@CoadyBenjamin I think you want someone nailed so that rules Bayindor out.
The other 2 are pretty close in terms of best combos of match up. Perri with a score of 142 and Hermansen with 143, not significantly different.
Go with however you like best/think is most nailed.
@SolioAnalytics Tool looks great. I heard discussion on a pod about predicted xGI for every player over the season. Is that data available on the site.
The original matrix provides a number which is the sum of the best GK fixture each game week for the 2 teams. It is therefore the predicted sum of the 2 GKs if there rotation was perfect and the only factor was opposition.
Newly promoted clubs have all been given an PPG of 4.27.
Conclusion.
There is less variability in combinations than expected.
The key take away is to avoid drafting Sels and Areola together.
Matrix is best used to check your 2nd keeper rotation after picking your 1st.
Home and away.
When looking at differences between home and away performances there was minimal difference GKs scores at home ore away.
Surprisingly 14/20 teams preformed better when playing away than at home.
It has not been factored in to the matrix.
Low GK PPG allowed teams.
As expected, Liverpool and Arsenal do not allow opposition GKs to score many points.
Brighton surprisingly very poor fixture for a GK.
Man City were average last season, coinciding with their general underperformance.
FPL Draft - GK fixtures matrix.
This matrix shows the best pairings of teams for GKs with complimentary fixtures over the 38GWs.
TLDR:
1. Not huge variation across the season.
2. Avoid drafting Sels and Areola together.
3. Palace have some good combos.
High GK PPG allowed teams.
As expected the relegated teams were high on the GK PPG allowed.
Not as expected Man Utd were top with 4.27. This could be due to a combination of poor finishing and taking lots of long shots allowing multiple easy GK save points.
Method
In order to best predict points I used the data from 24/25 season, and totalled the number of FPL points each team conceded to opposition GKs.
I only included matches where the GK played a full 90mins (not subbed) and took an average to give PPG conceded to GK.
Background.
GKs have 2 primary ways of point scoring:
1. Clean sheets - more likely to accumulate vs poor opposition attacks.
2. Saves and BPS - more likely to accumulate vs strong opposition attacks.
Therefore a paradox in using fixtures to predict points.
As usual raw data used from @Draft_FC.
Each of the 38 plots on the overall line only has 40 data points to determine the correlation. I would expect the line to smooth out with more data points.
Inspired by the Athletic’s football pod talking about the statistical correlation between PL position after 10GWs and final position - Spoiler it’s 77% - I decided to look into the data for my 10 person draft FPL league over the past 4 years. 🧵
Looking more closely at the business ends of the table (we have a reward for first and Sacko punishment for last, nothing for 2nd to 9th).
By GW25 whoever was in 1st stayed there until the end.
By GW22 whoever was in 10th stayed there until the end.
End 🧵
Looking at the overall correlation there is a spike in correlation at the 10 GW point - Up to 65%.
Predictably lower than the real life PL due to being a lot more luck based.
This number goes up to 80% at GW 29.