On Scaling In: Confusion (Part One) regarding pyramiding, I discussed the possibility of the Avg. Profit/Loss% performance metric being a subpar measure of system performance. To get a better understanding of system performance of buying on pullbacks, scaling, and default mean reversion trading systems, I decided to use two other performance metrics (three if you consider exposure a performance metric).
SPY
Default(Buy 100% of equity when DV2 < 50, Sell 100% of position when DV2 > 50):
- Exposure: 49.61%
- CAGR: 13.03%
- MDD: 27.82%
Purchase at first Dip(Buy 100% of equity the first day the close is less than the close of the day DV2 < 50, Sell 100% of position when DV2 > 50):
- Exposure: 25.52%
- CAGR: 9.36%
- MDD: 19.80%
Scale In(Buy 50% of equity when DV2 < 50, Buy 50% of equity more the first day the close is less than the close of the day DV2 < 50, Sell 100% of position when DV2 > 50):
- Exposure: 37.64%
- CAGR: 11.50%
- MDD: 21.94%
Nasdaq-100
Default(Buy 2% of equity when DV2 < 50, Sell 100% of position when DV2 > 50):
- Exposure: 69.75%
- CAGR: 33.44%
- MDD: 27.70%
Purchase at a Dip(Buy 6% of equity the first day the close is less than the close of the day DV2 < 50, Sell 100% of position when DV2 > 50):
- Exposure: 68.07%
- CAGR: 37.03%
- MDD: 29.98%
Scale In(Buy 1% of equity when DV2 < 50, Buy 1% of equity more on each subsequent day the close is less than the close of the first day DV2 < 50, Sell 100% of position when DV2 > 50):
- Exposure: 68.43%
- CAGR: 38.03%
- MDD: 27.22%
Individual Nasdaq-100
Default(Buy 100% of equity when DV2 < 50, Sell 100% of position when DV2 > 50):
- Average of Exposure: 44.58%
- Standard Deviation of Exposure: 6.43%
- Average of CAGR: 15.76%
- Standard Deviation of CAGR: 10.68%
- Average of MDD: 54.95%
- Standard Deviation of MDD: 19.95%
Purchase at first Dip(Buy 100% of equity the first day the close is less than the close of the day DV2 < 50, Sell 100% of position when DV2 > 50):
- Average of Exposure: 24.22%
- Standard Deviation of Exposure: 2.69%
- Average of CAGR: 9.81%
- Standard Deviation of CAGR: 8.22%
- Average of MDD: 46.23
- Standard Deviation of MDD: 18.34
Scale In(Buy 50% of equity when DV2 < 50, Buy 50% of equity more the first day the close is less than the close of the first day DV2 < 50, Sell 100% when DV2 > 50):
- Average of Exposure: 36.82%
- Standard Deviation of Exposure: 2.18%
- Average of CAGR: 14.42%
- Standard Deviation of CAGR: 9.66%
- Average of MDD: 49.77%
- Standard Deviation of MDD: 19.06%
Conclusion
The differences between each method seems negligible. Personally, I plan to use the purchase at first dip method. Even though the scaling in method and the first dip method have similar CAGRs after factoring for exposure (the first dip method has marginally higher CAGRs for 2/3 tests), the scaling in method will have higher commissions and slippage costs. The first dip method also had a lower MDD on 2/3 of the tests, but that has little to do with my decision. In Scaling In (Part Three) I reach similar conclusions, testing these entry methodologies across two other simple mean-reversion trading systems.
Adaptive Trader 
Woodshedder discussed an appropriate benchmark on your previous post. Howard Bandy recommends that an investor print equity charts of the various strategies he/she is considering and pick the charts he/she is most comfortable with. I suspect many of us would select charts that have a high MAR (CAGR / MDD) which Woodshedder alluded to.
I am not a scale-in type of guy but many of the strategies favoured by the Connors Group, for example, tend to achieve higher returns with scaling in.
Have you researched whether the success or failure of scaling in is somewhat dependent on the volatility of the security traded?
Thanks for the idea; I haven’t tested it but I plan to now.
I agree, checking the system equity curve somewhere in the system development process is definitely necessary before live trading, but for comparing systems, I prefer to use performance metrics since the software I use re-scales the Y axis (profit) based on the size of the equity curve, which means that two systems may have very similar equity curves, despite one system being significantly better than the other.