Football predictions or predictions in any sport are hard to get right. There will never be a ‘lock’ or ‘sure thing’ in a sports event. Forget about what your buddy says or the statements made by touts and tipsters in the newspapers, over the web, or on TV. No ‘straight game’ is predictable to a 100% certainty.
Worse still, are the confidence levels applied to individual games by tipsters and handicappers. Some matches get a handicapper’s ‘Mega 10 Star Double Whammer’ rating. Other games get a standard ‘3 Stars’. There really can be no justification for this measure of confidence. The value on offer is tough enough to find as it is and no game warrants a stake size of two or three times greater than that of another game.
If you are betting on the NFL / football point spreads or making bets where the odds are hovering at around evens / +100, then you will be doing very, very well to win 57% of games over the long term, which means you are still losing 43%.
Why you lose those 43% (greater in most cases) can be due to ‘on field’ factors e.g. bad calls by officials, the team are not ‘at it’ or good enough on the day. Yet it can also be because of ‘chaotic’ factors. Any number of external influences can impact a sports event; the personal lives of one or more of the players, distractions in the crowd, management and franchise decisions, or even an altercation with the bus driver taking players to a stadium.
The vast majority of tipsters out there are scammers and charlatans, however, let’s assume that you do narrow down your choices and find a handful of honest handicappers. How do you know from their results, if they will be any good at helping you with your general sports or football predictions?
One of the issues that many bettors face when evaluating tipster results is what is known as the Law of Small Numbers. This is in effect a cognitive bias, where deviation from rationality occurs and individuals create their own ‘reality’ or ‘understanding’ of the data, from a relatively small sample.
Kahneman and Tversky’s Hospital Quiz
In the mid-seventies, psychologists Dan Kahneman and AmosTversky set up an experiment:
A town had two hospitals. In the bigger one around forty-five infants are born each day. In the smaller one, around fifteen were born each day.
We know that around 50% of infants are girls and 50% are boys. Yet naturally, the exact numbers of boy and girl infants born / percentages each day will vary.
For a whole year, the hospitals each recorded the days on which more than 60% of the infants born, were boys. Which hospital do believe had the higher numbers of such days?
- The smaller hospital
- The larger hospital
- Roughly the same (within 5% of each other)
The number of days where 60% of infants born were boys (girls being 40%) was around three times more in the smaller hospital because of the greater volatility in birth ratios. A bigger sample is unlikely to deviate much form 50%.
On the face of it this finding would seem quite logical but only 22% of subjects got the right answer.
So the conclusion drawn was that the vast majority of people will incorrectly rely on a small sample size as representative of a much larger sample of data.
So for instance, a small sample, which seems randomly distributed, would back up the idea that the wider population from which the sample is taken, will also be randomly distributed. By the same token, a small sample that seemingly shows a meaningful pattern e.g. nine out of ten coin tosses turning up heads, convince the observer that the population will display the same meaningful pattern. In this case the assumption would probably be that the coin is biased.
This human tendency to perceive meaningful patterns within random data samples is called ‘apophenia‘.
Kahneman and Tversky called the belief in the Law of Small Numbers and the wider group of ‘mental short cuts’ that people tend to take when making decisions with ‘uncertain facts’, heuristics.
One example of a heuristic is ‘gambler’s fallacy’, which does in fact come from a belief in the law of small numbers.
The gambler’s fallacy, is when people believe that if something occurs more frequently than usual over a period, then it will happen less frequently in the future and vice versa. The problem is that in situations where a sample is truly random, this ‘comfortable balance’ for the human mind, is false. This misconception is most commonly associated with gambling, where essentially the gambler believes that a ‘deviation in one direction will soon be countered by a deviation in the other direction‘.
Sports bettors can be quite vulnerable to believing in the Law of Small Numbers. It really is a classic scenario, that bettors very often incorrectly assume that long-term profitability will follow from small samples of bets.
Check out the results below – a profitability chart of one hundred bets on NFL games (points spread).
The above chart is quite typical of what a tipster or handicapper might show you as his record for football predictions. But check out the chart below after one thousand bets.
As you can see after a larger sample, over a longer period, the true picture is revealed. No profit. The first 100 bets shown in the first graph, now looks insignificant.
To ram home the point, these charts were in fact created by a random number generator with a 50% chance of a win in each game and a profit expectation of -2.5%.
With a profit expectation of -2.5%, the results were doomed to fail from the outset in terms of profitability, but notice how after over 600 bets they were still in profit.
Also notice that the pattern of results does not look random and flows in quite a pleasing wave pattern.
Take a look at the ‘one thousand bet’ charts below. Each chart one was randomly generated. The wide range of outcomes drives home the concept of how easily one can be misled by seemingly meaningful patterns of results.
Check out the chart in the middle. It looks just like a solid and profitable results chart from an excellent handicapper. Yet this sequence of results happened just by chance – effectively by ‘sticking pins’.
The table below gives you an idea of the likelihood of being in profit after a period of time with an expectation of -2.5%.
Note how over one thousand bets, there is still just over a 20% chance (odds of around 5.0 or 4/1) of showing a profit. Even after 10,000 bets, there is still a 0.51% chance (odds of 201.00 or 200/1) of being in profit. With a negative expectation, however, you are destined never to be profitable.
Try and remember this in your betting, regardless of whether your focus is on football predictions or predictions in other sports. It is crucial to understanding the nature of betting and being profitable.