After spending way too much time importing stats into a database, the first thing I did was try to determine whether or not our defence is forcing bad shots. They are, to some extent. I re-did the zone thing from my last update slightly:
This was done primarily to consolidate zones that had very little activity, and break heavy traffic areas down further. I'll probably change it again later, but this seems to work better for now. I multiplied the number of opponets' shots taken from each zone by the league-wide success rate for that zone, which gives us an expected number of goals from that zone:
| Zones | Shots | Goals | Success % | EGA |
|---|---|---|---|---|
| 1+2 | 23 | 6 | 0.261 | 6.394 |
| 3+4 | 92 | 12 | 0.130 | 14.076 |
| 5+6 | 52 | 2 | 0.038 | 5.772 |
| 7+8 | 21 | 0 | 0.000 | 1.365 |
| 9+10 | 56 | 3 | 0.054 | 2.912 |
| 11+12 | 3 | 0 | 0.000 | 0.099 |
| 13+14 | 110 | 2 | 0.018 | 4.070 |
| 15+16 | 75 | 1 | 0.013 | 1.125 |
| 17+18 | 1 | 0 | 0.000 | 0.064 |
| 19+21 | 10 | 0 | 0.000 | 0.070 |
| 20 | 41 | 1 | 0.024 | 0.574 |
| 22 | 1 | 0 | 0.000 | 0.000 |
Adding up the totals from the last column gives us 36.521 Expected Goals Against. We compare this to the league average goals scored/conceded per club (currently 38), and get something that I hope allows us to separate goalkeeping and outfield defence a little bit more. This puts Fulham's SQA at around .962, compared to the league average of 1.000, assuming I did this correctly. This means that our opponents' shots do tend to be on the speculative side. Some of the difference between EGA and our actual total of 27 conceded can be attributed to goalkeeping. There's still more to it than this (how do we measure contested shots, for example?), but it helps give credit where credit is due. Konchesky, Hangeland, Hughes, Paintsil, and the rest - excellent work.
Much more to come on this as it happens.
Post new comment