Back to index

Data

The figures below are based on data from all Western Conference games played in 2014 plus an increasing number of Eastern Conference games. In particular, all games involving the DC Current are already included.

Scatterplot: completion percentage vs number of throws

The y positions (completion percentage) in the following two scatterplots are randomly jittered a bit to reduce overplotting. Excluding players with less than 5 throws.

plot of chunk scatterplot_comp_perc_vs_throws plot of chunk scatterplot_comp_perc_vs_throws

Barchart: player’s share of team’s total season goals

Figure includes players with rank 15 or better for total season scoring. Due to ties this may not produce exactly 15 players. Players with goal share > 0.05 are labelled. plot of chunk barchart_share_of_goals

Barchart: player’s share of team’s total season assists

Figure includes players with rank 15 or better for total season assists. Due to ties this may not produce exactly 15 players. Players with assist share > 0.06 are labelled. plot of chunk barchart_share_of_assists

Barchart: player’s share of team’s total season D’s

Figure includes players with rank 15 or better for total season defenses. Due to ties this may not produce exactly 15 players. Players with defense share > 0.06 are labelled. plot of chunk barchart_share_of_ds

Scatterplot: player’s share of total season goals vs D’s

plot of chunk share_of_goals_vs_ds

Barchart of no. of passes in a possession

I have excluded possessions that end due to end of period. What this means via example: Vancouver Nighthawks have had 226 possessions that end in a goal and the median number of passes was 5.

Main observation: vanNH turn it over more and sooner than wdcCT. But also take fewer passes to score. plot of chunk barchart_passes_per_possession

Densityplot of no. of passes in a possession

plot of chunk densityplot_passes_per_possession

Strip-and-boxplot of no. of passes in a possession (whole MLU)

wdcCT resembles pdxST in terms of making more passes than teams like vanNH. Only sfoDF have shorter possessions than vanNH, but luckily vanNH’s possessions end well more often. plot of chunk strip_and_boxplot_passes_per_possession

Strip-and-boxplot of no. of passes in a possession (vanNH and wdcCT)

Head-to-head demonstration of wdcCT making more passes, across all possession outcomes. plot of chunk strip_and_boxplot_passes_per_possession_finals

How possessions end, high-level.

y = how possessions end

x = proportion of possessions that end a certain way

Alt text

How possessions end, high-level and by line.

I use o_line to denote a line that was sent out to receive the pull and play offense. I use d_line to denote a line that was sent out to pull and play defense. Of course, if there’s at least one turnover, an o_line plays defense and a d_line plays offense. How do possessions end if we split out by which type of line is currently on offense?

Caveat: I am not (yet) adjusting for the full line changes we often see during timeouts. But that affects a small proportion of possessions.

In an absolute sense there are more possessions by o_lines but the distribution of how the possessions end isn’t very different at all.

x and y = same as above

Alt text

How possessions end, detailed.

We revisit the same figures as above, but with a more detailed look at how possessions end. Here’s what the codes mean:

y = how a possession ends

x = proportion of possessions that end a certain way

Alt text

How possessions end, detailed and by line.

x and y and meaning of o_line and d_line = same as above

Alt text