Predictions of weather forecasters are not to be sneezed at
Lessons for bettors from the field of meteorology
Coldness can kill you. Bad weather that spares your life can make you ill. Last week there was a lot of snow and this week there will be a lot of people sneezing and coughing.
Meteorologists warned of snow. They are good at forecasting weather. Doctors are bad at diagnosing illness, though this week they might be better than last week or next week.
Author Dylan Evans compared academic studies of weather forecasters and doctors in a book called Risk Intelligence.
Weather forecasters were asked every day for the percentage chance of rain on the next day. Doctors were asked the percentage chance that patients were suffering from pneumonia.
Among weather forecasters there was a strong correspondence between predictions and outcomes. It rained on about ten per cent of the days for which they said there was a ten per cent chance of rain, on about 20 per cent of the days for which they said there was a 20 per cent chance of rain, and so on.
For doctors there was essentially no correlation between predictions and outcomes. A patient given a 90 per chance of having pneumonia was no more likely to have it than one given a 50 per cent chance.
Most people think more highly of their doctor than of weather forecasters, which only goes to show that professionals do not always get the reputation they deserve.
Why are weather forecasters good at predictions and doctors bad? Evans repeated three possible explanations suggested by Sarah Lichtenstein, an academic who had studied judgement.
First, specialisation. Every day forecasters are asked the same question: how likely is rain tomorrow? Each patient a doctor examines, though, could show signs of a different potential illness.
Second, practice. Weather forecasters have expressed the chance of rain as a percentage for about 50 years. Until recently they keep the number to themselves because they did not think the public would accept talk of possibilities rather than assertions of certainty, but that is another story. Weather forecasters are used to thinking of chances and quantifying them in numbers. Doctors are not.
Third, feedback. Weather forecasters always get feedback, and they always get it quickly. Today they will be asked to estimate the chance of rain tomorrow. Tomorrow there will be rain or there will not. Doctors, on the other hand, might never hear what happened to a patient who left their consulting room clutching a prescription.
Weather forecasters process their feedback systematically. They plot graphs of predictions against outcomes. These help them to see where they could do even better. If, for example, one weather forecaster is underestimating the possibility of rain when they say there is a 70 per cent chance it will show on their graph. Then they can try to figure out what they are missing.
Perhaps better betting includes trying to be less like a doctor and more like a meteorologist. We can specialise, bringing everything we have learned elsewhere to bear repeatedly on one sport, or just one part of one sport. We get plenty of practice. We also get speedy feedback, though maybe we do not always analyse it as thoroughly as we could.
More by Kevin Pullein
Beauty is in the eye of the beholder but talent is not
Remember Gonzalo Higuain’s shot that opened the scoring for Juventus against Tottenham in a Champions League last-16 first leg? He was facing away from goal then turned to volley a free kick as it came past. It was wonderful to watch.
The first leg ended 2-2 and the second leg is on Wednesday.
Higuain scores for Juventus once every five or six shots. Most of his shots are taken from nearer the goal and when the ball is easier to strike. As he turned then swung his foot at that free kick the odds against him scoring must have been about 30-1.
When someone scores a spectacular goal we drool over it and watch replays again and again. Look at that body shape, we say, what a perfect contact with the ball.
Every other time Higuain finds himself in that position, however, he will feel that he has performed the same technique, even though most other times the ball will not go in the net.
All professional footballers could have scored Higuain’s goal. The only difference between them is that in those circumstances some would miss more often than others.
A player’s skill when they score is not in what you see but in what you do not see – how many other times in the same situation their shot would go high, wide or spin off the boot and curl out for a throw. The lower that number the higher the skill.
Overall it is lower for Higuain than for many others – though the same as for Harry Kane, who scores once every five or six shots for Tottenham. So Tottenham fans may hope that Higuain, who missed Juventus’s last two matches with an injured ankle, recovers on Thursday and does not play at Wembley.
Where a team win the ball is a pressing concern
There is a technical justification for the tactic of pressing, which is why it has been used by coaches as varied in their football philosophy as Johan Cruyff and Graham Taylor.
You cannot score unless you shoot, and you cannot shoot unless you get the ball near to the opposition goal. If you press and win the ball in the last third of the pitch you have already performed the first necessity.
I recall Cruyff saying something like this: “When you lose the ball you want it back. Usually it is still nearby. Maybe only a few metres away. So why turn round and run away?” Most players do.
Sometimes I watch a match holding a notebook and pen and write down where each possession starts and ends. I have recorded only those moments in only a few dozen matches. Researchers who have recorded every incident in hundreds of matches found the same thing: the higher up the pitch a possession starts the more likely it is to reach the last third. If it starts in the last third it is already there.
Manchester City manager Pep Guardiola played for and adored Cruyff. Next time you watch Manchester City count how many times they win the ball in the last third. Usually it is a lot.
Follow us on Twitter @racingpostsport
Like us on Facebook RacingPostSport