How a Poker Pro Would Coach in the NFL

[@Gambler-uc4wj]

Loved the video! Lions fan and poker player here, and getting sick of people ragging on Dan Campbell for going for it so much on 4th down in the NFCC vs SF. He coaches to win the game and was the best coach in the league for in game decisions this season.

[@andrewmiller6309]

I remember watching former NFL player John Urchsel's HBO segment where he breaks down when a team should go for 2 when down 14. He is now a math professor at MIT. Funny that it takes a math PhD or a veteran poker player to point out the simple math mistakes happening in the NFL

[@tfwjr413]

Another thing to factor in is sample size. There are only 17 regular season games, and even fewer chances when this scenario comes up. This is opposed the the much larger number of chances you get when playing poker. In the long run, this strategy works. However, as you pointed out, if the coach gets fired, there is no long run.

[@Aaron-xs5bv]

Great point

[@DStern20]

In the second scenario, something that would be interesting to factor in but difficult to quantify is the information edge the defense has in knowing that you can’t go for a field goal and the impact on win probability

[@ryankirby383]

Definitely significant. They can play prevent a lot harder if you can’t kick, especially in very late game situations

[@NoKetchup32]

Not a huge football fan but love sports theory and analytics. Please keep it coming, Phil!!

[@PhilGalfond]

Can do, thank you!

[@BOnYTB]

Shout out to Dan Campbell, DET Lions HC, he’s probably the most aggressive HC rn in terms of going for it and fans in DET love it. It takes extra cajones when you’re not some super long tenured successful HC, he’s pretty new and still doesn’t care. He is going to coach his way and he has the full backing of ownership/F/the fans, full buy in from the players, etc. It’s awesome to see.

[@fluffysheap]

This is a good point. Maybe it's not the fans coaches are worried about but the sports writers. But as local news dies off sports writers will go with it and this will open up the possibilities. Everything is connected 🙃

[@kevinjohnson4498]

I think one major thing to consider for 2pt conversions is how does success rate scale with attempts. Generally teams will have a handful of 2pt plays in their gameplan and break them out for the rare cases they need the 2pts. If a defense is seeing the play/formation for the first time, there will be a higher % of success, and then the more they see that look they will be better prepared. So instead of 48% chance of converting each 2pter, it may be more like 48% for the first, 45% for the 2nd, and 42% for the 3rd. But when I played Madden I went for two 100% of the time and it definitely resulted in me winning close games more than I lost them.

[@PhilGalfond]

Yeah, this is a legitimate counterargument, but I think it should be outweighed by the math (which I did with generous assumptions). If 2pt conversions become a focus for your team in play and practice, you'll have more plays and formations ready to go.

[@loganpoole3621]

You say that ... but check out Ole Miss for example in college football. A team known to be using these analytics in their game, focusing instead on when to punt versus attempt a 4th down play. And their success rate was substantially lower than you might otherwise expect. Further complicating things are the psychological impacts on your offense. If you fail the two point conversion down by 8 leaving you down by 8, that means your offense will have failed their last attempt. You address this some in terms of the loss aversion conversation but this suggests the only individual vulnerable to this fallacy is the coach but its also the players. It also means your offense not only has to still score again, something they otherwise will have just had as their last attempt (and a success), but that they know going into it that if they DO score again they have to ALSO complete the task they just failed at as well. The extra point failing is substantially less likely but ALSO results in the last failed play not being one committed by your offense in the unlikely event the kick is indeed missed. This matters and affects what many coaches call the "Momentum" of the game. Analytics and "moneyball" approaches have value. But it's not so simple as it might seem.

[@Glitch47278]

@@loganpoole3621this is correct, if players were all on the same page and agreed with the strategy you’d be ok, but the pressure that accumulates on the players which impacts performance is not part of the math. Going for 2 and missing it, and losing 6-7 early in a game has an impact on your team performance as a whole.

[@Aaron-xs5bv]

Always go for it in madden lol

[@Brandon-youtube]

Was thinking the same thing. It might even scale from recent games (exposing your scheme on film). As Phil points out, probably doesn't change the binary "should I or shouldn't I" piece but does probably make it closer.

[@deano4138]

Fairly neat video, for using stats in an isolated vacuum with no context. It's the equivalent of a football coach telling a poker player that "two pair on the river is always a snap call because two pair is a winning hand 59% of the time" Immediately poker players would point out relevant context missing. How many BBs, what's the board, who's the opponent? Flip it it's the same. How does our o line match up against their d line? How has our team done in short yardage so far this game? What about the (2-3%) chance that we turn it over on the 2pt play and the defense takes it back for a 2pt score of their own? Also while the conversation rate is about 45-48%, most coaches believe this is inflated due to things like momentum and pressure (teams usually only go for 2 when trying to make a comeback after scoring, they are feeling the momentum and the defense has anti-momentum seeing their lead slip away.) In neutral, non high leverage situations like practice the conversation rate tends to lean closer to 35-40%. 1st-4th and goals from the 2 yard line also hover in that 30-40% range. That's enough football nerding for the day. Sources: played and coach low level college ball

[@kirkbastick8558]

Loved this, Phil, and would love a series of poker thinking applied to non-poker activities. I often find myself saying I’ve just flopped a boat against aces when I luck out in something. Perhaps a breakdown of optimal golf strategy with driver vs iron off the tee, or laying up vs going for the green in 2 shots on par 5s. Similarly, is it ever optimal to go for an attacking second serve in tennis vs just a conservative 2nd serve for the “make the opponent win the point” mindset. Other ideas: going for 3 point vs 2 point plays in the NBA, when it’s optimal to start a fight in the NHL to up the vol by getting two people sent off when you’re trailing deep in the game, but any application would be great. Consistently great content. I can’t believe there are so many poker players out there missing out on your channel!

