal is 5 times in 10 years =ok;
but use a 2oo day moving average also to profit.
SPY is a good/ pro benchmark....................................................................Good question

 

[sonoma]

Has anyone tried to get a larger sampling size by say tweaking a few parameters of your trading strategy so that you can have a larger sampling size? Does this count or will it defeat the purpose of having a sampling size?

For e.g using a 1 period MA and a 5 period MA and also a 2 period MA with a 5 period. Entry is when the price bar closes above both MA with the shorter MA on top of the longer MA.

If price is above 1/5 MA it is also above 2/5 MA so we can use total both results and samples? Thereby effectively almost doubling the sample size. Or does it defeat the purpose of having a sampling size by duplicating almost similar results in your opinion?

Better to bootstrap.