How We Make Earnings Predictions

We look at past earnings events and market information to estimate how a stock might move after its next earnings announcement. These estimates are not guarantees, and real results can differ.

1. Information the model considers

Before an earnings announcement, the model reviews 16 pieces of information about each company. These include recent stock and options activity, past earnings-related price moves, and measures of trading activity. The model can use different combinations of this information for different companies.

16 Information points

Flexible Uses relevant information

2–8 quarters Past earnings reviewed

One outlook Selected for each company

2. How past results guide the outlook

For each company, the model compares information from its 2 to 8 most recent earnings periods. It looks for patterns that have been associated with different stock moves in the past, then selects the outlook that best matches those patterns. Past patterns may not repeat.

Possible outcome Expecting price to go up

The model sees past patterns that point to the stock moving higher after the announcement.

Possible outcome Expecting price to go down

The model sees past patterns that point to the stock moving lower after the announcement.

Possible outcome Expecting a massive price move

The model expects a bigger-than-usual move, but does not predict whether the stock will rise or fall.

Possible outcome Expecting price move contained

The model expects the stock to stay within a narrower price range after the announcement.

3. Choosing the final outlook

Different information can point to different outcomes. The model compares the possible outcomes and chooses the one that best fits the patterns it found in past earnings periods. This is an estimate, not a certainty.

When the model is unsure The model selects one of four outlooks: rise, fall, a larger move either way, or a smaller move. If the available signals do not point clearly to one outcome, no clear outlook may be shown. A prediction should not be treated as a promise or as a recommendation to buy or sell.