[@WriteWordsMakeMagic]

I think that the second scenario is a bit more difficult to model because if you're up 2 vs when you're up 4, you might play differently knowing what your opponent needs to do to win. So the numbers might shift a bit more.

[@shakabraddah7898]

When you were talking about fans only remembering the bad things that happen - the losses more than the wins, couldn't help think of the poker parallel you discuss in other vids. How poker players seem to only remember and fret over the bad beats we take instead of all the times we got lucky. Couldn't help but to chuckle there. So true!

[@bajajoepoker]

Really enjoy your videos, Phil. This one was unexpected and an excited to be able to share with non-poker friends! Would love to see you dive deep on hero calls. Found a few with Patrik Antonius that were enjoyable. Probably the most interesting would be right and wrong hero calls (you like) vs you.

[@timmychuk]

I used to work in insurance and did a calculation of % chance of getting caught without insurance and the penalty and it was slightly +ev to not pay for insurance. I was factoring in police per capita and all kinds of fun stuff way more fun than the job itself.

[@imrahil_327]

This was super fun and interesting, looking forward to more! Thanks Phil!

[@Brandon-youtube]

The 'keeping their job' point is applicable in a lot more than areas than poker and football, glad you called it out. People are much less likely to take "risks", even when they are hardly a risk, so that the fault couldn't possibly be theirs. In reality, a poor outcome after taking the safe route is their fault for the reasons you have explained in the video!

[@Funktaro5]

Exactly ... using the Lions vs SF as an example, they were about 50% to convert each of those 4th and short attempts and with their kicker at 48 yards about 50% to convert each field goal attempt. The Lions went for it both times and failed to convert, so the coach has now gotten all the heat for that. If they kicked both times they'd also miss both at the same frequency (about 25%) but then it could've all been blamed on the kicker instead. Some coaches actually probably factor who would get the blame if they fail into their decision making. But Dan Campbell absolutely does not. Which really is the way it should be.

[@naswiipp]

Thoroughly enjoy your takes on poker and how you relate them to other aspects of life.

[@OwenPenner]

Hey Phil, I really enjoyed your insights into NFL coaching decisions through a poker lens. The math behind it is solid. It got me thinking - if you were coaching an NFL team, how would you incorporate poker logic into draft strategy? Especially considering those critical two-point conversion scenarios.

[@hugoengstrom2032]

Awesome video, liked How you made it so easy to understand while keeping it interesting.

[@lapp3r30shockeycardflips]

Awesome video Phil!!! Love it!!

[@wildcatchamp7354]

These videos are superb. Literally every single one of them. No idea why you have relatively few subscribers, there should be way more!

[@PhilGalfond]

Thank you so much! We'll get there :)

[@skoldpadda9]

Chip Kelly had a similar style of going for 2 early when he coached at Oregon and UCLA.

[@ldmb1966]

As a titans fan that dolphins game was the highlight of the season, thanks for spotlighting that haha. The mentality of "we're not even expected to be competitive, so f*** it" plays into the smart and aggressive playcalling sometimes i think, but you're right that too many coaches are conservative and play it safe.

[@PhilGalfond]

There are a lot of interesting decisions that involve increasing the variance when you're the underdog!

[@Aaron-xs5bv]

Loved this video. Really cool seeing poker principles applied to other areas of life. We need a breakdown of Dan Campbell’s calls after that Lions game now!

[@AdamGoodmanlv]

Great work Phil! How about applying poker concept to financial planning and/or investment advice for individuals?

[@FadingDanYates]

Interesting video. The 1st example is definitely being utilized more often these days. But the second one not so much and it makes sense. I’d love to see some kind of analytical analysis of onside kicks vs regular kick offs.

[@D_Rahs]

Onside kicks no longer viable. They changed a rule and made them much harder. A video on fake punts though may be interesting

[@noex100]

​​@DRAHS What rule got changed?

[@D_Rahs]

@@noex100 New kickoff rules (2018) mandate where players on the kickoff team can line up, and prevent kicking team players from getting a running start.

[@themikemac21]

Great Job Phil!

[@marctetreault3877]

Very cool video Phil, I love the concept.

[@PhilGalfond]

Thank you 😊

[@lautben]

I think the only variable you might not have considered is that these statistics are obtained under a metagame where the players mindset is accustomed to specific scenarios where they must go for two. If you make a player go for two every time now, half of the time he will miss and that will result in a downswing way more often, affecting his "going for two" skills and maybe even lowering that 50% to 45-40-35%. Thanks for the content Phil, awesome video!

