There are times when Fink Tank makes me nervous. And this is one of those times. You see, I don't have any control over what the Fink Tank model says. Neither do Dr Ian Graham or Dr Henry Stott. We are in complete charge of the methods it uses, but once we have pressed the button to output the data, that's it.

When selecting the Fink Tank player of the year we have long debates about the right way of designing the model. We are able to defend the decisions we make, but we can't tweak the results if they don't fit with our intuitions. Do that and it's all over for the Fink Tank. The whole point is to be objective.

So you can see why I am nervous as the PC whirrs away, ready to spit out the results. If Robbie Savage emerges as our player of the year, I have to live with it.

Fortunately, he doesn't. So what is our method? We take every player action and use something called a multivariate Poisson log-normal model to work out the correlations between all of the different actions. With these correlations, we know how important good passes are for scoring goals, bad passes are for conceding goals and so on.

This allows us to look at each player in turn. The computer work required to do this gives an insight into why you cannot do it with the naked eye.

You have to simulate the entire league season with the player in the team and with an average player in the same position. This involves doing it again and again and again. And again.

Now, some fans look at our data and are convinced that they know better than the model who has done well and who hasn't. But think what this means - that they have watched not only this player but every other in every game and can integrate that information in their head.

Before we can move from the model to the ranking, there are a couple more decisions that need to be made. First, should we adjust the numbers to reflect the different time spent on the pitch by each player?

We think that staying in the first team is part of being player of the season, so we did. Second, we can either look at a player's contribution to his team, or place him in an average team. The first approach really rates players who perform comparatively well in poor teams. The second rates more highly players who play in good teams. We use this second approach.

So what was the result? No prizes for guessing the winner - Cristiano Ronaldo wins easily, as he did last year. More surprisingly, the runner-up is David James. The large amount of time spent on the pitch is one reason he came close to winning the whole thing. Tomasz Kuszczak, of Manchester United, would have been player of the season if he had played as much as James.

This season, unusually, Steven Gerrard finished above Frank Lampard, who would have finished in third place if he had played as much as in previous seasons. The Fink Tank is behind the great Martin Samuel in insisting that Lampard is among the best - and the data bears this out season after season.

Again Robbie Keane makes it into the top ten. He has been one of the truly big players season after season without, perhaps, receiving quite the credit he should have.

Savage? Nowhere. I breathe a sigh of relief.

Explore Football

• Premier League

• Football League

• Scotland

• European Football

• International Football

• Live Match Tracker

• The Fink Tank

• Fantasy Football