Elliot Chin
136 posts
Elliot Chin
@ElliotKChin
Harvard 25, neuro + stat. SF. sometimes well thought out sports analytics takes. Co-pres emeritus, @Harvard_Sports.
Katılım Ocak 2015
221 Takip Edilen48 Takipçiler
Chicago Bears WR Dante Pettis is now wearing number 81. Last worn by Lachlan Pitts. #DaBears
English
Every stop on Caitlin Clark's road to the record book mapped out 📍
English
what
The Athletic@TheAthletic
For fans, comparisons between Taylor Swift songs and football situations will help reveal the genius in her lyrics. For Swifties, they're an entry point in understanding the drama that is the NFL. Which of her songs best describes this season? More ⤵️ theathletic.com/5252243/2024/0…
English
Using 2 points for HFA, here are all games of Mahomes' career in which he would have been a dog on a neutral field, chronologically.
- A bunch of early games in his career
- games vs Ravens and Bills when these were considered special
- four of his last 5 playoff games (3-0)
English
@ericeager_ nfl is the only sport in which people haven't learned to listen to the nerds yet
English
@JoshuaBurke21 @PFF_Moo yes, absolutely.
it's not a bad question, it's been asked before! but a model will take this into account. twitter.com/StatsbyLopez/s…
Michael Lopez@StatsbyLopez
No one asked, so I'll ask it There's a fairly important part of the go for 2 down 8 discourse that is constantly ignored What is it? Explanation at end of my current train ride
English
@PFF_Moo Honest question I’m unsure of: do the models account for the change in opponent strategy when falling behind another score or getting a sudden change situation?
English
Since analytics can't account for everything (and the gut feeling of some people on here apparently can account for the residual better than Dan Campbell), pick your favorite thing some models don't account for:
🧵
English
@inpredict Absolutely. Theoretically you could game theory yourself into a run a small % of the time there, but practically no defense is over-emphasizing the pass there as much as they should, so a run is difficult to justify.
English
Agreed. And this will likely not be the topic of discussion tomorrow.
I can talk myself into a run play on 3rd down, AS LONG AS you have a plan to get off a play quickly in case you get stuffed.
Calling a TO there is just brutal.
John Hollinger@johnhollinger
The run play + TO on the 2 was the real mistake. Would be able to get ball back rn otherwise.
Pasadena, CA 🇺🇸 English
@benbbaldwin one of those rare instances where the benefit may be small but so is the uncertainty! twitter.com/ElliotKChin/st…
Elliot Chin@ElliotKChin
+1 - that Brill & Yurko paper that made some waves last year (link below) showed that our 4th down models are not actually as good as we think they are! Meanwhile, it should be pretty obvious that going for 2 is better, even if the benefit is marginal arxiv.org/pdf/2311.03490…
English
"Go for 2 down 8" is unlikely to affect result of the game bc being down 8 in 4th quarter is a bad place to be and you're probably going to lose.
BUT there's no argument against it being the right thing to do. It's impossible to make a model where it hurts you. So just do it.
English
Following up on this, these are good arguments and I'm convinced
Seth Walder@SethWalder
@benbbaldwin I strongly disagree! First: That is a small effect decision but one where we can be certain which side is the correct result. Don't blow that! Second: in the event the Bucs stop the Lions and score another TD -- the only way any of it matters -- the WP change is very high!
English
@inpredict @StatsbyLopez @skepticalsports agreed, this thread reminds me of that "win probability is like blackjack" debate from last week - ideally this should be baked into the model!
English
@StatsbyLopez Win probability based analysis would presumably account for this though, correct? Like @skepticalsports analysis for 538. Which showed an advantage for going for 2 down 8.
Pasadena, CA 🇺🇸 English
No one asked, so I'll ask it
There's a fairly important part of the go for 2 down 8 discourse that is constantly ignored
What is it?
Explanation at end of my current train ride
English
sports analytics twitter is going to be (justifiably) unbearable when any 50/50 calls go to the ravens this weekend
English
+1 - that Brill & Yurko paper that made some waves last year (link below) showed that our 4th down models are not actually as good as we think they are!
Meanwhile, it should be pretty obvious that going for 2 is better, even if the benefit is marginal
arxiv.org/pdf/2311.03490…
Seth Walder@SethWalder
@benbbaldwin I strongly disagree! First: That is a small effect decision but one where we can be certain which side is the correct result. Don't blow that! Second: in the event the Bucs stop the Lions and score another TD -- the only way any of it matters -- the WP change is very high!
English
Super interesting project!
Nick Lopez@nicklopez86
Last semester, I worked with @ElliotKChin on a project applying counterfactual regret minimization to poker and testing the results against a genuine user pool. We created our solver and game entirely from scratch. Read and play against our bot here: github.com/nrlopez03/CS13…
English
Students in my #sportsanalyics class are going to complete critiques of analysis they read online throughout the semester, initial list of potential sites so far (I won't have a comprehensive list but I'd greatly appreciate any recommendations)
Ron Yurko@Stat_Ron
Does anyone have a list of #sportsanalytics websites by sport? I don't mean in terms of data, but rather articles (e.g., Fangraphs and Baseball Prospectus for baseball, and so on)
English
@tejfbanalytics seems like this might just be a "teams that rush X number of times always win" stat rather than being causal, no?
English
not much of a coincidence that some of the best offenses in the league (SF, LA, MIA, HOU and DET) have the lowest average formation width this season
both linemen width and linemen standard deviation are tells of run/pass so when play callers keep formations more condensed it's less of a tell
Anthony Treash@atreash_
NFL's highest/lowest average formation width in 2023 (PFF data)
English
Elliot Chin retweetledi
Elliot Chin retweetledi
HSAC featured in @AdamMinter's Bloomberg opinion this morning. Check it out!
"'Teams that are top-heavy tend to have worse winning percentages,' explains @meron_ben , a recently graduated member of HSAC who contributed to the analysis."
Adam Minter@AdamMinter
Want to fix the Yankees? Send in the antitrust lawyers (or their statisticians). bloomberg.com/opinion/articl… by me for @opinion
English
Elliot Chin retweetledi
Want to fix the Yankees? Send in the antitrust lawyers (or their statisticians). bloomberg.com/opinion/articl… by me for @opinion
English