[@Spreek2]

great vid, would love to see other sports leaks broken down

[@mjrnj]

This is a great open job application/audition..... Wishing you the best of luck in your new career Phil!

[@Crazy_Nancy]

Good video, Phil! I coached football for a living before I started playing poker full time. You make a great point about coaches not wanting to make unorthodox plays because of pressure from fans, players, and superiors. I had lots of ideas for how to run an offense that was totally outside of the box and would be a game-planning nightmare for opposing teams. But it’s a huge risk, because any downside variance will be met with calls for your job, boos from the stands, and a loss of morale from the players in the locker room. Definitely tricky to weigh all those things into your EV equation when deciding to go for 2, run a flea flicker style offense, or onside kick way more often than the average Joe.

[@jackripper8337]

Super ironic that this game out 2 days ago and the Lions played their 4th downs perfectly and are getting lit up for not kicking the field goals.

[@big221-30]

8% edge is HUGE

[@matthewgittins3338]

Probably out of the realms of easy math and your knowledge base, but there is some great work about where to place a penalty in soccer amd at what velocity. Whether the keeper should dove or stay, whether takers should vary placement location or keep to 1 spot....

[@nickmullen402]

LOVE IT! Especially the part about loss aversion. I've been saying forever that teams should focus more on the two point conversion and going for it on 4th and goal inside the 5 yard line. I think maybe there are some football specific subtleties that you kinda brushed over, for example the psychological edge/disadvantage that comes from making/missing the two point conversion in these situations, but I think these also favor going for two.

[@PhilGalfond]

Thank you! My belief is that the subtleties are negligible and cut both ways, which you allude to by saying that they can favor going for two.

[@BOnYTB]

You’ll love Dan Campbell (DET Lions HC) he’s prolly the most aggressive I’ve ever seen in terms of going for it.

[@nickmullen402]

@@PhilGalfond I think the first of two Dan Campbell 4th down calls illustrates the point. Up 24-10 at that point in the game, taking back a three possession lead was the standard play for a good reason. Even if the “points EV” was higher by going for it in that spot (and I’m not sure I buy that but we’ll assume Campbell knows the analytics better than we do), a very likely 3 points was far better than an uncertain 7 (with a possible 0) in that situation. The “momentum” or psychological edge really showed its worth in that game. I think it shows how “point EV” and “win probability EV” are not always the same

[@robw9963]

I always go for 2 in Madden from the start. It starts off the game with a big advantage if you make it, puts your opponent in a bad spot and it's more fun. Poker players strategy for running a business / investestments would make a good video. Keep it up.

[@Wooden_iguana14]

Ironic that the example you used was Vrabel when he did end up getting fired at the end of the season 😂

[@PhilGalfond]

The problem with recording many weeks in advance 🙈

[@seanaujila7014]

@@PhilGalfondit kind of proves your point as to why coaches don’t do it even more.

[@BOnYTB]

Vrabel is a really good HC too which is what’s crazy. The guy was coaching Will Levis lol. Pretty shocked he hasn’t been hired, maybe he wants a year off and/ or is waiting for an ideal job. Bc he’s a good HC.

[@Wooden_iguana14]

@@BOnYTB agreed. He shouldn’t have been fired.

[@jjjrjjjr1]

And Adams Strunk family want the taxpayers to fork over 2.3 billion for a new stadium. I know the 2.3 is in bonds and I don't pretend to understand high finance but still...

[@jackklompus3099]

Great video, as always, Phil!

[@cpasa798]

Raise your hand if you think of “Moneyball” when watching this video 🙋🏻‍♂️

[@lunchbox6576]

Me as well

[@berdyderg900]

Tfw you have to view reality through the lens of media you've consumed

[@cpasa798]

@@berdyderg900 simpsons

[@PkrDealer86]

Not me

[@brooksbutler8256]

Very neat video concept

[@johnd5619]

Super cool video Phil! Would love to see more

[@PhilGalfond]

Thank you!!

[@Joe-Bourbon]

I’d like to see a breakdown of always doing onside kicks. What does the recover percentage need to be for +EV? I’m surprised we don’t see more kickers training on this.

[@donbdoe5798]

Was talking to a friend about this the other day and predicted that in 10-15 years football teams will never go for the extra point unless it’s to take the lead or win the game

[@PhilGalfond]

I think there will always be some value in the 1pc, but if it shifts too much towards going for two, they might adjust the rules to make it harder.

[@BOnYTB]

