2071 Defense Breakdown

[TheseBoots]

Here is a look at the defensive pressures for the most recent season. This is something I have been doing in the IHOF so I will copy my thinking and methodology from there. You can click the image for a big version.




After the Eagles were much praised for their 'no blitz' victory over the Kansas City Chiefs in the recent(ish) Superbowl, I thought it would be interesting to take a look at the effect of blitzing during the IHOF 2130 regular season. This thread shares the initial results.

My idea was to investigate which teams, like the Eagles, are good at generating pressure without blitzing, but the numbers tell a somewhat confusing story, as the distributions in blitz numbers are wildly different to the NFL itself. The other pressure stats which include amount of drop backs faced, and hurries, knockdowns and sacks are all within a comparable range with the odd exception, but the overall calculation of 'blitzless pressure' is somewhat confusing based on the method I have currently implemented, being heavily weighted towards low blitz teams - at least on the surface level evaluation. I'm not convinced that tells the whole story at all, as for most teams there does seem to be some correlation between blitz rate and pressure rate. It's just when trying to remove the blitz rate from the pressure rate the that numbers seem to break down a bit and favor the zero blitz teams. Somehow oranges are being compared to apples =)

So, if there are any suggestions how to handle this please do feel free to share them!

(Following posts contain the assumptions & methodology)

[TheseBoots]

Here's how the data was compiled:

Data was taken from in-game data weeks 1-17.

A BLITZ is defined as 5 or more players rushing the QB. Anything less is just 'pressure'. This means that 4-3 teams always blitz if a player blitzes the QB in the game log. 3-4 teams, however, are only counted as blitzing when they send two or more blitzers.

So far as I know there are two potential sources where a blitz can be detected. One is the game log, but parsing all the game logs to detect blitzing is possible but beyond the scope of this 'curiosity' project. Therefore I used the film room data to determine how many times a team blitzed in each game. The film room is not always 100% accurate, but in a small sample test I made comparing the film room to the game log this data did indeed seem to be pretty good. My test was 100% accurate, but I do expect that we could find some variance there if we parsed all the logs.

(If there is a better source of blitzing data somewhere in the CSVs or HTML please let me know!!)

In addition, without parsing the game logs I don't know of any way to determine whether the blitz was against a run play or a pass play, so the blitz rate calculation is based on the total number of opponent snaps faced. This is probably sub-optimal, but at least every team is evaluated the same way. Nevertheless, teams which face a lot of passing (those who play in a division with heavy pass teams, or those who lead a a lot) will be somewhat penalized by this. (Note: on further reflection it counts both as Alaska blitzed more times than they faced drop backs...)

Hurries are taken from individual statistics - one team - all players - defense - hurries. The limitation to this approach is that players who change teams mid-season will have the whole season's KDs shown on the page of the team they end the season playing for. I tried to take this into account by referencing the game-by-game statistics for players who changed teams, but I may have missed something along the way. Nevertheless, the numbers should be pretty reliable.

Sacks are taken from individual statistics - one team - all players - defense - sacks

QBKDs are taken from individual statistics - one team - all players - defense - KDs.

Pressures is therefore team hurries + team sacks + team QBKDs, and pressure rate is pressures/drop backs.

[TheseBoots]

Looking over some of the number trends...

In 2071 teams blitzed between 0 (STC, CSP, SAO) and 709 (APB) times across weeks 1-17. This is wildly variable compared to the NFL last season, where the distribution range (across 18 weeks) was 91 (Jacksonville) to 277 (Minnesota). Zero blitzes just doesn't exist. And nor does 709

Hurries ranged from 56 (LAS) to 127 (WVA). Real NFL by comparison was 27 (Tennessee) to 74 (Dallas).

QBKDs had a distribution of 5 (TEX) to 103 (LVS). The NFL distribution was 29 (Carolina) to 72 (Kansas City).

IFL saw teams SACKing the QB between 20 (TEX) and 53 (CAL) times. By comparison, the NFL saw a range from 28 (New England) to 63 (Denver).

(I included the NFL numbers just for fun - I didn't expect to see them accurately recreated in the IFL, but it does potentially limit the application of metrics which can be used on real life data sets when the distributions are too far outside the normal ranges - like blitz rate).

Finally, the pressure rate varies from Los Angeles at the low end with 16.97% and Las Vegas at the upper limit with 39,27% (almost went outside the conditional format rules!). Remember that pressure rate IS calculated just for drop backs. Teams with zero or very low blitzing DO tend to be in the bottom half for pressure rate, although there are a couple of notable exceptions to this, primarily State College.

We also have a few AI teams to look over in this league, and while none of those teams had particularly impressive pressure defenses (apart from maybe SLC, who are just generally good under the AI control) it does give some insight into what Rex thinks is a balanced game plan, landing somewhere between 20% and 30% blitz rate.

But feel free to dig into the numbers and share you own thoughts.

[TheseBoots]

Link to the data: https://docs.google.com/spreadsheets...it?usp=sharing

[Mellman]

Nice one Boots, thanks for sharing this - gotta look at that blitz rate, absolutely crazy lol

[CoyoteVice]

728 dropbacks... Wow

[joe the commissioner]

this is awesome!

[rode68]

Never said anything here, but amazing job, Boots!

Thanks for doing this!