I'm interested in using Index Lab to return a single merit metric that evaluates the equity curve for each stock in a dataset. Basically, the index evaluation just needs to return the accumulated value for the last bar of the equity curve. If it could display these values as part of the By Symbol tab for each stock in a dataset, that would be great.
If it's easier to do this evaluation with a WL indicator instead of using Index Lab, that works for me too.
The problem I'm trying to solve is effectively evaluating the performance of a stock against a given strategy. Currently, I'm using Net Profit and Profit per Bar for this, which are listed on the By Symbol tab. But as one can see from the attachment, these performance metrics may be positive when inspection of the actual equity curve clearly shows this strategy fails to properly trade this stock. What's needed is a better metric for reporting negative slope behavior in the equity curve, and Index Lab may have some methods for doing that.
If one can think of an alternative approach (short of manually examining every equity curve for every stock I track), I'm also interested. Alternatively, I've considered using ScoreCard metrics for doing negative equity-curve evaluations as well.
Size:
Color:
QUOTE:
Alternatively, I've considered using ScoreCard metrics for doing negative equity-curve evaluations as well.
If it's about consistency of returns (i.e. less wiggle and downside movement from the linear regression line of the equity curve) then the performance metrics measuring this that come to mind are K-Ratio and Lake Ratio (or Ulcer Performance Index).
Size:
Color:
Yes, it's about the downside movement of the equity curve with less wiggle (equity line consistency). Let me take a look at the K-Ratio and Ulcer Performance Index, which I see are part of the extended ScoreCard metrics.
http://www2.wealth-lab.com/WL5Wiki/CommunityScorecard.ashx Perhaps these ScoreCard metrics will identify the problematic stocks in my datasets without further Performance Visualizer coding.
One plus with employing the ScoreCard metrics is that the WL optimizer knows about them. So if the strategy has parameters to address falling equity, the optimizer is empowered to adjust those parameters to curve the falling equity problems with certain stocks. I didn't realize the ScoreCard metrics were equity curve aware, but I like this approach because it empowers the strategy to employ PV parameters to address weakening equity problems. Thanks for all your guidance.
Size:
Color:
You're welcome.
Yes, Scorecards are equity curve aware and for this reason said metrics (like the K-Ratio) are not available in Raw Profit mode simulations.
Size:
Color:
I've spent some time studying the above, mentioned ScoreCard metrics and their ability the check for a consistent positive slope on the equity curve. Based on my observations, I would rank their effectiveness, best listed first, as: Ulcer Performance Index, K-Ratio, and Lake Ratio. The Ulcer Performance Index shows good promise; the K-Ratio is so, so; and the Lake Ratio is a bust.
What I would like to do is compare a list of stocks within a dataset against these ScoreCard metrics using their existing PV strategy settings. How can I do that? The thought occurred to me if a can get an optimizer to make one iteration employing existing PV settings for each stock and displaying these ScoreCard metrics in the Optimization > Results tab, that would do the job. Is there a "dummy optimizer" or an optimizer setting that would execute one iteration using existing PV settings for all stocks in a dataset?
My intent is to employ the Ulcer Performance Index and K-Ratio columns to single out the poorly performing stocks based on their not-so-consistently-positive-sloping equity curves.
Size:
Color:
QUOTE:
What I would like to do is compare a list of stocks within a dataset against these ScoreCard metrics using their existing PV strategy settings.
I think the new question goes beyond the general notion of developing/finding a suitable merit metric. As such you would want to start a new topic where we could continue discussing WL's features and how to make them accomplish your new objective.
Size:
Color: