Crisis Alpha: Our “Amulet” EA for Stock Crashes Collapsed. Why?

A beginner-friendly summary of the verification: “Crisis Alpha: Our “Amulet” EA for Stock Crashes Collapsed. Why?”.

What’s the Idea? Betting Against the Market

We’re always looking for clever ways to make our algorithmic trading strategies (EAs) more robust, especially during market turmoil. One exciting idea that’s often floated is “crisis alpha” – a strategy that actually profits when the rest of the market is crashing. The hypothesis for this experiment was simple: Stock market indices, like the S&P 500 or Nasdaq, often experience sharp, persistent downtrends. Think back to late 2018, early 2020, or much of 2022. During these periods, if we could consistently “short” these indices (meaning, bet on their price going down) using a trend-following approach, we might not only make money but also have a valuable hedge against our main portfolio. In other words, when our main long-only strategies are struggling, this short-biased strategy could potentially kick in and save the day! This concept is different from a previous attempt (Research 104) where we tried to use the Japanese Yen as a safe haven in FX. That didn’t work out, so we wanted to see if shorting indices directly would be a more effective way to capture crisis alpha.

How I Tested It

To put this hypothesis to the test, I built an EA specifically designed to follow downtrends in major stock indices. The strategy would identify when an index was clearly heading south and then open short positions, aiming to ride that downward momentum. I backtested this strategy across various market conditions, paying special attention to the periods of significant market crashes or corrections mentioned earlier (2018 Q4, 2020, and 2022). The goal was to see if it could consistently generate profits during these challenging times and provide that much-desired negative correlation to a typical long-only portfolio.

What Happened? A Catastrophic Failure!

Well, sometimes even the most promising ideas turn out to be duds. The results for this “Crisis Alpha” short-index strategy were, to put it mildly, disastrous. Over the entire backtested period, the strategy posted an overall loss of -29.6%. To give you some context, that’s nearly a third of the starting capital gone! Even more telling was the Profit Factor (PF). The Profit Factor is a common metric in algo trading, calculated as total gross profit divided by total gross loss. A PF greater than 1 means the strategy is profitable, while less than 1 means it’s losing money. Our strategy’s PF came in at a dismal 0.13. In other words, for every dollar it lost, it only made 13 cents back. That’s a catastrophic failure by any standard.

Why It Failed So Badly

You might think, “But what about those crash years? Didn’t it at least profit then?” Sadly, no.

In 2022, a year marked by significant market declines, the strategy barely broke even, making a paltry +0.1% from just two trades.
In 2020, the year of the swift COVID-19 crash and recovery, it ended up -1.2% in the red. This highlights a fundamental problem: index crashes, while dramatic, are often incredibly fast and “V-shaped.” The market drops like a stone but then snaps back with surprising speed. A trend-following short strategy often faces a few critical issues:

Late Entry: By the time a clear downtrend is established enough for a trend-following EA to enter a short position, a significant portion of the move might already be over.
Getting Squeezed: When the market inevitably bounces (even temporarily), these short positions get “squeezed” – meaning they face quick losses as the price moves against them.
Upward Drift: Between the crashes, the market has a natural upward bias. This “upward drift” slowly but surely eats away at any small gains and adds to losses over time, making it incredibly difficult for a short-biased strategy to stay profitable. Despite having a slightly negative correlation of -0.11 with our core long-only strategies (meaning it did tend to move opposite to our main portfolio), the massive overall losses rendered this correlation meaningless. As we’ve learned from previous research (like Research 25), a negative correlation is only useful if the strategy is actually profitable. Otherwise, you’re “just losing in a different place!”

What I Learned: Stick to What Works

This experiment was a harsh but valuable lesson. We’ve re-re-confirmed a critical principle in our algorithmic trading journey: shorting, whether it’s in FX (as seen in Research 25 and 48) or stock indices, consistently proves to be a “drag” on performance. It’s incredibly difficult to make money consistently betting against the market. The robust, reliable edge in price action lies almost exclusively in long-only trend following. The market tends to drift upwards over time, and riding those upward trends is where the consistent profits are found. Therefore, this “Crisis Alpha” index shorting strategy is rejected and will not be incorporated into our confirmed systems. No changes to our existing, profitable strategies are needed. Sometimes, the best lesson is knowing what not to do!

How this connects

This verification builds on earlier ones (what failed before and what I tried this time, comparisons between approaches).