We use a four-state multinomial logistic regression model in order to estimate the probability of corrections and crises in the Norwegian stock market. The probabilities are subsequently translated into market exposure through a systematic trading algorithm based on the Kelly criterion, aimed at yielding risk-adjusted return in excess of a buy-and-hold strategy in the Oslo Stock Exchange Benchmark Index (OSEBX).
We conclude that financial indicators carry predictive content for the occurrence of market downturns. Particularly, we find that the Price-to-book and Price-earnings multiples, the VIX and the Commodity Channel Index are suitable determinants of stock market development. With a realized Sharpe ratio of 0.86, our candidate strategy outperforms the market at a significance level of 95 %. Thus, we find evidence against semi-strong form of market efficiency in the Norwegian stock market during the period march 2014 - march 2019.