Jun 21st, 2015, 19 comments
Statistics and Trends are to many the most important factors to consider when looking for bets. Such historical data is pushed out by the media, writers, tipsters with great enthusiasm to please the masses and the average punter rarely stops to think that such data is available to the bookies as well.
Are statistics useful? Are they of any great relevance? Well, any mathematician will tell you they can be but, the problem is that they need using very carefully and things are not as black and white as is usually made out.
As an example, I expect that Manchester Utd have a good record against Wigan, or Bolton, or Fulham, or lesser team X but a stat that Man U have beaten lesser team X in 7 of the last 10 meetings is totally worthless information on its own.
So what is missing? The second part of the phrase below....
STATISTICS vs EXPECTED PROBABILITY ...Expected Probability. This is vital if you are to profit from analysing stats and trends. If the odds suggested Man Utd would beat lesser team x in 8 out of 10 games, then the 7/10 stat should be read as Man U have failed to beat ‘lesser team x' 3 times in 10.
As a more detailed example - let's look at the Coral CUP at the Cheltenham Festival. My data is a little out of date but that doesn't matter in illustrating the principal.
In the 2008 RP Cheltenham Guide - 'LONGBOW' points out that the race is a bad one for favs with only 1/10 successful in the past 10 years. That forms the STATISTIC part.
However, as is almost always the case, he has ignored the EXPECTED PROBABILITY.
If we look at the SP of the fav in the past 10 years and reduce the total book to 100%, we get a true idea of how many favs we would expect to have been successful. Last year the fav was 6/1 in a book of 142%. Previous years - 5/1 (142%), 10/3 (128%), 5/1 (122%), 4/1 (137%), 11/2 (132%), 7/1 (141%), 13/2 (139%), 6/1 (136%) & 9/2 (141%). If we crunch the maths, the expected number of winning favs would have been;
1.1 Winning Favourites So, our trend has actually proved that the number of winning favourites in the Coral Cup are about average. The way that 1/10 winning favs is portrayed as a negative pointer is simply incorrect and misleading.
Also, no horse older than 10 has won in the past 10 years. This statistic has a lot more logical truth to it as in this ultra competitive handicap, older exposed horses are always going to struggle. However, with only 5 runners in the past 10 years, when you crunch the numbers based on the SP's, then there is greater than a 99% chance that there would have been no winner older than 10!
Trend analysis is a much mis-understood tool. It can be very valuable but the press sometimes use it to please the masses without completing the task.
I don't have the space to go through it all but trends such as 8/10 won last time out are often more than explained by chance alone when you look at the SP of all horses that apply.
Another Example (Fictional);
a) In the last 10 years, 8 of the winners had finished first or second on their previous outing.
b) Backing all third favs in novice handicap chases at Taunton in Jan & Feb where the jockey has worn at least some yellow (except on Tuesdays) returned a LSP of +£33.20.
Which one of the above is stupid? Both of them are..
Seriously, when presented with trends, ask yourself if there is any scientific basis behind them or whether they are purely circumstantial. If there have been 8/10 winners carrying less than 10st 11lb - then read this as there have been 2 winners above 10st 11lb.
Feb 24th, 2018, 10 comments