@@PhilGalfondYep. Extra points became so close to 100% (it was from the 2 yard line, so only a 19-20 yard FG) they moved it back 13 yards in 2014 all the way to the 15. So what was once a 19-20yd FG (I say 19-20 bc while the end zone is 10yds deep, the holder can sometimes not be EXACTLY 8 yards behind the line of scrimmage, human error is a thing, and obviously a couple inches isn’t a huge difference…I’m too juvenile to not re-read this and chuckle) became a roughly 33yd FG. The ball for 2P conversions is spotted at the 2 yard line. If it ever became a thing in the NFL where teams started to crush it at 2PCs and go for it almost every time, bc they’re converting at like 65%+, they’d surely move it back from the 2 to the 5 or something to re-balance it. Once something becomes a no brainer decision and you’re converting something at such a high rate, the NFL tends to balance it out. One example would be the “Tush Push,” although there are other circumstances as to why it’ll get banned. If people watch the NFL, they’d know how big of a thing the “Tush Push” became this year with Philly. They were converting 3rd & 1, 4th & 1, etc. at an OBSCENE rate. They have a very big and strong mobile QB, and have/had (Jason Kelce retired) amazing key players on the offensive line. The most pivotal on the OL was probably their Center, Kelce, one of the greatest ever at his position and someone who’ll be a 1st ballot hall of famer. Super strong. It also helps their QB, Jalen Hurts, is about 6’1 225 and literally can squat 600+ pounds. There’s also a video from a couple years ago of him deadlifting 620lbs. Dude is a beast. As a SO in HIGH SCHOOL he was squatting 500. He was a regional finalist in his weight class for squatting in HS. His legs are basically tree trunks, and bc of how strong his lower body is, he has a huge advantage in this play because he can keep his body moving even against crazy amounts of force. The NFL didn’t want basically a cheat code to get 1-2 yards in the game, especially one that wasn’t very appealing to fans, and was also probably pretty dangerous considering you’re having multiple 250-300+ pound linemen push the QB violently in to other 250-300+ pound linemen. So they said they’d ban it after this season. PHI perfected it essentially, as far as I know they never had any major injuries running it but I could be wrong obviously. I do know other teams both couldn’t replicate the success of it, and got a lot of players injured trying. In week 4 this season, the Giants had TWO players sustain injuries during a failed tush push. Other teams admitted Philly ran it the best, and while some didn’t complain, way too many coaches, owners, front office people, and even fans complained. To change a rule they need 75% vote of the owners, (24/32) and obviously the PHI owner would’ve vote against it, so really 24/31. They pleaded w the NFL because no team could do it even CLOSE to as well, but obviously the other owners are going to eliminate such a massive advantage. It honestly makes sense to ban it even though I agree it’s a play they figured out and perfected so it’s unfortunate. The thing is, you’re not allowed to PULL a player, so there’s no reason you should be allowed to PUSH a player like that. So as much as I like to see people innovate, I feel like they kinda had to. In 2022, they had a 93.5% success rate running this play. Everyone knows exactly what play they’re running too, because the formation is a dead giveaway, but even with the defense having perfect information…they STILL converted it 93.5%. That’s absolutely ridiculous. Now they never really ran it unless it was a yard or less to gain, maybe 1.5-2 yards occasionally…but 1yd was the most common by far. The QB sneak has always been a thing, where it’s a regular non shotgun formation, and the QB just immediately dives forward or solely drives himself forward as soon as he gets the snap. It’s a successful play, probably underutilized on plays with 1yd or less to go for a first down or TD…but teams may be worried about injuring their QB. Good starting QBs are making 10-15% of the salary cap, which is a crazy amount when you consider there are 53 players to pay. So even if the sneak was 5% better than just running or passing it, you’d probably only do it in really important situations. The “Tush Push” or “Brotherly Shove” is next level though. They lineup with multiple players around/behind the QB, and as soon as it’s snapped, they have their biggest TE, and WR both push the QB as hard as possible. They’re also of course lined up with all OL beside the people pushing him so there’s max blockers for him. It’s not like it’s literally unstoppable, it has been stopped, but compared to any other play to gain a yard, it’s so much better it’s insane. As I said it was 93.5% last year…you’d think the % would go down drastically giving defenses an entire year to study it, have a full training camp and then practices during the season, etc. to stop it. Nope. 37/40 in 2023, 92.5%. Jalen Hurts scored 11 1yd rushing TDs this season. But ya even if the tush push wasn’t kind of exploiting a rule that should’ve been on the books but wasn’t, if half the league started doing it and converting 1 yard, 1.5 yard plays at even like 75-80%, let alone what Philly was doing in the low mid 90s, they would’ve changed something to stop it. Nobody was doing anything wrong with the extra points, they still changed it. In an article published in Dec ‘21, they put X% on a graph. In 1974, the NFL moved the goal posts from the front of the end zone to the back, a 10 yard move. A lot of that had to do w safety concerns, guys running in to them. But obv % suffered moving it bak. In ‘73, XP% was up to 97.7%. After the move, it plummeted to 92%. Ot kept doing down a bit, until 1978 it was 93.1%, back down to 91.3% in 1979, 94.9% in 1980. Then it was a steady climb from there…within 15 years it was 97%, and by the mid 90s it was near 99%. In the 2010s, it was over 99%. That’s when they moved it in 2015 back 13 yards. From ‘15 to ‘21, it’s gone from 99.3% to 94.2%. But it’s not correcting. It’s been between 93% and 94.6% in the 8 seasons after the change. A decent explanation for why, is that 2PC attempts have gone up from 11% in ‘14 to 25% in 2022. Since 2015, XP is 93.7%, and 2PC is 48.1%. League averages. So just by pure EV, on average, 0.9371 points < 0.962 points. This was written in January ‘23 so it doesn’t have this seasons data, but that might be a good thing with the tush push affecting things considering it won’t be around anymore. You really need a HC that is very well versed in the analytics, but also knows his team well, and is willing to alter his decision vs the analytics if it’s close for the benefit of his team. Like if you’re a big underdog and would be one in OT, and you suffered numerous injuries in the game, entire team is gassed, etc…you should be much more aggressive going FTW. Like let’s say you score a TD to go down 1 with basically 0s left, but you’re on the road and would be huge dogs in OT, you should probably go for 2 even if there’s a penalty that moves you back to the 6-7 yard line. If your defense really had suffered numerous injuries to key players, ESPECIALLY if you don’t have a good K, you can defend going for it from the 7. Then there’s being more conservative than analytics tell you. If your defense is destroying their offense and they can’t move the ball…and you are up like6 points on 4th & 8 where your decision is to kick a 43 yard FG, or punt it with like 1:30 left with the opponent having 2 timeouts, analytics probably say to kick the FG especially bc it’d end the game. But if your defense is crushing, their offense hasn’t been able to move the ball, AND your FG kicker is maybe banged up a little, your Punter is fantastic and is really good at pinning the opponent inside their own 10, that’s when smart coaches know to deviate. I’ve also seen charts that are straight up absurd. One from ESPN says to literally go for it on 4th & 1 from your own 1. 4th & 3 from your own 25. They don’t even mention situation, this is a chart. So it’s essentially saying in a 0-0 game on the first possession of the game, to go for it on 4th & 1 from your own 1. I’m not sure what smart person would do that, but it’s legitimately absurd. I also laugh seeing messages online saying how lazy the analytics are on TV. As if they cannot put in to account time, score, that teams average conversion rate, etc. NFL teams have their own models that take in to account their personnel, what % they are in that situation, time, score, opponent stats, everything you can think of. For sure some of the robots that spit out go for it or don’t are wrong occasionally, personnel matters, if a team is atrocious in that spot, etc. it all matters. That’s why having a HC that has a strong analytics department and understands them, but also understands his own team…it’s so massive. Long post, sorry about that! #GoLions

