Premier League 2015/2016

For the past 2015/2016 season, I have been gathering shot data. I have used this to explore the various zonal conversion rates for each team.

Link to Document: https://public.tableau.com/profile/alex.rathke#!/vizhome/PremierLeague2015-2016SeasonReview/Dashboard1

Furthermore, I want to evaluate my Expected Goal (xG) model using this season’s shot data. What are Expected Goals (xG) and how does one calculate it you ask? If you are wondering, please read this post.

Teams

This slideshow requires JavaScript.

My xG model fits the data well as we can see from the R-Squared (R²) value of 0.80. As for the individual teams, Aston Villa, Crystal Palace & West Brom were the only three teams to under perform their xG for the season. This result is a bit skewed because of Crystal Palace’s astonishingly low win rate during the second half of the season. West Brom under Tony Pulis have  not been a very attacking team so that result is not exactly surprising.

Lastly, let’s examine Aston Villa. Or should we even bother? Everybody knew how bad Aston Villa were this past season. For me, all the problems started last summer with the recruitment process. I know a lot of people might argue that is an easy statement to make (after the season finished). I am not so convinced. When you buy four players (directly from Ligue 1) and try to merge them into one team asap it can get complicated. Language and culture for me are just one factor. Then there’s the competitiveness of the Ligue 1 versus the Premier League. Mini rant over. If you were to ask me though if there was one player that I would buy from the relegated Villa squad, who would it be? It would be Jordan Ayew (younger brother of Andre Ayew – Swansea City). I have loved his movement and skill when watching him play for Ghana. Thank you very much to Football Radars for the two analysis charts.

Players

When we look at evaluating my xG model and see how the Top 10 Premier League goal scorers do, the R² value is not great. That is due to better players very often “outperforming” xG. I suspect the correlation would jump up to 0.82 – 0.85 maximum when we include all players. Let us have a look if that is true.

Screenshot (177)

The second image includes all players from the 2015/16 season. As we can see the R² value jumps back up to 0.83. Of the players that surprised everyone this season, how did they do? Take a look at the graph and see for yourself.

Screenshot (189)