In the March 31 Portfolio Update, we took up the question of whether it might be a good idea to change the Calendar Options strategy for different market environments. The answer was no, primarily because our market-neutral income strategies are based on statistical probability—i.e., by tilting the odds in our favor, we expect to profit over the long-run if we apply the strategy consistently.
Nevertheless, extreme conditions can shed light on how a strategy might be improved to better manage risk and adapt—inherently—to such extremes. So I went back and analyzed our first-quarter trades to see if there were any common reasons they didn’t do well. I noticed three things:
- January and February each included two highly correlated positions. We’ve used wide double-calendars and double-diagonals, with some success, to diversify risk when trades are centered very closely, but this approach didn’t provide much protection in the steep sell-off and even sharper recovery the S&P took us through in the February cycle.
- Although the March entry trades passed our filter for volatility risk, that filter proved inadequate with implied volatility in a strong downtrend.
- Reducing delta bias by 1/3 to 1/2 was not a sufficient adjustment target when the market was trending up.
From this analysis, and rethinking the implications of our shift last year to a staggered-entry, portfolio-based approach, I came up with a few strategy changes that backtesting showed improve performance in periods when the existing approach underperforms, without significantly limiting returns in the best periods.
There are three ideas behind the changes:
- Making sure positions are spread out over a range of strike prices;
- Taking a more bullish stance when making adjustments;
- Being more conservative about volatility risk when IV is trending down.
We’re going to start requiring new positions with the same expiration to be centered a minimum of about the one-week standard deviation (based on implied volatility) from the center of the current portfolio risk curve. As with adjustments, we want new positions to reduce, but not completely eliminate, our delta bias, which means entry trades also can’t be centered too far from the current underlying price. In other words, the price has to move a certain distance from the center of the portfolio risk curve before a new position can be added.
When it comes to adjustments per se, we’re also going to start making upside adjustments wider. Instead of reducing delta by 1/3 to 1/2, we’ll roll to a strike high enough to cut about 2/3 of the bearish bias. Because implied volatility tends to fall when stocks are climbing, we need to be more aggressive with our adjustments in an uptrend. Since IV tends to rise when stocks are pulling back, boosting our P/L curve, we’re not changing the adjustment rules for rolling down.
The Trend is Not Your Friend…
…at least when it comes to calendar spreads and falling implied volatility. We’ve been willing to enter positions when implied volatility was as high as the mid-point of its 3- to 6-month range, and that worked okay until the VIX broke below the 20 mark—but we’d be better off if we’re more cautious when volatility is trending down. When IV is in a strong downtrend, we will not open a new position at a volatility any higher than 1/4 of the recent range.
The new restrictions we’re placing on entry trades mean there will be fewer opportunities to build a portfolio of three positions each month within our 20-day time window—but that isn’t as bad as it sounds. The time window we’ve been using was based on the optimal range for a single position. If we’re entering multiple positions on the same underlying to manage risk, subsequent entry trades can be closer to expiration than the minimum for a stand-alone position. (Our one winning trade for March was opened just 15 days before expiration and yielded a profit of more than 30%.) That gives us a bigger time window, and thus increases the number of entry-trade opportunities compared to the old time window.
Even so, there will be cycles where we trade fewer than three positions, and in rare circumstances, none. (The new restrictions on trade entry would’ve kept us 100% in cash during what very likely would have been a disastrous month in October 2008.) This runs counter to the common practice of trying to maximize returns by being fully invested at all times, but the reduced risk more than makes up for the additional opportunity cost. Members can still expect to see plenty of trades, though: at least two entries were triggered in most months backtested, and the cycles that had fewer were either extremely volatile (September through December 2008) or so calm (December 2009) that a single position gained 40%, yielding a Model Portfolio return of 10% with three-quarters of the funds safely in cash.