# Interpreting the Odds Ratio of a logistic regression model

I'm currently working on building a logistic regression model with the aim of predicting whether a given stock index will go up or down the following day. The table below shows the 3 models I've ran and the associated odds ratio's. The dependent variable in this study is the Direction of the stock market which can take 2 values - Up or Down. However, I'm struggling with how to interpret the odds ratio's of the independent variables. Specifically, how do I interpret the odds ratio associated with the variable Mu? The odds ratio for the variable Mu is 0.00000, how would do I interpret and explain what an odds ratio of 0.0000 means?

• It looks like your zero-value is the parameter estimate, which is a very different situation than having zero as the estimated result. Getting at the question you asked, an odds ratio of zero means that $p/(1-p)=0$, so $p=0$. The prediction is no chance of the event happening. For log-odds of zero, $log(p/(1-p))=0$, so $p/(1-p)=1$, meaning that $p=0.5$. – Dave Dec 23 '19 at 12:49
• Why are you categorizing the result as "up" or "down"? Surely it would be better to model the proportion of increase or decrease. If a stock costs (say) \$100 a share then a rise of \$0.01 is meaningless. But a rise of \\$1.00 is not. – Peter Flom Dec 23 '19 at 13:02