Pinned Tweet
Inspired by @ericeager_ article using nflfastR data I decided to look at which point differential for extra point vs go for 2 would be most impactful and the answer will surprise you: go for two up 6. (1/n)
The case for going for 2 down 7 points, by @ericeager_ sumersports.com/the-zone/why-y…
1
1
28
29,822
x.com/903124s/status…
Instead of man in box I use a finer grain approach of using tracking data to estimate yards gain by pre-snap position and given today play-action pass ratio it’s basically indifferent to pass or run more
Instead of man in box I use a finer grain approach of using tracking data to estimate yards gain by pre-snap position and given today play-action pass ratio it’s basically indifferent to pass or run more
Why hasn't "pass go brr" happened?
It's probably a game theory thing... NFL offenses and defenses already sit close to a Nash equilibrium.
Equilibrium: about 55% pass. The NFL: 54%.
Data: @SumerSports
Assumptions and methodology (in video form) ⬇️
2
0
19
2,344
My contrarian view is if slope of (pass efficiency/pass%) and (run efficiency/run%) are similar then any arbitrary pass percentage will yield similar result regardless of how passing is more efficiency than rushing
I think the fact it "stares us in the face" is why I am less in the camp "it's obvious we should be passing MUCH more".
It doesn't even take understanding EPA... Every coach in the league knows yards per play is much less in the run game.
I think someone in analytics should good faith steelman trying to answer the question "why do we run so much" and see what we find
2
2
23
5,573
Using @SkillCorner open ACB data we can extend it to data without labeling e.g. recently released NBA subset of tracking data (compiled by @SKPearlman)
1
4
49
3,990
For recent baseball vs cricket debate here is a very good piece by @himganj153 about evaluating cricket performance similar to WAR framework that baseball people should be somewhat familiar
hganjoo.github.io/t20basics/
hganjoo.github.io/t20basics/
1
1
10
1,593
People says momentum i.e. previous scoring don't matter in football and I beg to differ. After account for score state/rubberband effect, teams score 14+ points previously perform 1 point worse than expected
3
3
31
10,069
As pitch/FX era only has pitch speed and break only, using different pitches in player’s arsenal to derive spin rate(as it’s correlated) and vertical/horizontal break contribution to calculate spin axis, here is season normalized stuff of some notable pitchers since 2008
2
0
15
2,323
Polls margin of error don't correlated to actual error of result unless it has very small sample size @DrewSav @ADincgor @RealCarlAllen
1
2
3
632
Very rough estimate show about 0.03 run per 9 increase for each percentage point increase of league wise pitch type usage (after reclassify to today definition) @tangotiger @pitchprofiler
1
1
9
1,473
Susan Collins is actually the most "disloyal" Republican senator of all time (in accumulation) and still top 10 on average regardless of affiliation.
Here next I'd explain why it's rather not trivial to put politician voting record in context 1/n 🧵
Here next I'd explain why it's rather not trivial to put politician voting record in context 1/n 🧵
Susan Collins voted with Donald Trump 94.6% of the time in 2025, according to VoteHub's Trump Scores.
1
0
1
1,199
