Earlier this season Mike Backherms from Big Cat Country supplied us with his Offensive and Defensive Success Rate data for the 2015 season. His work is inspired by Bud Elliott (Tomahawk Nation) and Bill Connelly (Football Study Hall). This data is great for understanding how well offenses and defenses perform in certain situations, providing key context for week by week or season-long stats.
Offensive Success Rate measures the percentage of plays that a team gains enough yardage to put itself in a good position on the next play. Success is determined by the situation at the start of the play. On first down, a successful play gains half of the yardage needed to move the chains. Second down is considered a success if 70 percent of the needed yardage is gained. Third and fourth down success is determined by whether or not a new set of downs is earned.
Plays are also divided into standard downs and passing downs. Standard downs are 1st and 10s, 2nd and 7 or fewer yards to go, and third and fourth downs with 4 or fewer yards to go. Any plays longer than that are considered passing downs. Success rate helps tell which teams are consistent in certain situations. Long plays can skew yardage totals, while success rates measure how consistent teams are in the situations listed.
Mike makes a few adjustments to the data. The first is redefining success on 2nd down to be 50 percent of the yards remaining for a first down. Mike also does not include kneel-downs to end halves, and considers yards gained on quarterback scrambles to be yards gained on passing plays.
Defensive Success Rate uses the same situations as above, but success is determined by preventing the offense from gaining the yardage needed to be considered successful.
For both offensive and defensive success rate, Mike has broken the data into an overall metric, running plays only, and passing plays only.
Check out the data through Week 15 here:
Follow us on Twitter @ITPylon.
Do these measures predict future wins better than traditional measures of success like points scored and (adjusted) yards per attempt?
(DVOA, which also uses success rates, does not seem to be very useful for predicting wins until its black box is tweaked to account for hindsight).
If these data don’t improve on win prediction, what are their purpose? Thanks.