I can't find the actual picks that this LLM made, are they listed anywhere? As squirrelmaster commented [1], would be very worthwhile to see how this picked in regards to the (betting) markets: against-the-spread (ATS) and the money-line.
Pickem is straight up normally so would be money line bets essentially.
The spread is designed to make the bet a 50/50 so it would need to take into other factors or compute it’s spread to determine if there is an edge on Vegas.
In my experience, you can get extremely good performance in this style of sports prediction with extremely basic methods. E.g. doing some sort of basic Elo rating system + choosing the higher rated competitor to win every game. This is simply because almost all human competitors will frequently try to predict upsets (as its more fun), but it will harm their performance overall.
I would be cool to see if something like this can do can outperform a benchmark Elo system but even that is an extremely high bar that most sophisticated models can't beat.
Also in sports (especially a sport like American football), you have circumstances that change from week to week: i.e., a quarterback gets injured, and will sit out for a game. This won't show up in the Elo rating, but normal football fans would be aware of it and take it into account in making predictions.
also in sports, you have a league (multi-billion dollar advertising business) and league-controlled referees who are inconsistent and may or may not have a narrative they are trying to push (aka, so-and-so team will do better advertising/TV ratings if they make it deeper into the playoffs)
how do you account for inconsistent/bad calls/missed calls when there is zero accountability? you can only review some things sometimes, and supposedly "new york" is watching in a booth to tell the refs "bad call/good call", yet announcers + players + coaches all disagree with refs on calls all the time that 100% affect the outcome of the game
i personally don't think the NFL + NBA is to be trusted. i don't want to sound like some crackpot conspiracy theorist but i feel like there's enough evidence that... it's not a serious product. it's entertainment first and foremost.
i don't know what corporate boardroom focus group research was done to identify the fact that viewers respond well to something they hate (bad calls), but... boy, they sure are not trying to use technology to remove them from the game
9 comments
[ 5.1 ms ] story [ 25.1 ms ] threadThe performance is very impressive though, NFL games aren't easy to pick.
[1] - https://news.ycombinator.com/reply?id=39019606
The spread is designed to make the bet a 50/50 so it would need to take into other factors or compute it’s spread to determine if there is an edge on Vegas.
I would be cool to see if something like this can do can outperform a benchmark Elo system but even that is an extremely high bar that most sophisticated models can't beat.
cowboys 2 seed
eagles
how would that have worked this season? those high elo teams lost badly multiple unexpected times no?
how do you account for inconsistent/bad calls/missed calls when there is zero accountability? you can only review some things sometimes, and supposedly "new york" is watching in a booth to tell the refs "bad call/good call", yet announcers + players + coaches all disagree with refs on calls all the time that 100% affect the outcome of the game
i personally don't think the NFL + NBA is to be trusted. i don't want to sound like some crackpot conspiracy theorist but i feel like there's enough evidence that... it's not a serious product. it's entertainment first and foremost.
i don't know what corporate boardroom focus group research was done to identify the fact that viewers respond well to something they hate (bad calls), but... boy, they sure are not trying to use technology to remove them from the game