/cdn.vox-cdn.com/uploads/chorus_image/image/29931005/182110359.0.jpg)
The Birmingham Mail published an interesting statistical look at the rest of the season today and concluded that Aston Villa will finish 10th with 42 points. In the end, their results may prove correct, but the reasoning behind them is awfully flawed and shouldn't be used when justifying their point.
According to the article, they have examined how teams perform against "bands" of other teams in the league. In other words, the teams currently sitting 1st-4th are one band, 5th-8th are a second, and so on. From the performances against those bands, they predict the rest of the season. It's not the worst idea in the world, but in the case of Aston Villa, it is a bit skewed. Here's a look at how that breakdown would work for Villa. You can sort the table to view the matches played, points won, points/match, matches remaining, and expected points left for each of the five bands:
These numbers would indeed garner Aston Villa another 10.83 points. Round up to 11 and you've got a season total of 42. Remarkably, Villa have actually done better against the top teams in the league this year than they have against those in bands 2 or 3. It would be statistically very hard to argue that those results are anything more than a fluke. Through about 28 matches played, we have enough data to say that Arsenal, Chelsea, Liverpool or Manchester City are the top teams in the league. We also have enough data to say that Aston Villa are, rather definitively, not one of those teams.
Therefore, it would be wrong to expect Villa to continue their 1.16 points/match ways against the teams in the top band. Do we really think Villa will average a little better than a draw against Chelsea and in a visit to Manchester City? That seems incredibly unlikely to me.
Villa very well may make 42 points. They very well may finish 10th. But the information used to determine this in the Birmingham Mail's computer-aided study is not the right information. Don't pin false hopes on that report.