[@KaCee-v6o]

@@BOnYTBi can’t believe u just wrote all that on a KZbin comment , u alright over there?

[@jackklompus3099]

@@BOnYTBwow great post…ty for writing all of that! Learned a lot

[@gtheofanopoulos]

Phil delivered as always. Let me suggest an analysis of a concept in basketball: European coaches always foul on the last seconds of a game, when their team is up by 3 pts, to force two free throws for the opponent. That is to cap the opponent's attack at 2 pts and force them to foul immediately after. NBA coaches rarely use this strategy. I believe the Europeans have the better understanding of math in this situation, but the result is not good for the viewer, as the match goes to a barrage of free throws at the end of such games.

[@PhilGalfond]

Awesome, thank you! I'm not a big basketball fan. I'm curious to look into this.

[@qsdailydose8970]

Great analogy and I just put it back to poker great stuff

[@tomtsu5923]

Great video Phil. As a pro, I've been considering results oriented thinking and football (game of variance) for a long time. You should do the whole break down (before I do, could write a book lol).

[@aaronbogage9560]

I totally agree with the mindset of going for 2 in these situations, and @Phil I have seen your comments regarding how momentum cuts both ways, which I also agree with. But there is more context to consider, than just the score, the time remaining, and the momentum. Does the offense have great success in the low red zone? Is the offense entirely healthy, or maybe did their QB or star WR get injured during the game? Is the defense missing a key interior defender, which might make it easier for the offense to score via rush attempt? Is the defense the best in the league in terms of stopping short yardage rushes? IMO all of that context should be utilized primarily when deciding on what the play call is for the 2pt conversion, rather than deciding whether to go for 2 or not.

[@bpbenda]

Excellent video, Phil! 🏈

[@PhilGalfond]

Thanks!

[@RaidenRadio]

I often think about this in regards to 3rd and long situations and teams not passing deep to try and pick up the first down. If you complete it the drive continues, if you don't you'll be punting, and if it gets picked you were punting anyway. So it's frustrating to see very safe, short, low % plays and I wonder how much INT stats figure into this.

[@alexvo6873]

Would be nice to hear Phil analizing 4th downs.

[@construirelequebec]

Awesome video Phil. I used to play poker, not anymore, but I still enjoy your content. This is 100% in my sweet spot ! Love the idea. We want more of these !!!

[@sambattista9697]

Loved this! Would like more like it

[@sundarpappu2601]

You’re the man Phil

[@busomite]

I think the keenest insight here is that the coaches goals and the team goals aren’t exactly aligned, they are closely aligned, but there’s some daylight there.

[@justinkushi]

In the down 10 regular example why are we limiting them to only kicking a fg and going to ot and not acknowledging the times they score a td and win? If we give them the same 25% td rate their win % goes up to 62.5%.

[@DejaquezernDenaster]

Awesome video!

[@nealstradamus]

Phil you must also factor in the defense feeling disrespected and turning up their play a notch when faced with a 2pt attempt early in the game. It can feel like a slap in the face. The momentum can turn quickly.

[@SimonSchatten]

awesome idea!

[@jberczi6]

Would you ever consider doing a follow-up to this with a true ground-up game theory approach?

