If we go through an extended period of relatively low volatility in the S&P it’s easy to get periodic large moves in either direction that then feel like a shock since we have been nodding off. We then want to know how to react. Is this a breakdown/breakout so a new trend has begun or is it the pullback we should buy?
Let’s set up some conditions:
- The true range must be greater than the highest true range of the last 10 days.
- The close must be lower than yesterday’s close.
- The close must be greater than the lowest close of the last 2 months.
The third condition prevents us from buying volatility increases during a meltdown. We’re going to look at different holding periods so we will require rule #3 to be true for every intervening day as well.
|Holding Period||Total Events||% Profitable||Total Profits|
The data since 2007 shows us that recovery takes a few days. After five days one can assume that the shock has worn off and our returns will merge into the average for the period. The average five-day return is 2.482 points. For the entire period, the market gained 1200 points over 2307 days. That works out to a similar average daily return but the drawdown circumstances are radically different. There is only one meaningful drawdown – Dec. 8, 2014. Other than that the equity curve is beautiful.
The approach will not get us into a strong rallying market since we may be left waiting for a shock and one may never arrive. This does speak to our current conditions.
On the flip side we could ask if a high volatility breakout is a good way to get long. Here are the conditions:
- The true range is greater than the highest true range of the last two weeks.
- The close if the highest close we have had in the last two weeks.
I could use one month instead of two weeks or 1 week instead. Neither will change the result(s). Using the two-week example we have 37 cases and over a 5-day holding period, we would have lost a total of 177 points. A monthly look back suffers an 83 point loss. Successful high vol. breakouts are difficult to find for the S&P 500.
Lastly, we could add an additional filter to only buy pullbacks if we have gone through a low volatility period. The condition would be something like: Highest 10-day close minus lowest 10-day close is less that 2 average true ranges. When I add that filter, it does reduce the number of occurrences (t0 37), but the probability of winning drops to 25%. The total number of points earned = 446. Perhaps counterintuitively, a big pullback matters just as much in a higher volatility environment as it does at the end of a lower volatility period.