Williams’ VIX Fix
This is a follow up to Mind Money Market’s post re: the Williams’ VIX Fix (WVF). The WVF is a VIX-mimicking indicator that, unlike the VIX itself, could be applied to any intraday vehicle (ETF, futures, etc). Read more about the WVF (pdf).
MMM’s post showed that the stock market has been very bullish following high WVF readings (i.e. high volatility). To illustrate, the graph below shows results from 1994 trading the S&P 500 using MMM’s strategy (strategy rules to follow).
The strategy is in red versus buy & hold in grey. It would be wrong to think that because the strategy trailed buy & hold over most of the test that it was inferior; the strategy only spent about 13% of all days in the market so we’re not comparing apples to apples (and not including the significant impact of return on cash).
Strategy rules: go long at today’s close when today’s WVF reading will close in the top 5% of all readings over the previous 30-trading days, otherwise move to cash.
See end of post for assumptions about dividends and trade frictions. Note I’ve used the ETF SPY to calculate WVF, but assumed trades were place on the index itself.
Numbers for the number-lovers…
Clearly the market has exhibited strength following very high WVF readings. Because the strategy spends so little time in the market it’s not going to keep pace with strong bull markets, but when the high-volatility signal has triggered, it’s portended good things.
Things I like about this strategy:
- High return (actual and risk-adjusted) given limited time in market
Things I don’t like about this strategy:
- Because WVF relies on intraday data (in this case, SPY), a very long backtest is difficult (although we could use futures data for a longer look at the S&P 500)
- At just slightly lower levels of “extreme” WVF readings (say, the top 5-10% of readings rather than the top 5%) the market has actually been bearish rather than bullish. It worries me when small tweaks to strategies lead to such big differences and hints that a good bit of the positive results in this post are the result of curve-fitting
In a follow up post I’ll look at (a) trading this same strategy using the VIX itself (rather than WVF), and (b) applying this strategy to other equity indices.
Test assumptions: (a) S&P 500 returns daily dividend-adjusted (interpolated from quarterly data), (b) results are frictionless (do not account for transaction costs or slippage) but could be closely reproduced in today’s market using certain mutual funds less part of an annual expense ratio, and (c) I’ve ignored return on cash.
. . . . .
Filed under: Trading Strategies, VIX & Volatility | 5 Comments