How to use metrics to choose a strategy?
Author: innertrader
Creation Date: 3/4/2017 9:40 PM
profile picture

innertrader

#1
I have a strategy ("S1") consists of 10 equity symbols that are either active or not. This strategy returns less on an APR% basis than B&H, but reduces exposure, and drawdown (DD) by two-thirds. Both the Ulcer Performance Index (UPI) and WL scores indicate that S1 is a better strategy than B&H on a risk-adjusted basis.

However, exposure is less than 60% with S1. I would like to supplement this strategy with a second strategy that would increase exposure (i.e., invest cash) while providing an overall improvement in metrics. The second strategy I selected for that purpose ("S2") is activated on symbols when only when S1 has them inactive.

I am trying to decide if adding S2 to S1 (S1+S2) is a good move on a risk-adjusted basis. The table below shows some key performance metrics that highlight the changes adding S2 makes. S1+S2 increases APR% above S1 alone, and above B&H. There is a modest increase in DD and the number of trades more than quadruples.

THE WL score goes down modestly, primarily due to increased exposure and DD. However the UPI goes up almost 50%, which would imply S1+S2 is a much better risk-adjusted strategy than S1 alone. However, DD Close nearly doubled, which neither the WL score or UPI considers. Not sure how much that metric should influence this decision.

I'm looking for comments on whether the data clearly indicate that I should adopt S1+S2 in place of S1 alone. If not, what other things should I consider?

profile picture

innertrader

#2
Table
profile picture

innertrader

#3
Corrected table entry
profile picture

Eugene

#4
I'd consider that you add S1 and S2 to a Combination Strategy to evaluate visually how well they're playing together. Does S2 smooth the portfolio equity or not? Will it support S1 when it stumbles or only increase the max drawdown? etc.
profile picture

Dave B.

#5
"However the UPI goes up almost 50%, which would imply S1+S2 is a much better risk-adjusted strategy than S1 alone."

Correct me if I am wrong, but an increasing UPI is LESS desirable than a decreasing score.
profile picture

innertrader

#6
The Ulcer Index (UI) increases as the depth and duration of drawdowns increase. So like Standard Deviation, a larger number is less desirable.

However, the Ulcer Performance Index (UPI) divides the risk-free return by UI. This calculation is similar to the Sharpe ratio. So a higher UPI number is more desirable.

There's a thorough explanation from the creator here http://www.tangotools.com/ui/ui.htm.