[@Syntax991]

when to force buy in counter strike could be an idea

[@placeholdermcgavinport5343]

I haven't watched this yet but immediately I have to say this is a genius premise for a video

[@OhCanadaMoose]

Im familiar with the go for 2 twice when down 14 argument. But what I dont understand is you still need 2 TDs either way, so what if you just go for the extra point on the first one and go for the 2 on the second one instead of 2 on the first one?

[@bobsburgers8885]

Follow-up video suggestion: Analyze the decision to take the ball or not in the super bowl OT. Going first gives you the chance at the sudden death 3rd posession, but the other team can negate this by going for 2 if they score a TD after you score. Or maybe you should take it first and go for 2 if you score a TD? This is clearly the correct play if Mahomes will score at every opportunity (including a 2 point try); how high does his success rate need to be in order to do this? This is an interesting game theory situation.

[@mksteele99]

This 100000000000000000000000000000%. I'm adamant that receiving in OT was a huge mistake by the 49ers but my 49ers friends don't think it matters. @phil galfond

[@jackripper8337]

I have had this argument with people for like 10yrs now. 95% of people just don’t get it. And I had a similar argument where game ends on final play for TD down by 1 do you kick the extra point or go for 2? Lions did it this year against Dallas…. It was a mistake. They only make 2pts 48% and have a 50% chance of winning in OT. Honestly all these small +EV situations in life, it drives me insane when people don’t take the best possible situation. Great video bro!

[@BOnYTB]

Doing some NBA stuff would be cool. Some teams manage to mess up super basic stuff, like the 2 for 1. Or when to foul when you’re up in a late game situation can be fascinating. If you’re up 3, and there’s 8s left, opponent inbounding from half court…if you have a foul to give you HAVE to foul unless they just instantly rise up in to their shot. Ideally you can run off 2-3 seconds there. Say 5-6s left, up 3, opp inbounding from half court. If the opponent is dumb enough to go for a 2, let them. Literally play matador defense and let them have the free layup or dunk. Bc now they’re going to be down 1 with say 3-4s left, and now they have to foul you. Especially if they don’t have another TO left it’s super obvious…if you make both they’re down 3 again with 3-4s left and no TO. They have to go full length of the court and hoist up a 3. Even if you make the 1st and miss the 2nd FT…you’re still up 2, and now they have to get the rebound (you get it, it’s over) and then attempt a half court shot or longer. I’ve seen teams foul up 2 with like 10s left which is very interesting. If you can foul a really bad FT shooter (say 60%) it may be very reasonable. They have to make two FTs to tie (.6 * .6) so they’ll have a 36% chance to tie it. And even if they tie, you call a timeout, and you get to draw up a play with 10s left where you can take the final shot. Worst case scenario is OT really…and especially if you have a fantastic half court offense with lethal iso scorers, you’re going to score 1+ point in that 10s possession a good %. If all fails, you go to OT where you’ll be a favorite (wouldn’t do this unless I have the better team) so while I’ve legit only seen this done once (they missed an FT, so they pulled it off successfully) and was a bit baffled at first, in specific scenarios I think it can be a great idea.

[@Bonzi94]

I ALWAYS think this! legend

[@MauricioDongo-x7t]

I could be wrong but wouldn’t that 48% success rate get lower if more teams go for 2 more often? Larger sample size and defense preparing more for 2pt conversions.

[@checkrazor76]

Some great insight here but also missing out on some negatives. First kickers field goal percentage is way up since the inception of the longer extra point as it now works as a practice kick of sorts and and not attempting the extra points could effectively keep kickers cold. Second as teams attempt more and more two pint conversions the success rate goes down. Offenses innovate and defenses only take a season or two to adapt and stop offensive schemes. (Falcons high 5 and Dolphins wildcat come to mind ). So while this might in theory work now it wouldn’t be a lasting strategy

[@firstiesify]

Super cool! Loved that breakdown. Could you do chess next if you are familiar with the game. Both have vastly branching decision trees and dense game theory. Love your content man. Thanks ❤

[@glennyu6855]

isn’t the 45% chance of 2pt conversion is conditional on the other team’s prior on the likelihood you will go for the 2pt conversion? if you are the type of coach that goes for it in this spot every time, 2pt conversion might be more like 40%?

[@twistedspine7300]

great fun video. i have a couple of thoughts. first, when a team who wins the coinflip in OT receives the ball first and scores a TD the game ends immediately. you should consider making a video on OT specifically on how often the team who wins the coinflip should skip the FG attempt and play for the TD because of the added value of the other team never getting a possession. second, going for 2 more often as head coaches figure this out will create a lot more 1 or 2 point wins and losses in games which historically is a lot less common and would increase the value of a +3 point dog dramatically and punish -3 point favorites. could this potentially shift the meta for sports betting? it might even change how bookies set the lines.

[@josephpleone]

Absolutely loved this. More please

[@PhilGalfond]

Amazing. Thank you!

[@nickadoo]

phil you are so cool and I love each and every piece of content you release

[@PhilGalfond]

What a kind comment. Thank you so much!

[@nickadoo]

No - thank YOU @@PhilGalfond! I haven't even finished the video yet but I am recommending the Commanders take you on as head coach (or the Ravens if you prefer)

[@ilovebrandnewcarpets]

I’m almost 40 and I still play Madden (I know, “grow up”). Anyways, my buddies always give me a hard time because I ALWAYS go for 2. Now that might be a bit too crazy…but mayyyyybe? Just sent this to the group. TY Phil

[@PhilGalfond]

Happy to help! Man, I "retired" from Madden a while ago, but that was my favorite. You've got me missing it now.

