RSM318H1 Lecture Notes - Lecture 3: Survivorship Bias, Yield Spread, Expected Return
Document Summary
However, some features of stock return data means we may see predictability in returns based on the available data. May see positive autocorrelation in available data if we only look at stocks that have performed well. Actual autocorrelation depends on mechanism of which stocks survive. Irrational investors can make stocks over or under priced. Creates predictable stock returns that may be profitable opportunities. Realized return can be decomposed into (conditional) expected return and unexpected return. Expected return is conditional on information observed today. Represents expectation going forward a. b. c. i. ii. Conditional expected return could be related to risk. Can estimate volatility using past data or option data. D/p ratio, e/p ratio, b/m ratio, other valuation ratios. Normally distributed with mean 0 and standard deviation 1. Approaches chi-squared distribution with 1 degree of freedom. Cannot use the standard t-tests for predictive regression. Imagine we have to predict next period return can collect historical data and run predictive.