Betting Masterclass Volume 4: Mark O'Haire on the football stats that don't matter

Way back in March 1950, Swindon beat Bristol Rovers by a solitary goal in a largely uninspiring contest at the County Ground. The victory for the Robins was just one of 15 accumulated by Louis Page's side that season as the Wiltshire outfit posted a 14th-placed finish - a disappointing return having hit the heights of fourth in the preceding campaign.

As thousands of fans grew frustrated with Town's first-half display against Rovers that day - the goal came in the second period - one man in particular found Swindon's pattern of failed attacks frustrating.

Growing disillusioned with his side's apparent slow and inefficient forward play, Charles Reep, an RAF Wing Commander, unexpectedly pioneered football performance analysis for the future by taking out a pencil and notebook from his pocket and noting down details of all the attacks from his team in the second-half to see where his side were going wrong.

With these simple tools Reep started to invent a method of recording what happened to every possession on the field and his eventual conclusions - having gone on to record and analyse more than 2,000 games, up until the mid-1990s - suggested that moves consisting of three passes or fewer were more likely to result in a goal than longer passing plays.

The theory was scorned by those within football blindly tied to the merits of possession football - but a handful of managers, including Graham Taylor and Dave Bassett, based their style of play on Reep's ideals with spectacular success. The primitive findings are credited to this day with the invention of, and continued fascination with, the English long-ball game.

The strategy has been debunked since and Reep and his disciples have been criticised for a perceived misinterpretation of data, as well as a wilful ignorance of essentially all other factors that influence a particular football match. For example, why should a tactic that worked well in December in Hull be applicable to an El Clásico in August?

xG explosion a sign of how far stats have come

Nevertheless, Reep's work deserves recognition and appreciation. His attempts outweigh his achievements and his process can be considered to be the first steps towards today's age of analytics in football and betting. The Expected Goals (xG) explosion has dwarfed previous incarnations of basic shot data and the landscape has never been greener for budding football punters armed with a range of data points to attack the weekend's action.

But there are certain rabbit holes we should still steer clear of. Possession is often used in television and deserves its own place in the sphere of stats, but it's a figure that should not not be used in betting to measure success or team strength. As Leicester's title-winning side showed, it's about what you do with the ball, not how long you harness control for.

One golden rule: ignore H2H stats

However, there is still a more meaningless trend that should largely be ignored - head-to-head (H2H) stats. There is a widespread belief in football that certain teams will always play well against others, irrespective of their form or league position. But the bogey team is as much a figment of our imaginations as the bogey man.

Past results in a fixture have no influence over future happenings. Derby fixtures can be classified in a different bracket, and of course, the psychological aspect of H2H records deserves a mention too, but the bland fact is, teams who seem to have some sort of spell over their opponents have simply been lucky the majority of the time.

The best place to search for evidence of bogey teams would appear to be the Football League. In the Premier League, teams, particularly those such as Manchester City and Liverpool, have good historical records against the majority - not because they are bogey teams but because they are consistently the best. In the EFL, unlike the top-flight, the best sides leave at the end of every season, meaning clubs who stay within a division are likely to play at a similar level.

The figures that follow are from a study of Football League games played last season between teams who had met at least five times in the previous 10 seasons. If they had met that often, it would seem reasonable to assume that they were usually of a broadly similar standard.

In that campaign, 42% of games finished in home wins, 31% in draws and 27% in away wins. In fixtures that had not produced a home win on any of the previous five occasions they were played, the proportion of home wins was 42%. In fixtures that had not produced an away win on any of the past five occasions, the proportion of away wins was 27%.

In other words, teams who had been doing very badly in a fixture, did no worse this time round than everyone else, and vice versa. It is the best evidence that bogey teams do not exist. And, obviously, we should not bet on phantoms.

***

Be sure to check out the other volumes in our Betting Masterclass series, listed below:

Volume 1 - Ed Hawkins on Test Cricket
Volume 2 - Ed Hawkins on Twenty20 Cricket
Volume 3 - Ed Hawkins on how to bet on ODI Cricket