With half the English Premier League season gone, the January transfer window is opening. Some time ago, NYA developed a powerful algorithm for determining clubs’ top priorities for improvement. Based on historical tables, the algorithm projects how changes in goals scored and goals conceded would affect a team’s likely finishes in the league for several seasons to come. I’ve dusted off the algorithm and updated it to offer some guidance in this window.
The original version of the algorithm places a cash value, at the margin, on goals scored and conceded based on a club’s current league position. The values, which are based on payments from competitions, tend to be biggest for clubs on the edge of the relegation zone or on the verge of qualifying for Europe. But to simplify matters, I’ll just stick to positions for this exercise. I want to know whether enhancing offense or defense would make moving up the table more likely.
An underlying insight here is that a single goal can change a club’s fate for years to come. A club that finishes high in the table one year will have more money and a good chance of ending up just as high in the following year. So there’s a snowball effect for every change a club makes. With that in mind, let’s look at some pictures.
First, here’s a plot summarizing where teams can expect to finish given their goal totals, based on the past 18 seasons:
Now, here are the positions they might gain over the next five seasons by increasing the number of goals they score:
And here are the positions they might gain over the next five seasons by decreasing the number of goals they concede:
So what should the teams in the current window do? For simplicity, I’m going to assume that they’re on track to finish the season with double their current totals for goals scored and conceded. That’s a pretty fair assumption these days, given this graph from four recent seasons:
Using the numbers from the earlier plots, I can estimate how many places a team might gain by scoring one more goal or conceding one fewer. The values in the chart below are the predictions for total positions gained in the table over five seasons for each marginal goal. For clubs where the difference between strengthening offense and defense was less than half a position, I called it a draw.
For 11 of the 20 clubs, the algorithm finds a clear priority. So, could Sunderland really gain almost 0.6 positions per year just by conceding one less goal in the next 19 games? Perhaps, if that change significantly decreases the probability of relegation. On the flip side, if Manchester United can give up one less goal, it might raise the likelihood of Champions League football – and the money it brings – sufficiently to increase the chance of future finishes in the top four as well. The caveat here is that the algorithm bases these figures on historical data; Sunderland may conclude there are so many bad teams in the league this season that the risk of relegation is unusually low, even for a team with its poor goal totals.
As for the leaders, Chelsea and Manchester City, both clubs have goal totals that would normally be good enough to win the title. The algorithm doesn’t see much benefit to bolstering their squads on either offense or defense. Everton and Aston Villa are the oddballs; their goal totals are peculiar enough that there’s insufficient information to predict where they’ll finish.
This exercise suggests more of a rule of thumb than hard-and-fast guidance for clubs; each club will have a more specific idea of where its weaknesses lie. But there are two important lessons here. First, a club’s biggest weakness isn’t always the most pressing area for improvement. If the objective is to finish higher in the table, transfer priorities will depend not only on the makeup of the club in question but also on the context provided by other clubs’ positions. Second, transfers in a single window can affect a club’s fortunes for years to come. Keep that in mind as you watch Jim White a month from now.