Introducing wCP

Quarterback completion percentage is a weird thing. “Good” quarterbacks are expected to complete a bit more than 60% of their passes, but the percentage on its own has no context.

Completion percentage doesn’t convey a quarterback’s accuracy very well because dropped passes, throw aways, spiked balls, and other anomalies are all counted as attempts. On top of all that, passes of different distances or lateral depths aren’t equal.

Some sites and analysts, like the folks at Pro Football Focus, rate accuracy by calculating an on-target percentage. That seems like a good start since it controls for events out of the quarterback’s control and passes that aren’t meant to be completed, but it doesn’t factor in the pass type or depth. Furthermore, determining a dropped or a pass that should be caught is very hard to consistently do without biases.

Pro Football Focus, does, however, provide zonal pass numbers for every quarterback without the passes thrown away, tipped at the line, or spiked. After pulling all of the data for quarterbacks who played in at least 25% of their team’s offensive snaps, I was able to determine league average completion rates by the zones they provided.

2013 NFL avg cmp rate

Using these zones and each quarterback’s individual performance, I was able to determine the percentage above or below league average to each zone. I then weighted the above or below average rates by the proportion of attempts thrown to the various zones and aggregated them to get one number.

The idea is to compare all passes of a similar distance (including lateral distance) and determine which quarterback has the best weighted completion percentage (wCP).

Finally, I set the baseline to 100 (like some baseball metrics) so that anything above average would be greater than 100 and below average results would be less than 100. A perfectly average rate would be 100. The results of the test are below (AFC East starters are italicized).

2013 wCP

The league average completion percentage was 61.2%. Philip Rivers had the highest raw rate in 2013, but Aaron Rodgers had the best weighted rate. It wasn’t even close, really.

Rodgers outlier wCP

Meanwhile, the AFC East didn’t feature one quarterback with a wCP over 100. Take a look at where the starters all finished the 2013 season.

AFC East wCP

This metric is still in its infancy stages, but it could lead to a better way to view completion rates. I’m going to try to add more seasons to get a better baseline for the averages. Those added years could also lead to some predictive applications in the future.

Lastly, once I gather more data, I will be able to determine whether more or fewer zones should be included in this study. (For example, is there a significant difference between all passes twenty or more yards downfield and those thirty- to forty-yards past the line of scrimmage? Is there a significant difference between passes between the numbers and along the sideline on short throws? Are passes aimed along the sideline different than those just outside the numbers?)

Understanding the context of plays is really important. Weighted completion percentage tells a much better story than just completions divided by attempts.

About these ads

2 thoughts on “Introducing wCP

  1. Have you thought about capping the wCP so it does not go over 100? essentially, figuring out a way to cap it at 100 and deciphering how you could show some sort of value over adjustment where if a QB is below average his grade essentially would be negative might lend more credence and a more accurate depiction of his overall accuracy based on you initial model.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s