2076: A season in (too many) numbers - part I

[TheseBoots]

With the regular season in the can I can run the results through a few tools I have been slowly working on in my lab. Let's have a different look at Turnovers compared to what is available in the game to get us started.



Turnover luck seeks to evaluate whether a team had relatively good or bad fortune compared to the expected, and can give quite a different picture to the data available in game. If you look at the Team Statistics -> Scoring Turnovers you will see that STC had a -17 turnover differential this year, second worst in the league. You might think they must have been unlucky, but they weren't! In fact, they were incredibly lucky to get -17!! Almost 50% luckier than the second place team in terms of turnover luck, KCY! The worst TO differential this year in terms of raw numbers was SAO on -23, which is bad, but they were genuinely unlucky by a wide margin. More than twice as unlucky as CAL, the second most unlucky team.

[TheseBoots]

So, what about defense? Well, let's have a look at which teams were just good at stopping their opponents without conceding a score.



Stop Rate totals the number of times a team's defense caused their opponent to punt, or forced any kind of a turnover, and expresses it as a percentage of the number of drives faced. London, Columbia, Kansas City, West Virginia and Alaska were all very good by this metric, and across the Top 12 teams, 8 of them made the play offs. It's probably helpful to have a lot of stops, but not by itself a particularly strong indicator of how the season will pan out

[TheseBoots]

Which defenses were the most horrible to play against? We can get a sense of it by calculating the 'Havoc Rate' of the defense by summing sacks, forced fumbles, interceptions, PBUs and TFLs and again, expressing that as a percentage of the defensive snaps played. Note: Havoc Rate is a metric you can find calculated for the NFL and college game. QB pressures are not included generally, and anything over 20% would be considered very good/exceptional out in the real world. FOF8's engine generates realistic numbers.



And who is creating all this mayhem? I'm glad you asked!



This is the top 20 havoc-generators - raw numbers only, so not like PRPct. As you can see, it's a pretty solid mix of the defensive positions, although no DT made it this year and they probably have the hardest time generating the stats they need to get onto the leaderboard. For the record, Hayden Mahoney (TEX) 20 and Brayden Leggett (PIT) 18 were the highest rated DTs.

I'll be posting some more new analytics stuff over the next couple of days. If there is anything you are interested about feel free to ask, and as you can probably tell, I'm more of a data guy than a visual designer so if you have any suggestions for improving the visualizations they would be very welcome!

[TheseBoots]

You can click the embedded images for big versions. And for anyone (anyone??) who listens to the IFL Chat shows might know, I Often wonder what LBs do for you. A bit of everything it turns out. I probably underrate the value of a good one. I still don't like paying them.