[@r0lfus]

That was great!! 👍 Would be interested to see if you can come up with anything for golf?

[@gerrilla]

Phil, considering you are a PLO pro, I would have thought you would touch on variance. 2pt conversion is a much higher variance play. There isn't a large enough sample size in a given season, or even several seasons, to correctly realize your equity. In a given season, a football team might encounter these scenarios only a handful of times. Taking the higher variance play can have a large impact on their overall win rate- for better or for worse - and as you said, coaches want to keep their job. That said, I would still go for the 2pt conversions 😅 Great video!

[@SteamBoatZae]

Makes total sense. Would love to see this scenario ran out on a real in depth simulation on NFL seasons(not Madden) & to try to account for all the aspects of the Season. Such as Injuries, strength of offense/QB, intelligence of Coach / team moral / momentum / Road or home game, so many factors & variables but would love to get down to the Roots & real outcomes of this. Someone could definitely be onto something game changing if it really backs up the numbers & doesnt fall through as a bs experiment.

[@u2FasT4Me]

keep it up!! i fucking love you, you're advice and videos Phil Galfond!!!!!

[@PhilGalfond]

Thank you!

[@messymarv1111]

Break down poker player logic and startups, currently I’m building a bitcoin startup and have been playing poker for years,studying, but have recently transitioned out of poker, but all the skills that I learned from poker has allowed me to really understand decision making and dealing with the “traditional” business people that I have to work with and deal with and they are really bad at bluffing, but would be cool to see a video by you on how poker players would go about building startups (tech) and teams

[@jamievardy5890]

I've thought about this so many times, and from a sports bettor's perspective, I always think about the EV of decisions and understanding the flow of the game that I feel some coaches lack the insight to understand...

[@D_Rahs]

Great video. Could be interesting to break down Jeopardy daily double wagering and final Jeopardy wagers in game theory style. The players always screw it up. Especially Final Jeopardy.

[@D_Rahs]

@philgalfond This spam? (The telegram me message?)

[@Bresingeri]

I’m totally here for Phil Galfond, the new, nicer Haralobous.

[@hirollerm2916]

Intuitively, the reason why the first example works even though 2pt conv is 45% is that basically you are risking 2 points to gain 1 point for the whole process. So you will do it more often than not. The second example is less clear. You are not considering that a team down 3pts can score a td to win; and also that after missing 2pt conv and then scoring a td it leaves you up only 3pts and not 4pts if you would have taken the single (and we know it can only take seconds for the team to come back and get into fg range to tie).

[@MusicPhrase]

Good video, Phil, but I do have a question/critique. As poker players, we know all about sample size. We also know how the results of a sample size can drastically change based on our environment and competition. Thus, I don't think using 48%, or whatever the number, as a 2pt conversation rate can be some universal constant assumption. It's very likely whatever data is being used is a small sample size and is likely to vary significantly from team to team and opponent to opponent, etc. For example, if the Houston Texans were basing their 2 pt conversation rate only on the 2023-24 season, that's going to be too small a sample size. If they base their rate on many seasons, that won't be accurate because their team and opponents have changed drastically. Furthermore, their 2 pt conversation rate can change drastically based on each opponent. Their average might be 48%, but if they are playing Ravens red zone defense vs. Panthers red zone defense, that's could be a huge difference in number. So, these are some things maybe to add into consideration. How trustworthy is the 2 pt conversation % really at the end of the day? My win rate might be 30 BB/100 in a super fishy live game, but only 2 BB/100 in a tough online game.

[@BOnYTB]

NFL teams have models that take personnel in to account on both sides, time, score, situation, league average, their %, how often their opponent stops it, then how often it believes you’ll succeed given all of these things. They don’t just use tiny sample sizes, there are many many factors including simulations based on personnel.

[@MusicPhrase]

@@BOnYTB No, STILL A SMALL SAMPLE SIZE. Only 127 two-point extra attempts in 2023 for the league. That's not a big sample. It still just ends up being a 'best guess' at the end of the day..furthermore, it's not based on a large, super relevent sample size to those two teams playing, if those 2 teams have only played 2 games. It's a small sample based on league average stats. Of course teams and sportsbooks and everyone have this data, but it doesn't mean the data will be an accurate predictor for a given situation. Most all of the stat predictors are just best guesses, based on non relevant data. Purdue loses to FDU last year after being a 24 point favorite. But, the odds line is just a guess. We'd have to watch the teams play 1,000 times to know where the spread truly lies. Also, the success rate can change based on a team's preparation for you. None of these numbers are iron-fail proof. Take an extreme example: If you knew an NBA team was literally ONLY going to shoot threes and never a 2-point attempt, then you could play everyone on the outside and the players who shoot 40% all of a sudden drop to 20%.

