There is one signal in the Hedge strategy that has never failed in its history, and that is the Bear Market Signal.
When I was designing the hedge strategy, I quickly discovered that each component that worked well to signal when to go in or out of the market during a bull market was not effective during a bear market. For example, the phase angle threshold that is usually set to avoid B bounce in a corrective move starts to trigger on dead cat bounces. Other dataset conditions that usually signal the end of a correction also tend to buy back the top of bear bounces. In other words, the market behaves very differently. To address this, we decided to introduce a Bear mode that would deactivate the regular modality of the strategy and switch to a more defensive mode where it will try to hold up in any significant bounce, but opt out rapidly when the wind turns.
To change the behavior of the strategy, we needed to be able to recognize when we were in a bear market and when we were out of it. It's relatively easy to determine when we are in a bear market: a drop of 20% in the index. I tried to play with that bear market definition by enabling that mode at 21-22-23-24-25% line, but the optimal threshold was always when activating it at 20%. My conclusion is that it's somehow a self-fulfilled prophecy: when we start to talk and hear about bear markets in the news by the historic but arbitrary definition of 20%, people start to behave like bear market investors. Determining when we are out of a bear market is more challenging. A bear rally can sometimes take us out of the 20% drop zone before another downturn that brings us deeper into this bear. Moreover, in massive bear markets like the one in 2001-2003, it can take years to recover from the losses, so going back over the -20% threshold is just not realistic.
Upon examining the years 2003, 2009, 2018, and 2020, I observed that all market recoveries from bear markets shared a similar signature: strong, consistent exponential growth (in log space) with low internal volatility. Here's what I mean:
[Disclaimer: Nerd interlude] Consequently, we dedicated a lot of time to modeling the signature of these recovery phases using a dynamic modeling approach. Although I cannot disclose the exact equations I developed, my approach was based on the principles I published in 2007 on the topic of computationally efficient model-based predictive control. I chose this method because I knew that Jim Simons, during his hands-on involvement with Renaissance Technologies, achieved great success using philosophically similar modeling techniques. [End of Nerd interlude]
Recently, we adapted this signal for the S&P 500. We didn't make any significant changes except for lowering one threshold to account for the S&P 500's lower volatility compared to the NASDAQ. One advantage of working with the S&P 500 is the availability of complete data (full candle) dating back to 1962, which allowed us to backtest the model over a much longer period. Surprisingly, the model never failed when tested on this new (old) period. To clarify, strong bear market rallies can sometimes trigger this signal, and although it might not be a big concern since such rallies generate significant profits, the model never failed in the sense that no bear market ended without seeing this signal from the low point. Here are all the bear market instances. Each arrow on top of the indicator represents a signal, and the arrow's color indicates the signal's strength (blue < green < yellow < red). As shown, there was always a signal originating from the push out of the bottom or at the end. Also, except for 1967, where the signal was blue, all other pushes featured a red signal at the end.
If history repeats itself, this is what I currently expect to see before the end of the current 2022-2023 bear market.
There is another interesting feature of this indicator. The upper blue line represents the level of the rise since the current bottom, which is calculated from the lowest daily close to the current local high. In all the available data, we have never seen a 21% rise calculated this way that did not signal the real end of the bear market rally. In other words, all bear rallies faded before reaching this level.
This indicator was personally useful for me in this current bear market. It helped me strengthen my conviction that this bear market was not over in July 2022 when many people were writing that we were in a new bull market. It also helped to determine the potential upside left in a bear rally. This is why I decided to give access to it as part of the S&P 500 package with the hope that it will also contribute positively to your own investing game.