What is survivorship bias?

With success stories of professional bettors living the good life all over the internet, how easy is it to get consistent profits in the sports markets? Joseph Buchdahl, who has evaluated over 100 tipsters in his career as a betting analyst, has the answer. Are winners perhaps overestimated? Read on to find out.

When searching the internet for stories about betting performance it appears that losing bettors are a rarity; most of them just alk about success.

A quick Google for ‘sports bettors’ throws up famous sports bettors living the good life, the 10 most influential people on twitter for sports bettors and numerous guides, methods and techniques for becoming a winner. There’s almost nothing about failure at all, with the exception of a couple of articles explaining why most gamblers lose, and even one of those claims that it’s poor money management rather than poor picks which is the primary cause.

Actually, most bettors have negative expectation because of the bookmaker’s profit margin and insufficient forecasting prowess to overcome it; poor money management just magnifies the losses. Anyone might be forgiven for believing, firstly, that making a living from betting is easy, and secondly, that the bookmakers must all be about to go bust.

Plainly that isn’t the case, so what’s missing? The stories of all the losers; put simply, they are rarely told. This article revisits the idea of survivorship bias, previously introduced by Mirio Mella, and why it pays to avoid being fooled by it. 

History is written by the winners

History, it seems, is always written by the winners. The losers are ignored, because we choose not to see them (they tell a less interesting story, or one that elicits cognitive dissonance), because they’ve disappeared from view, or simply because we don’t bother to count them. It’s easy to be impressed with success if that’s all there is to see.

As Nassim Taleb, in Fooled by Randomness, narrates on the fantasy of monkeys attempting to recreate the poetry of Homer on a typewriter:

“If there are five monkeys in the game, I would be rather impressed with the Iliad writer, to the point of suspecting him to be a reincarnation of the ancient poet. If there are a billion to the power of one billion monkeys I would be less impressed…”

According to a Harvard Medical School study of nearly 8 million bets and over €60 million wagered, only 13% of players returned a profit.

Survivorship bias is the logical error of concentrating on the people or things that ‘survived’ some process whilst inadvertently overlooking those that did not because of their lack of visibility. Survivorship bias can lead to overestimating the chances of success because failures are ignored.

Its name was first coined during the Second World War when Abraham Wald, a free-thinking mathematician, solved the problem of where to put additional armour plating on the Allied bombers that were experiencing heavy losses.

Initially, engineers assumed that, through an examination of the bullet holes of returning aircraft, those areas that showed the highest concentration of bullet holes – in particular the wings and fuselage – needed the extra reinforcement. Of course, it didn’t work.

In fact the holes revealed where the planes were strongest, since these were ones returning. No one had given any thought to the planes that were lost. Counter-intuitively, following Wald’s advice engineers put extra armour plating where the bullet holes weren’t, with immediate results.

The illusion of cause and effect

When attempting to measure the probability of success in highly uncertain environments like sports betting, it’s no good just studying the sample that has succeeded. If we do, we risk turning causality on its head. Rather than suppose that success is caused by skill, survivorship bias ensures that we perceive the winners to be skilful because they have been successful.

As Michael Mauboussin explains in his book The Success Equation, the problem here is that inference is drawn from results, not the forecasting processes themselves. Where luck plays a major role in what happens there is very little connection between the forecasting process and the result.

If all you care about is results, you’re liable to draw erroneous conclusions. On the contrary, don’t study only the winners to see what caused them; study the forecasting process to see whether it consistently led to success.

What I learnt from evaluating tipsters

As a verifier of sports tipsters I became very familiar with the influence of survivorship bias. Between 2001 and 2011 I accepted 120 betting advisory services which had pre-existing track records. Aggregating these pre-verification performances together, 24,725 picks delivered a profit over turnover of 17.4%.

By contrast, the 90,451 picks that I verified thereafter made just 1.1%. There was no reason to doubt the credibility of the earlier histories. So why did so many tipsters regress towards the mean? I was simply being treated to the survivors whose previous winnings had largely arisen by good fortune. Those tipsters less fortunate had already disappeared, and there’s not much point asking to be verified if you’re no longer tipping.

What are the realistic chances of success?

If we only study the ‘winners’, we’ll never get a reliable idea of just how hard it is to succeed in sports betting. In an attempt to find some answers to this question, a team from the Harvard Medical School analysed the betting histories of over 40,000 of customers registering and playing with a leading online sports bookmaker in 2005.

Placing nearly 8 million bets and wagering over €60 million, only 13% of players returned a profit, with aggregated losses approaching -10% for the group as a whole, very much in line with the bookmaker’s typical margin.

Only 245 players showed profits of over €1,000. Of course for the majority of customers, betting on sports amounts to little more than guessing. Calculating from first principles, an unskilled bettor placing about 200 even-money wagers at a bookmaker with a 10% margin would typically have just a 10% chance of returning a profit. After 1,000 similarly-priced wagers, however, this probability would fall to just 1-in-1,000. The longer an unskilled punter bets for, the less chance he will have of being a winner.

Improve your chances with Pinnacle

Your chances of winning can be improved by betting with a bookmaker that has smaller margins. In this respect Pinnacle consistently offers the best in the industry, with typical margins of just 2% and sometimes even lower.

This time, an unskilled bettor would still have about a 1-in-4 chance of returning a profit after 1,000 even-money wagers. Furthermore, Pinnacle seeks to educate its customers through its series of betting articles, including betting strategy, forecasting methodology and bettor psychology, to improve their chances of success still further.

And unlike other bookies, Pinnacle actively welcomes sharp bettors instead of restricting them. Nevertheless, don’t let yourself be fooled by survivorship bias into believing that winning in betting is easy and everyone is doing it. To be a successful bettor takes hard work, and the longer you hope to stay ahead, the more work you will have to do to ensure that remains so.

Source: pinnacle.com