• This website uses cookies. More information.
  • The This Is Anfield Forums community is moving to a new home. Click here for more information on the transition.

LFC Analytics/Data Thread


It's all for me grog
Jan 28, 2020
Spent some time yesterday and today in the FiveThirtyEight stats, particularly focusing on the spi_matches sheet and the Soccer Power Index (SPI) rankings. Based on a scale of 0-100, they are not fixed for the season; they tend to fluctuate from week to week. Also interesting to note that the maximum SPI in 16-17 (for the whole Premier League) was 88.4; this year, it was 95.75. Objectively, I knew that the quality in the League was better this year than in 16-17, but I guess there is something to teams winning the League with 98/99/100 points being better than those in the high 80s/low 90s!

With all this in mind, I was eager to see how a team's SPI progressed (or regressed) between 16-17 and 19-20. At the same time, looking at SPI gains alone can be misleading; for example, our gain of 11.62 SPI points between 16-17 and 19-20 is less than Burnley's gain of 12.0, but would you argue that Burnley are a better side? I think not.

So, I came up with a formula (that is definitely not perfect), but figured it would better reflect not just a team's improvement but how close that team was to the top in those respective seasons. So, a team like us which went from an average of 81.13 in 16-17 to 92.75 in 19-20 would get its due versus a team like Burnley that went from 57.72 to 69.72 despite the latter being a bigger gain. Levels!

Anyway, here's the formula (for a given team):

(Team Average SPI 16-17 + Team Average SPI 19-20) / (League Maximum SPI 16-17 + League Maximum SPI 19-20) * 100 = SPI Graded % Margin

Below: The top 4's SPI progression with SPI Graded % Margins (click for full size).

It's not perfect: for example, I hardly think Chelsea are just 2% worse off than we are in terms of their progress in that time vs. ours (interesting they regressed between last year and this year), and it's also interesting to note that we had a higher SPI than City's centurions. Moreover, City had a higher SPI this year than they did in that year, which I have a tough time wrapping my head around. That said, given the increase in maximum SPI levels increased between 16-17 and 19-20, perhaps the whole "the League is shit" talk was all bollocks. Also fun to see that the Mancs can't crack 90. :cool: