Saturday, August 12, 2006

Put/call ratio and stat arb

It looks like put/call volume ratios (suitably cleaned, subtracting market makers, etc.) have predictive value for the underlying stock. A model portfolio going short/long the 20% of names with the highest/lowest ratios would have made 60% annualized returns over the last dozen years, with no down years (trading costs might reduce this to 50%). A subscription to the CBOE data necessary for this analysis is now available for a $600/month fee. I wonder what the weekly or monthly Sharpe ratio is for this strategy?

More good evidence against the strong efficient market hypothesis :-)

NYTimes: A new study has found that a portfolio based on the preferences of options traders has consistently beaten the overall stock market. In reaching that conclusion, the study paves the way for what may be a very profitable stock-picking strategy.

The study, “The Information in Option Volume for Future Stock Prices,” appears in the fall 2006 issue of the Review of Financial Studies. Its authors are two associate professors of finance: Jun Pan of the Sloan School of Management at the Massachusetts Institute of Technology and Allen M. Poteshman of the University of Illinois at Urbana-Champaign.

...Until now, there has been no comprehensive study of option traders’ track records as stock pickers. That hasn’t been for want of trying: the requisite data simply was not available to researchers. The volume figures that options exchanges report publicly, for example, reflect a combination of several kinds of transactions, muddying the overall picture. The volume number reported for a given option, for example, may reflect new purchases by options traders, but it may also include the sale of positions previously acquired. That makes it hard to tell whether traders actually favored a stock.

Using a private database provided by the Chicago Board Options Exchange, the two professors were able to deconstruct an option’s total trading volume into various categories. They excluded trades by market makers, for example — dealers at the options exchange who buy and sell securities for the general purpose of maintaining liquidity. They narrowed the database further to focus on just that portion of an option’s daily trading volume that reflected new positions by other traders, on the assumption that these transactions offered a clearer signal of what traders actually thought of the underlying stock. The database covered the dozen years from the beginning of 1990 through the end of 2001.

For each option in this database, the professors calculated a daily volume ratio of newly acquired put options to newly acquired call options. A high ratio meant a strong consensus among options traders that the price of the option’s underlying stock would fall, while a low ratio showed a widely shared expectation that the stock would rise. The professors found that the stocks whose options had the lowest ratios consistently outperformed the stocks whose options had the highest ratios.

Consider this hypothetical portfolio constructed by the professors: it held the stocks of the 20 percent of options with the lowest put-call ratios, while selling short the stocks whose options had the highest such ratios. (The portfolio was readjusted weekly, adding and deleting stocks because of changes in the ratios.)

The professors reported that before transaction costs, this portfolio produced an annual average return of 62 percent over the dozen years covered in the study. This contrasts with an annualized total return of 12.3 percent for the general stock market over this period, as measured by the Dow Jones Wilshire 5000 index. Still more impressive was the fact that the portfolio earned double-digit returns each year, even when the overall market declined. Based on their data, however, the professors had no way to determine how options traders were able to achieve these results.

The portfolio required frequent transactions because the price moves correctly anticipated by options traders lasted for only a couple of weeks, on average. So transaction costs would have eaten up a big chunk of the return. But Professor Poteshman estimated that in the hands of an institutional investor, for whom such costs would typically be quite low, the portfolio’s return would still have been as much as 50 percent annually.

Even for individual investors, who would pay higher costs, Professor Poteshman estimated that the annualized return would have been well into double digits.

DESPITE the strong results of the strategy, it would have no more than academic interest if investors had no access to the private C.B.O.E. database that the professors studied. In July, however, the exchange began selling subscriptions to this database to the public.

A subscription isn’t cheap: $600 a month. That helps make the professors’ strategy impractical for small investors. Still, the study shows that potentially valuable information can be found in options traders’ behavior. And institutional investors, including hedge funds and mutual funds, can easily exploit it.


Anonymous said...

Sweet and now we get to watch a nonequilibrium process as this arbitrage dies?

Steve Hsu said...

I'm already seeing hits on my blog from search engine queries on words like CBOE, option, volume, etc. by hedge funds, banks and analytics software companies.

The weakly efficient market (i.e., with some finite equilibration timescale) at work! :-)

Seth said...

Ah, so this post is itself a sort of arb play? Cut yourself a piece of the (advertising) action as the big money piles onto the trade?

(Of course, the big money discovered this long before the NYT published it ...)

Steve Hsu said...

I don't know if anyone really had access to this data previously, other than these academics. So, there might not be anyone trading on this exact strategy. People have looked at put/call ratios for a long time, but as explained in the article the data is noisy and you need to know who is trading to clean it up.

Don't forget to click on the appropriate text ads to transfer funds from luxocrats to humble physicists :-)

Anonymous said...

I guess we also have to look at why the professors, if looking at 50%pa, would make their research public? For the greater good of the Wall St hopeful? Surely they must have seen the value in what they were sitting on?


Steve Hsu said...

I imagine that for CBOE to give special access to data that is not in the public domain, they would ask researchers to sign an agreement saying they won't trade on the results. Otherwise it would be an unfair activity of the exchange.

Now that the data is available by subscription, this condition presumably does not apply.

Anonymous said...

The OEX p/c ratio tends to be correct. Equity put/call ratios don't accurately measure sentiment.

E.g., A put credit spread is made up of puts only, but is bullish. Selling puts is also bullish. Likewise, a put debit spread is also made up of puts only, and is bearish.

Anonymous said...

I should also add Interactive Brokers has a free EOD options commentary with highest/lowest put/call ratios.

Michael Benjamin said...

Greg, IB doesn't tell you how they derived their PCR data, and I suspect it's not the same way as in the article.

Has anyone gone any further with this idea?

Blog Archive