Session 9: Random Walks and Momentum

If markets follow a random walk, the price change in the next period should not only be independent of past price changes and completely unpredictable. Rather than debate the theoretical underpinnings of the random walk hypothesis, we look at the evidence on whether price changes in consecutive time periods are correlated and come to wildly divergent conclusions, depending on the time period in question. With very short intervals (minutes or hours), there is little detectable correlation, with much of the observed correlation being caused by market microstructure effects (the bid ask spread and liquidity). With daily or even weekly returns, the correlation turns negative, with paper profits to be made of the price reversals. As you go from weeks to months, the correlation turns positive with price momentum carrying the day. Finally, as you look at returns over many years (3 to 5 years), price reversals become the rule rather than the exception. This instability explains why it is so difficult for momentum investors to keep making money, since the key to making money seems to be avoiding the inflection points where momentum turns to reversal.

[marcomontecino]

This whole class is a hidden gem somewhat lost inside of Youtube. Thanks Mr Damodaran for posting this whole playlist and sharing your knowledge.

[marcludolph]

"Are they crazy?" hahahahaha cool

[StasSlu]

"You can make money on momentum, until it stops working".