Their sports stuff lags well behind their political modeling, but in this instance, it's interesting how closely they track with the odds. That could simply be a product of the truncated remaining schedule...if your model is off a little here or there, it'll distort things far more with 20 games to go than with 8.
The big issue with SPI is that it's xG (and xG variants) based, and while I like x-G, there's a growing appreciation for the fact that some players consistently outperform that metric, and others underperform it, and it owes partially to the next horizon in determining the likelihood of a goal being scored: the positioning of the other players. If you're shooting from 15 yards straightaway on your preferred foot, those chances aren't created equally; if it's Jamie Vardy played in with none save the keeper to contend with, the xG from that situation is necessarily going to be higher than Nathan Redmond having to hit a first time ball through a thicket of players.
Hence the overrating of our attack: we get a fair number of shots from relatively dangerous areas, in situations that mitigate the danger somewhat, taken by players who aren't noted finishers. Same thing is happening with Palace; SPI rates them extremely well in attack, because they're currently seventh in xG, yet they're 19th in goals from open play. Some of that is Benteke's struggles (and he has historically rated pretty well in xG), but they also just have a lot of players who struggle to hit the target...when your three leaders in shots taken are a confidence-shy Benteke, plus Zaha and Townsend, you're going to generate and then squander a lot of half-chances.
