Logit
Logit is the log odds of a probability, defined as ln(p divided by (1 minus p)). It is commonly used in statistical models.
Definition
Logit is the log odds transform of a probability. If p is a probability between 0 and 1, then logit(p) equals ln(p divided by (1 minus p)).
Why it matters
Many forecasting models operate naturally in logit space because it turns multiplicative changes in odds into additive changes in logit. Converting model output back to probability gives a predicted probability you can compare to implied probability.
How it connects
• Logit is the same concept as log odds.
• Odds are p divided by (1 minus p). Logit is ln(odds).
• Bayes updates often use likelihood ratio to update odds, which corresponds to adding in logit space.
Common pitfalls
Using p at 0 or 1: Logit is undefined at exactly 0 percent or 100 percent.
Confusing scale: Logit is not a probability. You must convert back to probability for interpretation.
Tools
Use Odds, Log odds, Probability Converter to translate between representations.