Forecasting Earnings Using k-Nearest Neighbors
Easton, Peter D., Kapons, Martin M., Monahan, Steven J., Schütt, Harm H., Weisbrod, Eric H.
The Accounting Review · 2024 · Volume 99 · Issue 3
Abstract
We use a simplek-nearest neighbors algorithm (hereafter, k-NN*) to forecast earnings. k-NN* forecasts of one-, two-, and three-year-ahead earnings are more accurate than those generated by popular extant forecasting approaches. k-NN* forecasts of two- and three-year (one-year)-aheadEPSand aggregate three-yearEPSare more (less) accurate than those generated by analysts. The association between the unexpected earnings implied by k-NN* and the contemporaneous market-adjusted return (i.e., the earnings association coefficient (EAC)) is positive and exceeds the EAC on unexpected earnings implied by alternate approaches. A trading strategy that is long (short) firms for which k-NN* predicts positive (negative) earnings growth earns positive risk-adjusted returns that exceed those earned by similar trading strategies that are based on alternate forecasts. The k-NN* algorithm generates an empirically reliableex anteindicator of forecast accuracy that identifies situations when the k-NN* EAC is larger and the k-NN* trading strategy is more profitable. Data Availability:Data are available from the public sources described in the text. JEL Classifications:C21; C53; G17; M41.
Keywords
earnings, forecasting, machine learning, nearest neighbors
Enable JavaScript for citation tools and additional paper interactions.