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.
x = how possessions end
y = proportion of possessions that end a certain way
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
New observations?:
x and y = same as above
To be clear here, within a team * line combination, the bar heights add up to 1. So we get the breakdown of how possessions end when, e.g. a Portland Stags’ o_line
had the disc: 48% end in a goal for them, 30% end in a turnover nominally due to offense error, and 19% end with an explicit stop by the defense.
One thing that I see is that vanNH’s o_line
and d_line
are more similar to each other than those of the other teams.
x and y and meaning of o_line
and d_line
= same as above
We revisit the same figures as above, but with a more detailed look at how possessions end. Here’s what the codes mean:
x = how a possession ends
y = proportion of possessions that end a certain way
x and y and meaning of o_line
and d_line
= same as above
x and y = same as above
x and y and meaning of o_line
and d_line
= same as above
self
scores means a holdopp
scores means a breaknobody
means time ran outrecv_team
indicates who received the pull
Of all the goals scored, what proportion were scored immediately off the pull, i.e. in exactly one possession? What proportion were scored by a d_line in exactly two possessions? Etc.
x = the number of possessions before a goal is finally scored
y = the proportion of points scored after x exactly possessions
color indicates hold = receiving team scored vs. break = pulling team scored
x = same as above
y = cumulative proportion, i.e. proportion of points scored within x or fewer possessions
color same as above
x = the number of possessions before a goal is finally scored
y = the proportion of points scored after x exactly possessions (relative to all points scored by a team)
color indicates hold = receiving team scored vs. break = pulling team scored
Observations not updated after recent large intake of games!:
95% of all goals are scored in 4 or fewer possessions. Note: for this, I have removed points that end with time running out vs. someone scoring.
When an o_line receives the pull, what proportion of the time do they score without turning it over? Think of that as a special case of a more general goal conversion rate: Of points with x or more possessions, what proportion have exactly x possessions?
Note: for this, I have removed points that end with time running out vs. someone scoring.
Yes, I actually put 95% confidence intervals down there, to make it clear there’s no evidence that the conversion rate differs by which possession it is. In particular, it doesn’t seem to oscillate due to possessions being handled by an o_line vs. a d_line.
x = number of possessions
y = goal conversion rate
Although I can resolve individual throws from the data I have, I have not done so yet. Actually I have now, so this figure will become obsolete soon! For now, I measure the “length” of a possession in terms of the number of events, which is highly and positively correlated to the number of throws. There are more events in a possession than throws, since there are events for defensive fouls, time outs, pulls, etc. However, this affects all teams uniformly, so this “cheat” shouldn’t distort the story below re: possession length.
The Stags’ possessions are notably the longest of all Western teams. The boxplots below give the median possession length.
Possessions leading to goals last longer than those leading turnovers. I guess that’s to be expected. For possessions ending in a goal, the Stags’ possessions are the longest. Somehow the Rainmakers beat them out re: possession length for possessions ending in a turnover.
Possessions leading to goals last longer than those ending with an unequivocal D. And both last longer than possessions that end because the offense gives it up. And still, regardless of the outcome, the Stags’ possessions are the longest.