Sunday, May 04, 2008

A combination momentum and mean reversal model based on earnings annoucements

Mark Hulbert of the New York Times just discussed 2 momentum strategies investigated by professors David Aboody, Brett Trueman and Reuven Lehavy.

Strategy A: pick stocks in the top percentile of 12-month returns. Buy them (individually) 5 days before their earnings announcements and sell them just before the announcement.

Strategy B: pick stocks in the top percentile of 12-month returns. Buy them (individually) 5 days immediately after their earnings announcements and hold them for 5 days.

Strategy A is very profitable: the annualized excess return is 47% before costs. (To be taken with a grain of salt due to the large transaction costs associated with trading momentum strategies, especially if small-cap stocks are involved.) Strategy B is very unprofitable: the annualized excess return is -43% before costs.

So what are the ways we can make best use of this research?

Naturally, instead of buying the top percentile after the earnings announcements, we should have shorted the stocks, thus making Strategy B a reversal strategy instead.

Furthermore, what about the bottom percentile of stocks? Should we have shorted them prior to the announcements, and bought them after the announcements? If so, we would have a very nice dollar-strategy for you statistical arbitrageurs out there!

6 comments:

JR said...

Ernie

An interesting article. Thanks for drawing it to our attention.

I've not yet read the academic paper cited but a few thoughts/questions that immediately spring to mind:

- A viable trading strategy would have to be profitable during both bull and bear markets. I wonder whether these strategies, particularly strategy A, was profitable during bear markets (in absolute &/or market relative terms).

In general, my own preliminary work on a related topic for UK companies found that speculative momentum in the run-up to earnings announcements was only evident during bull markets. (NB: I presumed the authors of the cited paper include inactive companies which could lead to serious sample bias
otherwise.)

- It is well-known that many strategies have become less profitable in recent years, a fact most often attributed to greater market efficiency. I wonder the extent to which any negative trend in returns (abstracting from cyclical effects) is evident in the authors' results. Again, any evidence of this would diminish the robustness of such a strategy to generate positive returns moving forward.

If anyone has had chance to look at the paper and could comment on the above points, I'd be very grateful.

- The phenomenon of post earnings announcement drift (where positive/negative earnings surprises take some time to be fully incorporated into the price) is well established in the empirical literature. A possible extension to your shorting strategy based on B would be to explore the efficacy of including the extent of earnings surprises (positive or negative) as a factor in a model. In particular, one might imagine that companies with negative earnings surprises may enhance returns of a post-earnings shorting strategy. Just a thought.

JR

Ernie Chan said...

JR:
Thanks for your insightful comments.

Indeed, many such momentum strategies are becoming less profitable in recent years. Generally, I find that competition has decreased the optimal holding periods of momentum strategies as well.

To see if this momentum strategy works only in bull market, one can check whether the short side of the trade is profitable: i.e. shorting stocks in the bottom percentile prior to the announcement. My research indicates that it is not -- thereby supporting your thesis that this strategy may work only in bull markets.

I have also found that PEAD strategies as you suggested are still profitable, subject to various refinement and modifications to counter the increasing competition.

Ernie

Donkey Face said...

Interesting...

I hope you don't mind me asking these questions:

Over what period of time did you perform this analysis?

What was your universe of stocks?

What's the Bloomberg function for getting historical earnings calendars?

Why can't I subscribed to followup comments on this?

Ernie Chan said...

I performed the backtest for the short side of the trade over the period 1997-2006. The universe is the top 1000 mktcap US stocks.

You can read the original paper quoted to find out what their backtest period is.

I don't use Bloomberg for this data, so I am afraid I don't know what function to use.

I will look into why the comments subscription is off.

Ernie

Anonymous said...

I backtested this strategy (both long and short side) using a dynamic S&P500 universe from Jan 2000 to May 2008 and the returns were dire.

My backtest shifted the announcement dates by 1 day to ensure no possible bias due to pre\post\intraday EA timing and my assumed execution was next day VWAP at $0.01 per share.

I think there are several issues with the study.

a) The highest percentile stocks are very likely to be the smaller more illiquid names.

b) The assumed transaction costs are way understated given the illiquidity of the names likely to be traded.

c) The study covers a very long period, much of which transactions costs were a lot higher than they are today and liquidity a lot more shallow.

d) Market impact has been totally ignored.

Ernie Chan said...

Dear Anonymous,
I agree with your (dire) assessment of this strategy. As I mentioned in the reply to jr's comment above, the short side of the trade does not work. The long side did not beat the index either. So all in all, there is not much to recommend it for.

In general, your observation about strategies proposed by academics is very acute. Their strategies usually get all their returns from illiquid stocks, seldom include sufficient transaction costs, and usually are backtested on such ancient historical periods that their relevance to present-day environment is highly questionable.

Ernie