Despite the use of complex and sophisticated forecast prediction models, accurately predicting share-price movements remains largely a mystery.
But what is commonly known is that you can maximise your stock market returns when the post-earnings-announcement drift is magnified.
New research has identified when this happens.
By comparing the results of three different forecast prediction models, anyone can accurately predict whether the post-earnings-announcement drift will be magnified.
The post-earnings-announcement drift is a well-documented phenomenon, which suggests that stock prices tend to drift upwards or downwards, depending on whether the earnings figures are above or below expectations.
In practice, let’s assume IBM released earnings of $0.30/share for the first quarter of 2009. Say there are two different groups of investors. The first group forecasts earnings to be $0.25/share and the second group forecasts a figure of $0.22/share.
In this case both predictions are lower than the actual share price, which means the prediction error is positive.
As a result of this positive prediction error, these two groups of investors will pull in the same direction, that is, they will both buy IBM stocks. This combined pull is enough to magnify the holding period return by an average of 2% over 60 days after the earnings announcement.
In other words the share price will increase 2% more than it would have, without the consensus in prediction error, in the 60 days following the earnings announcement.
However, if the prediction error was negative, that is if both groups predicted a share price that was less than what was announced, the share price drop in value will also be magnified by more than 2% (on average) over 60 days after the earnings announcement.
In this case you would recommend selling the stock before the 60 days is up.
But what happens if one investor forecasts earnings to be $0.32 and the other $0.25 or, in other words, both groups disagree about the share price value? They will pull in different directions and the holding-period return will remain stable.
In this case, you should leave the stock alone and neither buy nor sell.
Another way to think of this market movement is to consider the example of horses pulling a wagon. When the horses pull in the same direction, we know where the wagon will go. If the horses don’t pull in the same direction then the wagon is unlikely to go anywhere.
Where there is consensus about the prediction error being positive you should buy the portfolio of stocks. Where there is consensus the prediction error is negative you should sell the portfolio of stocks. And where there is no consensus you may as well leave the stocks alone.
It does not matter whether a profit or a loss is predicted or posted. What matters is whether there is consensus about different forecasts. If there is, the holding-period return is substantially magnified, often by two or three percentage points.
If there is no consensus, that holding-period return remains stable.