MY BROTHER arrived back at his office to find a man he had never met before, carefully and slowly painting his name on the door. He had written “Professor Anthony” and was halfway through our surname, when my brother tactfully pointed out that he was spelling it incorrectly. He would have to start again.

“You!” exclaimed the man, throwing down his brush. “You always spoil everything!” I know what he felt like. He’s got my brother. I’ve got David Moyes.

The Fink Tank Predictor has been performing exceptionally well. We have beaten the bookmakers season after season, and this year’s enhanced model is doing absurdly well. Except, that is, when it comes to Everton. The model that gives us the probabilities of game outcomes seems to be underestimating Moyes’s team.

More confusing still, our error does not appear to be consistent. This season and in the 2002-03 campaign, we underrated Everton but last season, if anything, we overrated them. What’s going on? Our attempt to see if there was any way to iron out this kink in our model led us to a fascinating conclusion. We are able to provide strong statistical support for a theory that Everton fans speculate about.

One of the equations underlying the Predictor, as developed by Dr Henry Stott and Dr Alex Morton, relates the average number of goals scored (GS) and goals conceded (GC) to the average number of points achieved. As they say on the terraces: average points per game = 1.25 + 0.68xGS - 0.6xGC. When this happens I’m over the moon. When it doesn’t, I’m sick as a parrot.

Fortunately, the equation holds most of the time, explaining 90 per cent of the variance in points scored. Unfortunately, there’s Everton. It seems as if Moyes has found a way of organising his team that makes them much better than usual at turning goals into results and just nicking the points.

In the history of the Premiership, the best team by far at turning goals into results were the Norwich City side of 1992-93. They actually came third, despite having a negative goal difference. They picked up an amazing 0.4 points per game more than expected, given the goals they scored and conceded.

Although they didn’t quite match this record, a couple of seasons ago Everton nevertheless did pretty well. They achieved 0.21 points per game more than expected and in a 24-game run at the end of the season this figure actually rose to 0.32.

But look at this season. Everton are, almost unbelievably, picking up 0.48 points more than you would anticipate with their goals record. Even if this does not continue at the same startling rate, it is still a real feat. With most sides, there is no correlation between their scoring rate and that of their opponents. With Everton there is. They score the goals they need. When they don’t need them, they don’t score them.

So what explains all this? Why did Moyes organise his side so well in 2002-03 and this season? And is it a clue that last season was quite different? Then, Everton were, just slightly, actually scoring fewer points than they should have, given the goals.

At first we were stumped. We looked at many, many different variables, but none provided us the explanation we were looking for. Then a keen observer of Everton suggested that Moyes may have been organising his side differently when he played Wayne Rooney. This could account for their reduced efficiency last year. Bingo! This turned out to be exactly right.

When Everton had Rooney in the side they were much less effective at nicking the points than when they did not. In the games since the start of the 2002-03 season that Rooney started, Everton scored an average 1.21 goals and conceded 1.21 goals. You would expect them to pick up an average 1.35 points per game. They picked up only an average of 1.06 points.

Now look at the games he did not start. Everton scored an average 1.22 goals and conceded 1.05 goals. You would expect an average 1.45 points per game, instead it was 1.78 points per game. It’s official. Everton are better off, far better off, without Wayne Rooney.

finktank@thetimes.co.uk