Merlin Strategy Performance for February 2020
Below is a performance summary for our Merlin trading strategy for the month of February 2020, net of transaction costs. To learn more about the strategy and to follow along in real-time with its trading signals, please visit the Merlin strategy page.
Merlin Performance for February
| Strategy | February | Year to Date |
|---|---|---|
| Merlin Margin | -8.25% | -3.96% |
| 60/40 Benchmark | -3.57% | -2.11% |
See Performance Disclosures: Monthly Performance.
Merlin Closed Trades in February
| Closed Trades | Return on Signal | R-Multiple |
|---|---|---|
| BDSI | -18.77% | -1.00 |
| MRCY | +14.36% | +1.50 |
| ROKU | +7.91% | +0.52 |
| RUBI | +28.81% | +1.51 |
| DRNA | +5.24% | +0.50 |
| CVNA | +18.12% | +1.50 |
| ZYXI | +24.48% | +1.50 |
| EGO | +25.73% | +1.50 |
| CRSP | -15.33% | -1.0 |
| REPH | +6.20% | +0.50 |
| VCYT | -14.33% | -1.00 |
| EXAS | +7.76% | +0.50 |
| CROX | -20.55% | -1.37 |
| AL | -13.25% | -1.00 |
| SIMO | -10.15% | -1.03 |
| ALNY | +5.25% | +0.51 |
| FAS | -19.34% | -1.15 |
| SKX | -9.46% | -1.12 |
| UDOW | -16.01% | -1.00 |
| JCI | -7.12% | -1.01 |
| FORM | -10.35% | -1.0 |
| APPS | -18.26% | -1.00 |
| NMIH | -19.35% | -0.998 |
| CVS | -10.83% | -1.00 |
| FB | -9.12% | -1.01 |
| KL | -21.04% | -1.00 |
| EW | -11.34% | -1.23 |
| MKTX | -9.48% | -0.998 |
| PAGS | +8.70% | +0.498 |
Performance Commentary
Merlin’s streak of market outperformance came to end in February with a busy month of trading and a -8.25% slide in the margin account. This ranks as the fourth largest monthly decline going back to January 2006.
Largest one-month Merlin declines:
- -9.98% in June 2011
- -8.97% in May 2012
- -8.77% in May 2006
- -8.25% in February 2020
On the surface, it looks like February was a large negative month with nothing to be happy about but in fact, it was just the opposite. I am thrilled at the performance of the strategy and believe it was an amazing stress test that the system passed with flying colors, and here’s why:
Merlin is a mean-reversion strategy
It’s important to first understand that Merlin is a multi-week mean-reversion strategy. That means the system is designed to buy dips, pullbacks, and retracements in individual stocks and ETFs. This is very different than a momentum trading system that seeks to buy strength, breakouts, and ride trends.
Naturally, when markets start pulling back, mean reversion systems get more and more interested in stepping up to buy the pullback. But when pullbacks evolve from being shallow opportunities to deeper cascades, in very short order and without bouncing, this is where they can run into big trouble.
The last week of February wasn’t just an ordinary pullback. The major averages melted down in a historic fashion setting a handful of records in the speed from which they fell from all-time highs. There were no bounces to speak of, it was just an elevator ride down 15% after closing at fresh all-time highs just 7 days earlier.
For mean reversion strategies specifically, this type of environment is deadly.
The great news however is that it was not deadly for our Merlin strategy, in fact, it performed much better than expected given what the market threw at it. It began buying stocks early in the decline, and it did get stopped out of many of those names just days later, but after everything was said and done, the systems position size and risk management procedures prevailed. Ordinary losses for an extraordinary market fall from all-time highs.
The limitations of backtests
Backtests are only that – backward-looking tests. There was no “optimize for future Covid-19 virus” button available when the strategy was being engineered and so when future high impact events do occur, they are important moments in time to measure how a strategy handles the surprise.
Surviving dynamic and sometimes extreme market environments is a hard challenge for all traders and in the world of systematic trading, the best way to boost the odds of survival is by designing systems that: are not overfit to historical data, are not too complex, and most importantly, are driven by logical and sound principles – all of which were a front and center focus for Merlin.
NAV is back to levels last seen two months ago
Merlin returns over the past four months:
- February 2020: -8.25%
- January 2020: +4.29%
- December 2019: +5.42%
- November 2019: +8.19%
Putting our performance numbers in perspective we can see we are back to levels seen just two months ago in early December.
Heading into March, Merlin continues to hold equity exposure, but much less than it did in the prior months. The sell-off that unfolded in the last week of February triggered Merlin’s “reduce risk” circuit breaker telling the strategy to slow down its buys until the environment can stabilize and proves itself once again.
Given the events that unfolded at the end of February and the stress test the market dished out to all of us, I think I’m the most pleased trader that just suffered an 8% down month.
If you’re looking for an evidence-based strategy for swing trading individual stocks and ETFs on an end of day closing basis, check out the details behind Merlin here.
I’m looking forward to signing up for the next six months when my first month period ends in a week or two. Thanks!
Glad to have you in for the long haul Todd! Looking forward to trading with you.
What about throwing in a system breaker on Merlin to pause or limit all new positions while the major indices are way over stretched on the highest timeframe from the longer term MA for example. If you looked at the 200 MA on S&P daily, weekly and monthly you would see price was at the most extreme level it’s ever been and crying out for a major correction to its MA, Corona or not. The extremes were more than the 2000 and 2007 levels. Another possibility is telling Merlin to switch more and more of its trades to safer haven risk off sectors when the indices start getting into these overstretched levels, and then revert back once they have corrected for example.
Hi Tony, yea I like where your head is at, thinking about ways to potentially mitigate against these types of risks and market environments in the future. There are a handful of tests and possible rules that come to mind like the ones you outline above that I will research to see how they impact the strategy going back over a deacde in backtests.
I personally believe when building systems that it’s ideal to keep the moving parts and rules to an absolute minimum in order to reduce complexity and potential data mining. So I basically want to see good statistical reasons with explainable causes before I add new rules. Those are the types of considerations i’ll be weighing with I go through these tests.
Thanks for the ideas!