An ITA Wealth member recently asked the question as to how the Tranche system that I am using in my real-time weekly performance reviews (to demonstrate the importance/influence of Timing/Review Date luck) compares with the results of automated back-tests.
Herb has kindly run a series of back-tests from the time that I started the Tranche portfolios (12-29-17) until the end of August 2019 (20 Months) to answer this question.
Four options were used for the back-tests:
- End-Of-Month (EOM) reviews using the BHS strategy (TRO set to Yes and BL/SH? set to No) using the original Momentum Ranking;
- As for option 1 but using Score Ranking (SR);
- As for option 1 but with weekly rather than EOM reviews;
- As for option 2 but with weekly rather than EOM reviews;
For the Real-Time tests the BHS system is used with TRO set to No and BL/SH? set to Yes (BL/SH? set to Yes is not available in the Back-Tester). The total portfolio is divided into 5 tranches with each tranche being reviewed every 5 weeks and with a 1-week separation between each tranche review.
A comparison of performance is shown below:
Here we see that, although there are detailed differences along the way, the 20-month returns of all the back-tests are essentially the same. The results of the EOM tests look smoother in the above plots but this does not mean that volatility is lower – it is just a result of the fact that fewer points lie on the EOM curves.
By comparison, the performance of the real-time total portfolio – average of the 5 tranches – (heavy red line) lags the back-test results by a significant amount.
Using VTTVX (dark blue line) as our benchmark, the back-tests beat this benchmark while the real-time tranched portfolio does not. However, the best performing tranche (Tranche 1 – dashed red line) matches the benchmark.
It is also important to remember that back-tests assume that (both Buy and Sell) orders are filled at closing EOD prices and assume no slippage (difference between bid/ask prices). Real-Time transactions occur after the review date decisions are made, Buy and Sell orders are not made simultaneously, and both include slippage. This is a reality of real-time trading/investing.
It is difficult to take a lot away from this data since the time period is still very short, but a few comments/observations might be made:
- Timing/Review-Date Luck plays a significant role in determining portfolio performance;
- Results are consistent with previous (longer term) tests that suggest that EOM reviews (rather than e.g. mid-month reviews) may generate the best results.
- Although weekly reviews come out slightly ahead at the end of 20 months, this may reflect the benefit of more frequent reviews rather than the actual Day-Of-Month review since the portfolio is only in “neglect” mode for 1 week rather than four (EOM tests) or five (Tranche) weeks.
Tranching probably degrades “optimal” performance due to the influence of Timing/Review Date luck. However, the more frequent reviews tend to smooth things out and to reduce overall volatility.