Cod Almighty | Article

by Andy Holt

30 August 2006

Welcome one, welcome all. Welcome to the first Ballpark Figures of the new season. Welcome, welcome, welcome.

Is it fair to draw the conclusion from your glum visages that you too are already fearing the worst for the season ahead and any lingering hopes of bettering last season's performance are lying shattered below the tea stain on your grandma's flock wallpaper? I know the feeling.

But fear not, as Ballpark Figures is here to save the day, patch together your broken hopes and determine whether Grimsby Town's performance in August will be any kind of indicator at all of their performance come the end of the season. My initial hypothesis is that it isn't and we can all stop panicking. But we'll see.

What I've done is take league position, points scored and points scored per game (ppg) for every season from 1919-20 to 2005-06 both at the end of August and at the end of the season (adjusting the measures to three points for a win for all seasons). I felt that these three measures would give a good indication of 'performance'. By plotting each end-of-August variable against its end-of-season counterpart, it would be easy to see if there is any sort of relationship evident, and thus whether we can tell how Town are going to do in the rest of the season given what we've seen so far.

So starting with league position, it appears there is some sort of relationship evident and it is in the direction we would expect: that league positions at the end of August correlate to positions at the end of the season, and vice versa. The R-squared value here, a measure of the strength of the relationship, is 15 per cent. If we use this relationship to predict Town's league position come the end of the 2006-07 season, we can expect to finish 15th.

Next, total points scored. Again there is a degree of relationship evident, with an R-squared value of 13 per cent, and again it is in the direction we would expect. Using this relationship to predict our points total come next May gives a figure of 60 points. Last season this would have given us a 12th-place finish.

And finally, points per game. Once again a relationship is evident, and once again it is a proportional relationship, as expected. The R-squared value this time is 12 per cent. This relationship predicts a ppg value of 1.30 for the season as a whole, again giving a total of 60 points for the 46 games.

None of these measures of performance, then, indicate anything near a play-off spot, and in fact we'll be lucky to get a top-half finish.

Of course, a major point here is that the above conclusions were all drawn from relationships with R-squared values of 15 per cent or less. And no statistician who knows what they're on about would dare draw any firm conclusions from so weak a relationship. Basically, then, final league performance cannot be accurately predicted by performance to the end of August. The same holds for total points scored and points per game. I conclude that we can breathe a huge sigh of relief and look forward to the rest of the season.

However, it might be worth noting that in the 80 seasons considered, only on five occasions (6 per cent) has an end-of-August league position of 18th or lower led to a end-of-season league position of seventh or higher. Sorry.

And one final point to bear in mind, to further temper your optimism, is that all of this analysis ignores the fact that the transfer window now exists and the team we have come 5pm on 31 August is the same team (bar the possibility of loans) that we'll have until 1 January at the earliest. This may mean that end-of-August performance is a better indicator now of performance at the end of season (or at least at the end of December) than it used to be. We may be in for a long, grey few months after all. Sorry. Again.