In sports betting, the idea that outcomes can yield to the application of information and knowledge about a team or player can give rise to an exaggerated sense of belief in one’s predictive ability. A little information is a dangerous thing.
Furthermore, our self-serving attribution bias ensures that we are more likely to associate forecasting successes to internal attributes (I am a skilled forecaster and my skill led to a correct call), whilst associating failures with external attributes (I was unlucky).
Despite being unpalatable to our craving for control, the reality is that, like weather forecasting and the stock market, sports betting is inherently a very uncertain business, where the evolution of a game is complex, chaotic, and conceivably even non-deterministic if we concede that what takes place might be largely influenced by the quantum world.
Of course, most punters appreciate that, on a bet-by-bet basis, luck (either good or bad) has a lot to do with whether we win or lose. But how much does chance influence things over longer periods?
To ensure that we are not fooled by randomness, a useful exercise is to analyse just how much inherent random variability actually exists in sporting outcomes.
One way we can do this is plot a time series of hypothetical betting returns from fair odds to see how much they vary over different time scales. Betting odds merely represent probability estimates for our expectations.
The wisdom of crowds ensures that, on average, these odds prove to be a very reliable indicator of ‘true’ probabilities. But randomness ensures that outcomes frequently deviate from idealised market expectations.
The sharpest punters recognise that whilst they are better at forecasting sporting outcomes than the remainder of their competition, most of what happens in sports is largely a matter of chance.
As Nate Silver, who demonstrated that a skilled poker player could still be showing losses after 100,000 unlucky hands, might say, the signal is weak and the noise is loud. Read his book – Fooled by Randomness.