Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
In this article, we investigate the predictive power of single predictors with regard to oil prices, using several constrained approaches that contain predictor-related, parameter-related and combined constraints. Based on these approaches, we obtain several noteworthy findings. First, the predictive power of several predictors can be significantly improved under a predictor-related constraint. Second, the predictive ability of most predictors can be improved using parameter-related constraints, but those improvements are not large. Third, combining the two types of constraints can achieve a remarkably better performance in forecasting oil price returns than an individual strategy both in terms of the number of predictors and the magnitude of improvements. Finally, our findings are robustly supported by the success ratio, alternative forecasting windows, other look-back periods and direct comparisons.