What should happen is not always what does happen in betting and other things
Pullein on Football
Scientists publishing research always say there is less than a five per cent chance their findings are wrong. How often are they wrong? More than 50 per cent of the time. There is a difference between theory and practice.
John Ioannidis wrote an article in PLOS Medicine in 2005 explaining Why Most Published Research Findings Are False.
Then others added more examples.
In 2011 the pharmaceuticals company Bayer HealthCare compared findings from 67 published trials with their own and got a mismatch in 50. In 2012 a biotech company called Amgen repeated 53 famous experiments and got different results in 47. In 2015 the Center for Open Science repeated 100 experiments that had been reported in prestigious psychology journals and got different results in 61.
The non-replication rates, as we might call them, were 75 per cent, 89 per cent and 61 per cent – all more than 50 per cent.
Polls of voters in the United Kingdom come with a claim, which is typically that 19 times out of 20 they will be accurate to within two per cent. At eight of the last nine general elections they were out by more than two per cent. The average error was five per cent. There is a difference between theory and practice.
Nate Silver, a well-known elections analyst, reckons that 19 times out of 20 polls in the United Kingdom are accurate not to within two per cent but to within 15 per cent – that is 15 per cent one way or the other.
What matters is not what you are told will happen but what does happen. In science. In politics. In football and betting, which I will get to in a moment. And in many other things as well.
A planner responsible for the construction of a building will make an estimate of how much it will cost. They will get quotes for everything involved in every stage of construction then add them up. In theory that is what the building will cost. In practice it is not. The building will cost more.
The planner could get an idea of how much more by speaking to planners of other, finished buildings. By how much did the cost of those buildings burst the budget?
Using your own estimate is called taking the inside view. Using other people’s experience is called taking the outside view. The outside view is more revealing.
It is what Silver took even before he became famous for his predictions of the 2008 and 2012 United States presidential elections. In earlier years of the century he wrote for Baseball Prospectus.
Among other things Silver predicted how young players would progress. He did not do this by watching the players, or even by studying their stats. Or at least not by studying their stats in isolation.
What Silver did was search through the records of past seasons for other players with similar characteristics at the same age. Players with similar height, weight and stats. How did they progress? Usually in all sorts of different ways. Typically, though, what became of them?
Silver predicted the development of youngsters better than scouts or analysts who read each player’s stats in isolation.
Football bettors theorise that what a team did before they will do again. In practice, though, what will happen? We can get an idea by finding out what happened in the past to other teams who had similar records. What did they do next? That is also taking the outside view rather than the inside view.
A popular theory among football bettors is that teams who featured in lots of low-scoring games in the past will feature in lots of low-scoring games in future, and that teams who featured in lots of high-scoring games in the past will feature in lots of high-scoring games in future. How does the theory stand up in practice?
I counted the number of goals in each Premier League team’s games in the first and second halves of the last 20 seasons. Generally speaking, one total told me hardly anything about the other.
Teams whose games in the first half of a season averaged 1.7 goals played in games in the second half of the season that averaged 2.6 goals. Teams whose games in the first half of a season averaged 3.7 goals played in games in the second half of the season that averaged 2.8 goals. You get the picture.
There is a difference between theory and practice.
The Survey of Professional Forecasters collates the opinions of economists in the United States. It started 50 years ago in 1968.
By how much will the United States economy change over the next 12 months? The Survey of Professional Forecasters publishes a range that is supposed to cover most possibilities.
Only in one year out of ten should the economy grow or shrink by some other amount.
In how many years has the economy grown or shrunk by some other amount? In about half of them.
Read Racing Post Sport every day for no-nonsense previews and expert sports betting tips
Follow us on Twitter @racingpostsport
Like us on Facebook RacingPostSport