Mean Reversion and the role of Normalisation in Investing
October 31, 2017
Imagine that you are offered a 3:1 return for every correct heads/tails call on a coin toss i.e. if you call heads and the coins lands head, you win 3x the amount you bet and vice versa. For the sake of simplicity, assume that you have nothing to lose until the first 20 turns. On the 21st turn, you are offered a double or nothing bet i.e. you bet all your winnings and if you call the outcome correctly, you take home double the amount, whereas if you lose, you give up all your gains (you cannot quit the game mid-way). As you reach the 21st turn, you have observed that the coin landed heads up in 15 of the first 20 turns. What would your call be? Heads or Tails (you are told that this is a fair coin that has not been modified in any way)? . Based on the results of the first 20 calls, you may be tempted to call Heads. However, you are aware that this is a fair coin, and therefore the probability of it landing heads up and tails up is the same. Since the coin has landed heads up more than tails up, ideally, it should have a higher chance of landing tails up and therefore Tails would have been a better call to make. This is based on a phenomenon called Mean Reversion – we explore this in more details in our note this month.