Author: shyam

Momentum Formation Period

Previously, we looked at skip-months and rebalance frequencies for momentum portfolios. A 1-month skip & monthly rebalance turned out to be ideal. However, we did these analyses keeping the formation period the same at 12-months. What if we changed that as well?

Turns out, there is no single “ideal” formation period where all stats converge. However, if set the rebalance frequency to 1-month, the average of the formation periods of the top performing portfolios works out to 12-months.

If the momentum fund is large enough, then it could probably be sliced into multiple sub-portfolios, each with different configs to avoid this magic 12-month formation.

Also, since the underlying process creating these portfolios is the same, the equity curves come out all bunched together. There maybe differences in month-over-month performances but they are all highly correlated.

Code, charts and statistics on github.

Categories Tags ,

Momentum Rebalance Frequency

Previously, we found that the traditional 12_1 momentum configuration, where you look at the previous 12-month performance while skipping the most recent month and rebalancing every month, was indeed an ideal config (MOM_1_1). However, there are momentum index funds that rebalance once in 6-months (MOM_[0,1]_6). Is there any performance give-up if you rebalance infrequently?

Turns out that the traditional config is quantifiably better than others. However, there’s isn’t much of a performance give-up even if you rebalance once in 6-months (MOM_0_6).

Besides, the analysis here doesn’t factor in transaction costs which would be a bigger drag on the monthly rebalance config. When you add the tax-advantage and low-cost of index funds into the mix, the current crop of momentum index funds don’t look all that shabby.

Code and charts on github.

Categories Tags ,

Momentum Skip Month

The original Jegadeesh and Titman momentum paper (pdf) used a “skip month” to manage the reversal effect (quant.stackexchange). However, why is it one month and not two, or three or four?

Here’s what the equity curves of different skip month configs look like.

The “skip one month” is indeed a magical config. Also, if you are optimizing for Sharpe, skip two.

Code and charts on github.

Categories Tags ,

Volatility, Volatility of Volatility, and Momentum

Momentum has proved to be the premier anomaly in different markets. And so has low-volatility. What happens if you combine both of them? Also, what if you also add low volatility of volatility into the mix?

There are a couple of ways to skin this cat. You can start with low-volatility and add momentum. Or, you could go the other way – start with momentum and then add a volatility sort.

tl;dr: go with low-volatility first, momentum second (VOLxMOM).

While a simple momentum sort gives the highest return, adding a low-volatility filter to it gets you a better risk-adjusted return.

The order of the sort – first volatility and then momentum or first momentum and then volatility – doesn’t seem to matter much for the Sharpe rankings but the former ended up with slightly better returns.

Code and charts on github.

Categories Tags , ,

Weighted Strike-Spot Ratio

Can options trading predict the underlying’s returns? Center of Volume Mass: Does Options Trading Predict Stock Returns? Bernile, Gao, Hu (SSRN) tries to answer the age-old question.

They construct an options volume weighted strike-spot ratio and use that the predict the underlying’s next-day returns.

We rely on the volume-weighted strike-spot ratio to characterize the central location of the distribution of trading activity along the moneyness of available option contracts on the same stock. The ratio of the contract’s strike price (K) and the underlying stock price (S) measures the option moneyness, whereby call (put) options are out-of-the-money when K/S is above (below) one. After normalizing K/S by subtracting one, we calculate the weighted average of the normalized K/S ratio across available contracts using as weights the number of lots traded on each contract during the same period (V WKS, hereafter). V WKS
reflects the center of mass in the options volume distribution along strike prices of available contracts and takes on higher (lower) values when the trading volume is tilted more toward OTM (ITM) calls and ITM (OTM) puts.

While their results look promising, we setup a very simple backtest to see if it can be used to trade the NIFTY. Sadly, both net Open Interest and Value traded fail to show any effect on next-day returns.

I guess this is one more for the #fail pile.

Code and charts are on github.

Categories Tags