[@MusicPhrase]

@@BOnYTB Announcer: "Welcome to the last second of the game folks where the Texans trail by 1 to the Ravens and are debating whether to send it to OT or go for 2 to try and win it." Head Coach to Offensive Coordinator: "Are you sure we should go for 2 pts?!! We do have our 3rd string quarterback in, who's been sacked 5 times and thrown 2 picks already, and our running back has a sprained ankle..all against the best Red Zone defense in the NFL." Offensive Coordinator responds, "well the stats say go for it coach! 48% success rate means we'll win most games." Head Coach: "We have stats for situations just like this?" Announcer: "Ouch and down goes the QB on that scramble, and the Texans lose the game! Fans are scratching their heads at that play call." Owner to Offensive Coordinator: "You're fired"

[@NFLed]

As a bigger fan of the NFL than poker, I enjoyed this analysis a whole lot. I play the online strategy game Strat-O-Matic Football head-to-head against other opponents and I often choose to go for 2 even early in the game unless my offense is poor and my kicker is almost automatic on XPs.

[@NFLed]

I see that someone posted a reply to my comment but when I click on Reply their reply does not show up (?).

[@lowellh.4949]

In the second example, it seems to me like you need to account for the team's probability of scoring a touchdown on the second drive while trying to get into field goal range. The difference between that probability and your chance of scoring a touchdown when you can't kick a field goal would be the deciding factor in if going for 2 would be beneficial.

[@jackhalas9493]

If you wanna be successful in 2 point conversion plays, assume your opponents will play man to man defense on the play. Use presnap shifts & motions as well as rub route concepts to free up open receivers

[@wydellbirchwood9146]

One thing you’re missing is that a field goal can be kicked from the 40 yard line… big difference between needing a TD vs a field goal from the 49 yard line…

[@AZ10s38]

The second scenario makes more sense than the first but the math is way off on both. Pressure is a massive factor in sports and you can be sure that teams with a lead have a MUCH higher 2 or conversion rate than teams behind.

[@PhilGalfond]

Why can we be sure of that?

[@AZ10s38]

@@PhilGalfond because that’s how sports work. There is a performance element that is hardly existent in poker and highly prevalent in sports. Poker players operate under uncertainty but they don’t have to deal with the pressure of performing. Completing a pass is a lot harder in a big moment, putting chips in the middle is not.

[@justinharris7650]

Was crazy as I actually thought about this like they will come a time when NFL teams will start going for two point conversions like normally every time I think I was watching some type of overseas football and it’s kind of how they do it was like yeah if only an NFL team went out their way to build a squad just for this to go for on two point conversions like they would win so many more games because all you need to do is score one and it’s the other team can’t compete and you’ll forever have a two point lead

[@iillinn]

Good points literally, good sir! I'd be interested to hear your thoughts on the mindset of needing to get even in a session. Is there any harm in thinking that way and thereby allowing it to affect one's decision making?

[@therealhossroot]

Yeah but what about river bluffs that keep falling into a single sizing😮

[@josephgill7140]

Worked in basketball analytics for close to a decade (50+ NBA players worked with). Decided to start transitioning that time into poker a few years ago because cards are, intrinsically, meritocratic over a long sample and somehow less infuriating.

[@snake0911]

There’s are so many mistakes made in football. 1. Scoring too fast before halftime when you can milk the clock. 2. Return a kick from the end zone 3. Not spiking the ball to kill the clock vs running a play after a first down with 40 seconds left and no timeouts 4. Not running the play clock all the way down when it’s clearly advantageous to do in the 4th 5. Reaching the ball out near the end zone on 1st and 2nd downs.

[@GLazarus604]

Coincidentally, I've been applying this logic to Madden '95 lately to great effect.

[@shesse5315]

loved it

[@yosik92]

Phil if “NFL coaches” did the right play for “their team” we wouldn’t have politicians and you (or any other pro) wouldn’t be making any money playing poker. Great game theory and more importantly life lesson!

[@davidallan4900]

In the 2nd example, haven't you discounted the scenario in which the team that kicked the extra point after the first TD drives down the field, intending to kick the FG on 4th down if necessary, and scores a TD and wins? They have the option of either winning the game or tying it - isn't that an informational advantage too? I think you've calculated it as if their only option is to tie?

[@m4ll3b64]

What if succes-rate changes with the rate of attempts? Are you not expecting other coaches and teams to adjust to your gameplan? How much can you exploit before opponents adjust?

[@carlossandoval701]

This is a funny video to watch after watching the lions blow the game last night. Still really interesting

[@jasonboutelle9501]

This is so fascinating as a poker/football enthusiast. Although I do think we undervalue the intangible affect of momentum. Its very real yet impossible to quantify in mathematical terms.

[@PhilGalfond]

It's real, but it works both ways. I think the impacts are often overstated and used as an excuse not to trust the math. Glad you found it enjoyable, thank you!

[@jasonboutelle9501]

What I mean is that when the team goes for 2 and misses they become so defeated that I think it becomes mentally and physically draining.

[@caseyw9864]

@@jasonboutelle9501 Yes, and if they make it then the inverse happens ... therefore is null. However, there are interesting studies that after interceptions/big touchdowns momentum is not actually a thing. you are no more or less likely to score/get a stop.

[@danielniese535]

Exactly, there are so many factors in football that cannot be quantified. Such as the mental state of every player and coach. These analytics are missing variables