FanPost

The Myth of the Penalty Problem

The Bills have a...gasp...PENALTY PROBLEM! Specifically; the problem the Bills have with penalties is that fans are still stuck in thinking that the Bills commit too many penalties. It's not true today, and definitely was not true last year.

Repeat after me: The Bills did NOT have a penalty problem last year

Since the close of last season, I've been harping on the fact that the Bills fared almost perfectly average across the season. I think the fact that Rex had so many other issues in managing the team has helped fuel the narrative that penalties continued to be an issue.

I've provided the numbers a bunch of times since the end of the season and my quest to end the narrative hasn't gotten as far as I'd like. Then it hit me. Numbers are boring. Colorful charts are where it's at!

Methodology

(Feel free to skip ahead to the charts)

One of the issues with convincing everyone that the Bills were pretty darn average with penalties is trying to encapsulate the entire league's data in an easily digestible way. I've given the raw numbers (league average vs. Bills average and whatnot). The problem there is it removes every other team from the narrative, and it's important to know how everyone else did. Also, the raw stats don't show context by failing to show "normal" all that precisely. And let's face it, with such obscure information as penalty rates, I can't kid myself into thinking that most people would know that 0.36 penalties per game more than perfect average is a fine place to be (where the Bills landed at).

The other missing piece is that I think deep down most people understand that "average" is usually a range rather than an exact number. Without knowing that range it's hard to conclude much of anything, and until now I didn't have that range to give you.

So I turned to Standard Deviations with some standard penalty metrics defined by rates per game. These metrics include: accepted count, yards assessed and total count (which adds declined and offset). This isn't perfect by any means, and I won't bore anyone with all the logistics (we can hash that out in the comments if desired). However, when looking at distributions, know that anything within one standard deviation from the precise average (above or below) is usually considered to be just as good as exactly average. The concept essentially turns "average" into a range rather than a single point which is just what the doctor ordered.

Accepted Penalties per Game

The first chart we have is what most people think of when they think penalties: how many? More specifically, how many were accepted and caused actual harm to the team?

A few conventions to note that I tried to stick to for all three charts:

The Bills are always bright red so they're easy to spot among peers. The red horizontal line is the precise average for the metric. The black lines above and below are one standard deviation away. This means the zone between the black lines is the real average result. Or put another way, the zone between the black lines might best be called "normal."

To the left we have the highest and right is lowest. For all metrics that means the best team is always the one on the far right. The teams with the worst "penalty problems" are huddled to the left.

Each graph has two other bars that are in "team colors." These teams represent the first ones going left and right that break out of the black line range (no longer in the range of "normal"). These are the teams that have some statistical significance for that metric.

Let's roll with chart one...

21462306_1399389573511264_3673700595760119745_n.0.jpg

A problem with conveying how average the Bills were last year jumps right out here. I've said a few times they were ranked 9th worst in penalties per game. That actually sounds bad because that puts them in the top third of "worst offenders." Please note though that teams ranked #6 - #27 are all within the expected range for penalty count. That means the vast majority of teams had penalty rates within the normal range.

In other words, very few teams in the league deviate that wildly from the pack. Five on the bad side and five on the good, and a few of those are just barely out of normal ranges.

Also note that the Bills were closer to the red line here than the black line. Which means they were closer to perfect average than statistically significant deviation. Put in terms of "impact," the Bills basically had five penalties more than perfect average across the entire season. Or 1 extra penalty every three games.

Repeat after me: The Bills did NOT have a penalty problem last year

The Raiders sure as **** did though. Fun thing in this chart, our division jumps out with Miami and the Jets being the cut offs to the left and right.

Yards Assessed per Game

21432878_1399392543510967_536560424696619581_n.0.jpg

Between count and yardage, I put more consideration in yards than most reviewers. Yardage tallies inherently create some weight by type of penalty. A team with high DPIs will shake out different than a False Start team for instance. As a result you see things shake out a little differently here than the last chart. Oakland still has the worst issue, but the Rams are now the cut off to the left, and Green Bay to the right. Note that in this measure, the Jets fall back to average rather than better than average, and Miami slinks further into the problem area.

The Bills move to #10 and closer to perfect average with this measure.

In terms of impact, the Bills had 2.76 yards per game higher than perfect average. If they had one less False Start every two games, they'd have have been just about perfect average.

Repeat after me: The Bills did NOT have a penalty problem last year

Total Count per Game

21462252_1399389566844598_320256875375139938_n.0.jpg

Some more shuffling as we get to the third and final chart. This is my favorite "standard" penalty metric. It differs from the first chart in that offset and declined penalties are included.

I put a lot of stock in penalty tendencies, as offset and declined are disasters that were averted only because something worse happened on the same play. So to me, this looks at the real rate of yellow flags. From a coaching perspective, this allows better trend analysis on problem areas.

Buffalo slips even closer to perfectly average in this category coming in at #11. I couldn't have planned this series any better with the results coming in as they did. Regarding impact, precise average here is 7.8 penalties per game. Buffalo? 7.94 per game. That's basically two and quarter penalties more than precise, perfect...BORING...average over an entire YEAR.

Repeat after me: The Bills did NOT have a penalty problem last year

Conclusion

Yes, Buffalo is on the "bad side" of the chart. However, the close clustering of teams means you'd be hard pressed to find meaningful differences between Buffalo and the 6th best team in the league. And that's independent of which metric you use above.

To illustrate what I mean, compared to the #6 team in each chart:

  • Buffalo had 1.16 more penalties assessed per game (compare to Green Bay)
  • Buffalo had 9.79 more yards assessed per game (compare to Atlanta)
  • Buffalo had 1.06 more total penalties per game (compare to Tampa Bay)
  • Coming back to trying to visualize this, Buffalo was about one holding call per game worse than the sixth BEST team in the league when it comes to penalties. Not a huge difference

Now, I do understand that such a small sample size creates some issues making perfect conclusions. For the record, using chart 1 as an example, I wouldn't actually say the Dolphins have a penalty problem. Nor would I say the Jets are super great with avoiding penalties. I'm merely saying those are the thresholds where the data suggests we're deviating more than "normal" away from precise average. However, using that chart, I would go on record and say that Oakland has some work to do, and the Giants have a couple things most teams could learn from. The overall idea is that most teams are actually performing quite comparably, with only a few outliers that merit any further consideration.

Some fun closing thoughts:

  • Denver's assessed count (chart 1) becomes interesting as they're statistically above average for total count (chart 3). This means they were luckier than most teams (disproportionate amount of offset and declined). The data suggests Denver shouldn't expect quite so many penalties to not count against them this year, which could mean a small increase in assessed is on the horizon
  • Oakland locks down their title with all three charts. They're actually more than TWO standard deviations above exact average in all three measures. For psychology purposes and human behavior, two away or more is usually clinical justification for a diagnosis and is seen not only as mathematically relevant like one away, but an indicator of a need for specialized support or treatment. Put in layman's terms; if one standard deviation means "relevant," two is usually the cutoff that means "holy ****"

One last time...

Repeat after me: The Bills did NOT have a penalty problem last year

Just another great fan opinion shared on the pages of BuffaloRumblings.com.