I am going to shock you. I and many other contributors here on the site have been focusing on incoming earnings, who is going to surprise, who isn't and what it all means for an investment strategy.
The question you should ask is: does it matter if a company surprises on earnings?
A substantial body of research shows that surprises do matter over the longer term, meaning a series of surprises (both positive and negative) can correlate to relative performance. But does it matter in any one quarter?
Let's look at one data point. How did the earnings surprises in Q3 2011 correlate to stock performance? I sorted about 1,500 larger companies by earnings surprise and then looked at the performance one month later. Note that I used the stock price on the day after reporting. In this way I wanted to replicate a naïve strategy, as they call them, of simply buying the biggest earnings surprise after it is reported. I divided the universe into groups of 150 and then looked at the average performance relative to the average earnings surprise.
The sad news is that you shouldn't bother to buy a surprise after the fact. Or at least it didn't pay to do this last quarter. There was absolutely no correlation between size of surprise and subsequent stock performance and the worst groups did as well (or poorly) as the best groups.
But before you chuck it all and go to an index fund, let me make a couple points. The market will never pay you for reacting to the past. To earn your money, you need to anticipate correctly. This was somewhat of a backward-looking test, in that it was reactive to results already reported. In my recent posts, I pointed out that we are using the estimate revisions to try to anticipate surprises. Owning a stock for a longer period, during which estimates are rising, and then seeing a positive surprise as confirmation of the trend should be more rewarding.