The idea behind my post-trauma work is that market participants take a while to rebuild positions – longer than it takes Vix to fall back to where it started from. Let’s set up some rules and study this behavior:
- Trauma is defined as Vix moving above its 10-day average + 3 standard deviations.
- The trauma is over if Vix falls below its level before the trauma day or if it falls below its 10-day average.
These examples come from early 2013. The trauma line is marked in red and we can see how usually the S&P (plot 2) is higher than where it began before the event. If the moving average becomes elevated then this will not be true. We can’t wait for Vix to fall below its start point since that may not happen in a significant bear market. In the next chart, we can see the P&L of buying the S&P’s after the event and selling it when it ended:
This is a surprisingly good result. It tells us that even though our definition of the end of trauma is not very demanding when it comes to Vix, there is a superior rally in S&P when Vix recedes. Yes, it is biased because the sample period is a bull market but higher volatility periods make trauma proportionately meaningful. In fact, the strategy works during the last 12 years except for one glaringly horrible trade – in October 2008. This definition of Trauma gives us a reasonable number of events so we don’t have to wait long before another one occurs. By buying after a Vix (mini) spike we essentially put a mean reversion trade on at an attractive entry-level – thus the profits.
There were 41 cases of trauma over the last six years and on average it took slightly more than 6 days for Vix to fall back below its starting point (or its average). That is a very short period but this was, for the most part, a bull market so we shouldn’t be that surprised. The profit for that 6.28 holding period works out to 1.94 pts/day versus an average daily S&P move of .95 pts/day. If we alter the rules so we must wait to exit after Vix falls below its level before the event then total profits almost double but risk is considerably higher and there is no guarantee that it will ever fall below that level.
Post Trauma Behavior
What we want to study are the returns after the trauma event is over. Let’s say we are allowed to buy the market the day after Vix falls below its starting point (or its moving average). This is trading for cowards. We never try to buy price weakness when Vix is elevated. Surely this can’t work …
The 5-day profit takes into account all cases where we are between 1 and 5 days (incl.) after a trauma event. Most of the time it does make it to 5 days. There are six occasions when we get another trauma event before we get to 5-days since the last one. If that occurs then we will usually lose money.
The post-trauma window between 5 and 10 days (incl.) earns us 806 points, but the next 10 get us only another 141 points. Clearly, we have found a sort of sweet spot. The shocking thing (and my entire point) is that we can make a very consistent amount of money by waiting until after a trauma event and holding the position for 10-days or until we get a new trauma event. There are no trailing stops (which would help) and no profit targets. 25% of the time we will be forced out early – before we get to day 10.
The 10-day post-trauma return is greater than one standard deviation of average daily return for the entire sample (2.29 pts/day) and the return profile is vastly better. If we combine the two periods so we buy the market the day after a trauma event and hold it for 10-days after the trauma is over we earn 1170 points out of a total move of 1250 but our holding period is only about 600 days out of a total sample of 1480. Our drawdown is lower simply because we wait for a 3 standard deviation move before we buy so we normally don’t endure those market declines unless they occur while we are in a post-trauma period.
Another way to use this data is to study the return profile of the next 10-days after the 10-day limit. By continuing to hold a position you earn only another 110 points or .3 pts/day – far below the average daily return. This can be used as a bear market warning zone of sorts. If we performed the same calculation for > 20 days and > 30 days then we would see returns get even worse.
There are other factors we could add to tell us if a trauma event is more likely to recur. We could define trauma differently so we get better buy signals the next day. By using this simple approach we are able to see the lingering effect of a trauma event well after it is over.