The strategy was published on the QuantStrat TradeR blog. Essentially, it’s another variant of contango strategies with averaging over parameters.
Again, to assess the market state, use the ratio of the near-term and far-term volatility indices and profit from rolling futures (if, of course, we have correctly identified the mode).
As the near-term index, take the 3-month VIX3M, and as the far-term, the 6-month VIX6M. We analyze the VIX3M/VIX6M ratio. To determine the entry point for a position, we average this ratio over XX days.
Since the average value heavily depends on the used averaging period, this strategy uses three different lengths of periods. This is done to reduce the probability of curve-fitting and the role of randomness.
For the length of the moving window, we take 60, 125, and 150 days. That is, we have 3 sub-strategies, each allocated 1/3 of the capital.
The original strategy was somewhat modified. We consider that since VIX futures are usually in a state of contango, buying VXX-like products in contango-strategies should be done very cautiously. Two conditions must be met:
1. The opening boundary is calculated either from the historical series or is further optimized.
2. If, in (1), the opening boundary is calculated from the historical series, F1/F2 futures should be included in the indicator calculation. In this strategy, relatively ‘distant’ VIX indices are analyzed – the 3-month and 6-month ones.
So, we only short VXX.
Additionally, the original strategy suggests an additional rule for entering a position – the current ratio value must be less than 1. We have slightly tightened this rule – not only the current value but also the average should be less than 1.
Strategy Rules
Calculate ratio VIX3M (3 month VIX)/VIX6M (6 month VIX). Calculate averaged 60, 125, 150 days ratios. Сreate 3 sub-strategies, allocating 33% of capital to each. If VIX < avg_ratio_NN_days and avg_ratio_NN_days < 1, short VXX. No position otherwise. Do this for each of the sub-strategies.
Strategy Performance
Test period: 2010 – 15 Dec 2023. Costs (brokerage commissions, slippage and borrow cost) are not included.
| Averaged Strategy | Benchmark: Short VXX | Benchmark: SPY | |
| Full Return | 14 280% | 5 850% | 549% |
| Annualized return | 43% | 34% | 12.95% |
| Max DD | -53% | -92% | -34% |
| Sharpe ratio | 0.96 | 0.42 | 0.70 |
Well! Looks like someone kept the old prototype code running. Along with many OTHER strategies. Glad to see you’re still alive and out there, VolatilityMadeSimple. How has your own proprietary strategy fared?
I’ve been…trading this live and iterating (and iterating…and ITERATING) on it. Suffice to say–there are quite a few new rules to catch various problematic drawdowns, though my own PA has suffered because I was long vol in March 2020, caught a huge spike up–and gave half of it back (added a new feature to make sure THAT doesn’t happen again).
I won’t divulge all the new bells and whistles publicly, but…suffice to say, the new strategy is very much improved compared to this one. I would be happy to share the rules privately in exchange for more knowledge, but…definitely not publicly!
Now…the 800 pound gorilla in the room is…how many strategies tracked before that fateful day on Feb. 5, 2018, didn’t get obliterated? Quite a few forced a position, one way or another, and though my model got out a week ahead of time, I’m pretty sure many more rudimentary strategies…didn’t fare anywhere near as well.
I’d be happy to talk about my trading experiences on a podcast as well. Care to have a chat?
Ilya, I’m quite flattered that you reached out, especially as a past reader of your blog. However, I’m not affiliated with VolatilityMadeSimple.
We’ve also been dealing with volatility strategies for quite some time (since around 2015).
Recently, I got curious about how the old VIX strategies survived events like 2018 and 2020. Reviving an old database of strategies was facilitated by existing testing infrastructure and relevant experience. Assembling a couple of dozen modules took a few days, as the rules are straightforward.
Then, I thought, why not regularly monitor their performance? Good things shouldn’t go to waste. Plus, we plan to gradually expand the basic package with new strategies. Tried to mention this in my first post. VolatilityMadeSimple seems inactive.
Regarding the spikes in February 2018 or during COVID – our strategy package isn’t much like the one presented, though there are commonalities – the main classes of anomalies with VIX products are well known. But there are also significant differences. Plus, we manage to hedge extreme spikes. So, February 2018 and March 2020 were even quite profitable. I sent you a private message.