Request: additional metrics

[mandelmus]

With some strategies, it'd be nice to be able to identify alpha and beta?  Would you consider adding metrics like percent performance against the "perfect" strategy or against some market proxy, like SPY, DJIA, or other symbol we can add as a "market"?

The "perfect" strategy is based on going long each bar if the bar closes higher and going short each bar if the bar closes lower ... based on look-forward information ... it's similar to taking the sum of the absolute values of the each bars' close - open.

[Michael R. Bryant]

For stock trading, a market index benchmark sometimes makes sense if you’re trading long-only. I will probably add something like that. I’ve discussed “perfect profit” in my newsletters and developed a slightly more sophisticated version, which I presented in an article. The problem with so-called perfect profit is that it’s so far away from any realistic results that it’s basically meaningless as a comparison. If you try to use it as an optimization objective (for example, trying to minimize the difference between the ideal equity curve and the actual one), it’s no better than simply trying to maximize net profit.

 

Mike Bryant

 

Subject: Request: additional metrics

 

With some strategies, it'd be nice to be able to identify alpha and beta?  Would you consider adding metrics like percent performance against the "perfect" strategy or against some market proxy, like SPY, DJIA, or other symbol we can add as a "market"?

The "perfect" strategy is based on going long each bar if the bar closes higher and going short each bar if the bar closes lower ... based on look-forward information ... it's similar to taking the sum of the absolute values of the each bars' close - open.

--
 
 

[mandelmus]

*pop*

Builder has been building a bunch of "interesting" long-only strategies on several stocks I've been analyzing.  Reviewing the trade histories, I noticed that some of these strategies entered nearly everyday -- which I don't mind.  But, I got to wondering if the strategy was actually identifying a valid strategy or just lucking-out by piggy-backing the general upward trend of the stock.  Looking at the past 6 months, I noticed that the stock has been trending up nearly the entire time.  So, I'm not sure how to assess those strategies against the stock's general trend.

It would be nice to have some additional metrics like symbol and market alpha and beta that would indicate whether, and by how much, a particular strategy is outperforming simple buy-and-hold.

[mandelmus]

Thanks Dave, I need all the support I can get!  :)

[Hugh Hammer]

You are dedicated.... I give you that much! :)

Hugh

On Wed, May 22, 2013 at 3:13 PM, mandelmus <gmb...@gmail.com> wrote:

Thanks Dave, I need all the support I can get!  :)

--
You received this message because you are subscribed to the Google Groups "Adaptrade Builder" group.
To unsubscribe from this group and stop receiving emails from it, send an email to adaptrade-buil...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[mandelmus]

I found the article, "Ideal Trades as Targets for Strategy Development" and downloaded the IdealTrades strategy for TradeStation.eld file.  I imported the eld and attached the strategy to a chart.  What do I need to do to get the strategy to output to the "C:\idealtrades.csv" file?  Is there a certain time frame I should select?

[Michael R. Bryant]

I don’t recall the details. While I’m happy to make my newsletter articles available free of charge, I don’t offer support for my free items. The articles are intended to explain the accompanying code and serve as a starting point for the reader’s own work.

 

Mike Bryant

 

Subject: Re: Request: additional metrics

 

I found the article, "Ideal Trades as Targets for Strategy Development" and downloaded the IdealTrades strategy for TradeStation.eld file.  I imported the eld and attached the strategy to a chart.  What do I need to do to get the strategy to output to the "C:\idealtrades.csv" file?  Is there a certain time frame I should select?

--
 
 

[mandelmus]

While trying to figure out the code to get the "IdealTrades.eld" to work, I found another nice article, along with an *.eld, on comparing trading systems with "perfect profit" potential.  Instead of perfect profit, Jeff calls it "Potential Annual Salary (PAS)".  Here is an excerpt from the article ...

 

"Conclusion

Of course there are many reasons to trade a market besides the PAS calculations. Different setups and trading techniques work better on different markets. Each market has its unique feel to be sure. Each market may be more tradable during different times of the day, which may or may not be possible for you to trade. The allure of trading what is trendy can be exciting too. However, taking into account which markets may hold the best potential for profit may be a good metric for you to consider. For example, in the video that accompanies this article I take a look at trading Apple stock. In my study of trading Apple on a 5-minute chart, I found that trading it was not worth it. Why? The buy and hold out performed the PAS calculation! If you are interested, see the video."

 

The "PAS Estimator.eld" will output a list of trades and compare it with buy-and-hold performance.

 

 

*I noticed that Mike is also a contributor to systemtradersuccess.com

[mandelmus]

I also found this article on "Key Performance Metrics" in which Jeff Swanson describes his favorite metrics.  I think Builder lists all of those metrics except a few (CAGR, RAR, t-test, expectancy score).

[Michael R. Bryant]

Just to clarify, Builder does in fact include CAGR (“Ave Annual Compounded Return”) and t-test (“Significance”), which is converted from the t value to the corresponding probability value in Builder. Builder includes expectancy (“Average R-Multiple (Expectancy)” ) but not the expectancy score.

 

Mike Bryant

 

Subject: Re: Request: additional metrics

 

I also found this article on "Key Performance Metrics" in which Jeff Swanson describes his favorite metrics.  I think Builder lists all of those metrics except a few (CAGR, RAR, t-test, expectancy score).

 
 

[mandelmus]

Oops, forgot about those ... thanks for fixing that

[mandelmus]

I've noticed that in many of my strategies, the "Ave R-Mult" is listed as 0.0000 for both IS and OOS.  Is that correct?

[Michael R. Bryant]

R multiples are based on the risk. If your strategy doesn’t have risk, the r-multiple will be zero.

 

Mike Bryant

 

Subject: Re: Request: additional metrics

 

I've noticed that in many of my strategies, the "Ave R-Mult" is listed as 0.0000 for both IS and OOS.  Is that correct?

 

[mandelmus]

So, is the "Average R-Multiple" the complete interpretation of Expectancy?

I found this formula at TraderMike.net ...

Expectancy = (Probability of Win * Average Win) – (Probability of Loss * Average Loss)

... which looks more like "Average Trade".

[mandelmus]

Are there metrics for "Risk Adjusted Return (RAR)" and/or % of time (bars) in trade vs. not in trade?

"The Risk Adjusted Return (RAR) calculation takes into account the time your money is at risk in the market. This is done by taking the CAGR and dividing it by exposure. Exposure is the percentage of time (over the test period) that your money was actively in the market. I like to see a value of 50% or better."