One Single "Tie-Breaker" Metric

[TradersRUs]

There are 2 closely related strategies (but not the same), that produce similar (and good/tradeable) results over 20 years of backtesting on a variety of different stock baskets. In some periods and/or baskets, one is "ahead" (in terms of not only net profit, but also all the typical metrics like Sharpe ratio, Profit factor, drawdown, etc. etc.) while the other is ahead at other combinations of periods/baskets (using the same metrics). The difference in terms of net profit between the two varies and can be as high as +-30%, but neither one comes out higher consistently.

How would you resolve the dilemma of picking 1 strategy out of the 2 to live-trade with?
Specifically -

Which single solitary metric can one rely on to be the "tie-breaker" in such a case?

 

[jelite]

Are you restricted to picking only one or can you allocate X% of risk to first and 1-X to second?

 

[Scataphagos]

results over 20 years of backtesting on a variety of different stock baskets.

"20 years" is completely irrelevant.

The market dynamics (and who is powering the move, whatever it is) are only relevant to NOW. (The market's "what" and "who is driving it?" shifts over time.) "20 years ago" market history is irrelevant to now... other than "chart support and resistance", "relative strength", of course.

Trade WHAT... without consideration as to "whom" you think is behind it.

KISS, baby!

 

[TradersRUs]

Are you restricted to picking only one or can you allocate X% of risk to first and 1-X to second?

They are too similar - one has a slight twist on the other. Trading both at the same time would produce mostly the same signals and be confusing.

 

[guru]

They are too similar - one has a slight twist on the other. Trading both at the same time would produce mostly the same signals and be confusing.

Some hedge funds trade thousands, if not hundreds of thousands strategies, allocating small % to each one. So you should be able to trade two strategies with 50% allocation each.
Otherwise pick one with the highest sharpe, or highest average %return, or even a random one. You won't be able to determine which one may do better in the future.

 

[2rosy]

The difference in terms of net profit between the two varies and can be as high as +-30%, but neither one comes out higher consistently.

How would you resolve the dilemma of picking 1 strategy out of the 2 to live-trade with?
Specifically -

Which single solitary metric can one rely on to be the "tie-breaker" in such a case?

lowest standard deviation of your holding time

 

[TradersRUs]

lowest standard deviation of your holding time

Thanks. This is the kind of specific response I am looking for.

I have not come across this metric before. Is there a standard name for it, or is this something custom you have come up with?

 

[guru]

Thanks. This is the kind of specific response I am looking for.

Pay attention to "highest sharpe". This is often recommended, widely used, and will be looked at if your strategy is evaluated by others.

https://www.academia.edu/8003677/Ernie_Chan_Algorithmic_Trading

 

[sonoma]

They are too similar - one has a slight twist on the other. Trading both at the same time would produce mostly the same signals and be confusing.

What are their correlations?

 

[DiceAreCast]

And you know that how? I am not sure whether you ever handled a backtest with your Visual Basic skill set and daily data from yahoo finance.

"20 years" is completely irrelevant.

The market dynamics (and who is powering the move, whatever it is) are only relevant to NOW. (The market's "what" and "who is driving it?" shifts over time.) "20 years ago" market history is irrelevant to now... other than "chart support and resistance", "relative strength", of course.

Trade WHAT... without consideration as to "whom" you think is behind it.

KISS, baby!