Thread by @ReformedTrader

1/ Momentum, Reversal, and Seasonality in Option Returns (Jones, Khorram, Mo)

"Option returns in individual stocks have momentum, short-term reversal, and quarterly and annual seasonality. This remains strong after controlling for other characteristics."

papers.ssrn.com/sol3/Papers.cf…

2/ Zero-delta one-month near-the-money straddles in individual stocks are formed each month and held to expiration.

Returns are excess returns calculated based on bid-ask midpoints.

IV-HV uses one-year HV
IV term stpread is calculated between 60-day and 30-day options

3/ "Straddle returns in the previous month are likely to be reversed in the following one.

"At lags 3+, there is momentum. The slope coefficients on lags 3 through 12 are all positive and all statistically significant.

"There is a complete lack of long-term reversal."

4/ "We see evidence of cross-sectional reversal at lag one and strong momentum for longer formations. The classic momentum strategy (2-12) is the strongest.

"Momentum's performance is close to that of the best predictors from the existing literature (IV-HV and IV term spread)."

5/ "The appendix contains additional results that were omitted from the main text."

6/ "Including controls has almost no effect on the coefficient estimates or t-statistics for past returns.

"Similarly, adding a past return measure to the set of controls has little effect on the coefficient estimates of the controls."

7/ "The short-run reversal effect (the coefficient on the lag 1 return) is generally strengthened by the inclusion of past returns at longer lag.

"The slope coefficients are always most positive for the '2 to 12' past return and generally decline as the lag lengthens."

8/ "Results suggest that momentum is pervasive across all options, not just those in the extremes of the distribution of past returns.

"In no case does the cross-sectional strategy significantly under- or out-perform the time-series version."

9/ Industry option momentum is constructed like industry stock momentum but is built using straddle returns instead.

"There is strong evidence of momentum but no short-term reversal in industry option portfolios.

"Factors appear to display momentum using all formation
periods."

10/ The authors also consider quarterly and annual seasonality strategies using past straddle returns.

"Both seasonality and momentum are important, but momentum may be somewhat more so.

"Including additional control variables has virtually no impact on the seasonality effect."

11/ "Almost all L/S factors based on individual straddles display a clear positive skew.

"Factor- and industry-based portfolios have lower or negative skews, though modest relative to the left skews of the short SPX straddle and the short eq. weighted stock straddle portfolios."

12/ "Five-year moving averages of returns are positive for all three strategies (short-term reversal, momentum, and seasonality) at all times.

"The momentum eect appears to have strengthened slightly in the second half of the sample, though it is by an insignificant amount."

13/ "For strategies based on individual straddles, momentum survives factor adjustment; reversal does not.

"Reversal is related to the tendency of high IV, relative to actual vol, to underperform. High recent straddle returns may cause IV to increase too much (overreaction)."

14/ "Individual straddle momentum is not spanned by industry or factor momentum, whether or not the additional controls are included.

"There is no industry momentum alpha after controlling for individual straddle momentum."

15/ "The tendency of straddles to reverse their most recent monthly return permeates the cross section.

"Like reversal, momentum is pervasive in the cross section. Unlike reversal, however, momentum is somewhat more prominent in stocks facing high impediments to arbitrage."

16/ "Since straddle returns are approximately market neutral, the variables should be approximately uncorrelated. Consistent with this, option momentum is strong regardless of past stock returns.

"In contrast, past stock returns appear unrelated to future straddle returns."

17/ "High past volatility swap gains, like past straddle returns, predict low future straddle returns. However, when both are included in a regression, only the past straddle return retains explanatory power.

"Momentum and reversal result from delayed responses to surprises."

18/ "Seasonality in straddle returns may be the result of unpriced seasonality in stock volatility.

"Volatility surprise is the superior measure of volatility seasonality: RV using daily returns is a less noisy proxy of true volatility than the monthly straddle return."

19/ "The seasonal average of implied volatilities is insignificantly related to future realized volatility.

"The coefficient becomes even smaller after controlling for the seasonal average of RVs. This indicates that IVs do not anticipate the seasonality in RVs."

20/ "Our finding that momentum can exist without LT reversal suggests these are distinct phenomena.

"Results also suggest that underreaction can exist without the disposition effect. Straddles expire in one month; an investor reluctant to exit simply does not have that option."