I was going to respond with the factoid about the World Series being named after a newspaper, but a Wikipedia check (and subsequent googling) revealed that it's probably false. Not news to you, but maybe of interest to others who, like me, had absorbed the popular misconception!
American Football is actually based on the delineation between sports played on horseback and on foot. So maybe it is an antiquated distinction but it makes sense. Maybe we can rename it infantryball.
The author is from London and the internet is global. Expecting an author to cater to every other culture with everything they do is a bit self-centered, IMO. It should be on the reader to understand that there are differences, not on the writer to make sure every single cultural difference is pointed out. It seems kind of weird to expect him to alter his headline to cater to a country he doesn't live in.
If you were writing an article about potato chips, it would seem kind of silly to expect you to add "(crisps)" in the title to cater to our English friends, so I'm not sure why you would expect that here, especially when the vast majority of the world calls it "football".
In fairness, it's not uncommon to specify "American football" for international audiences, so I don't think it's totally unreasonable. But on balance, adding a parenthetical word into a headline is just a little too much to ask. There's not much harm in bringing in someone who was expecting one but got the other anyway.
That's like saying 'Perhaps include "(France)" in the title' when talking about Paris. Football is well known all over the world and perahps the most popular spectator sport.
One of my professors was working on a PHD research project involving predicting soccer match outcomes using a neural net. At that time it was very...wrong...but I wonder if he ever made any progress towards better accuracy.
Inaccurate would probably be a better word. The results were fairly inconsistent, and for whatever reason he decided that he wanted to bet real money before perfecting (or at least as best as possible) the predictions and he lost a lot of money haha
But even that is very dated - one of the authors, Stuart Coles has worked at a gambling syndicate for many years. They at the very least incorporate expected goals into the model, but no doubt have all sorts of esoteric models by now.
Yep, Dixon-Coles is way out now. Generally, Poisson-based models don't work that well. They model some aspects of the game correctly but not others. And when it is bad, it is terrible (their paper was genius though, and changed the industry). You also wouldn't do the one-hot/regression stuff on teams, as there is so much player-level data...some kind of online, off/def model is useful though.
You also have the issue of modelling a bivariate as two independent univariates. The adjustments made by the authors were intended to get around this problem but there has been a lot more work on this aspect of the problem since then, and a few new solutions.
And you are right, expected goals was cutting-edge back in ~2011-12 but even some clubs have caught on now. Most gambling syndicates, afaik, actually produce their own datasets (the largest ones employ a huge number of people, I know one that started up five years ago and already has ~10 PHds on staff + traders + data people, etc.).
I will say though, it is surprising how inefficient betting markets were until recently. In the early part of the 2010s, I imagine that large syndicates were making, conservatively, 300-400%/year in the largest markets that were, apparently, "unbeatable". I am aware of very simple models, similar to DC, that could beat retail bookies on Brier for EPL as recently as 2017. And, ofc, smaller markets are far softer (or other sports). At the cutting edge, yes...it probably is complex (I have no idea). But a surprising amount is actually just doing the simple things well.
I do generally find these top-down team-strength models quite primitive though. Teams are made up of players, plans, all sorts of under-studied interactions. I would also say that there are enough quant jobs in football now that you don't really need to gamble to make a living. :)
Dixon-Coles, Ntzoufras, McHale, there are a few papers on Bivariate models from German authors (Groll...so google "Groll bivaraite Poisson"). I would also understand ranking algorithms (there is a book called Who's #1?). Bear in mind though, most papers are fictional/p-hacked/just bad.
I would also think carefully about what you are doing and why. Most people fail because they try to bet on markets that are, for them, unbeatable (for example, football data is expensive). It is far easier to pick off obscure markets (I did not do this because I had a junior high school maths education and needed some guidance).
The thing is, the size of the wager and the payouts are just as important. I was never a sports gambler, but spent two years counting cards at blackjack in rural casinos as my job. This plays out with the Kelly Criterion: https://en.wikipedia.org/wiki/Kelly_criterion
So you can have inefficient odds as the house, and still win. Did the models also provide optimal bet size given the probability of winning?
Larger syndicates also have the problem that it's not always easy to wager as much money as they'd like on a small handful of leagues. So it becomes about how you can accumulate data and insight into a broader set of competitions.
A lot of the newer syndicates spread themselves quite thin afaik. They maybe don't have the contacts to get liquidity/early prices so they do lots of sports (Tennis and Cricket being two that appear to be growing...again, afaik).
I have heard that the largest syndicate has groups that only cover one football team. I have no idea whether this is true or how/where you get the liquidity to justify this.
This is the reality of how the bettors move the lines (ie odds on offer), large bets or a lot of small bets will move the line. In effect the final line at kickoff (or whatever) is kind of a distilled, crowd sourced expectation for the result of the match. The key is to identify advantageous lines when the major books publish them and place bets before the "public" moves the line.
Remember, no one wins at gambling by picking winners, you need to look for value and find where the bookmakers have miss-priced a team in a match.
Then you have to deal with all the corrupt behaviours the sports internet bookies will deploy to limit their exposure to you, which is the other reason why you won’t win.
Professionals don't use retail bookies, they bet with Asian books who take action from sharps (Pinny, SBOBet, SingBet, IBC...although it varies, sometimes their lines are soft).
> Remember, no one wins at gambling by picking winners, you need to look for value and find where the bookmakers have miss-priced a team in a match.
No. You win by being the bookie or better yet, by not gambling. Sports betting is worse than a zero sum game because you gotta pay the bookie too. I'm betting that 99.99999% of everyone who bets on sports loses money.
31 comments
[ 2.5 ms ] story [ 70.8 ms ] threadAre you sure about the name basketball? Don't want to call it sackball or something like that? Or maybe shuttlebasket?
I always chortle at your "world" series too!
What?
If you were writing an article about potato chips, it would seem kind of silly to expect you to add "(crisps)" in the title to cater to our English friends, so I'm not sure why you would expect that here, especially when the vast majority of the world calls it "football".
https://dashee87.github.io/football/python/predicting-footba...
But even that is very dated - one of the authors, Stuart Coles has worked at a gambling syndicate for many years. They at the very least incorporate expected goals into the model, but no doubt have all sorts of esoteric models by now.
You also have the issue of modelling a bivariate as two independent univariates. The adjustments made by the authors were intended to get around this problem but there has been a lot more work on this aspect of the problem since then, and a few new solutions.
And you are right, expected goals was cutting-edge back in ~2011-12 but even some clubs have caught on now. Most gambling syndicates, afaik, actually produce their own datasets (the largest ones employ a huge number of people, I know one that started up five years ago and already has ~10 PHds on staff + traders + data people, etc.).
I will say though, it is surprising how inefficient betting markets were until recently. In the early part of the 2010s, I imagine that large syndicates were making, conservatively, 300-400%/year in the largest markets that were, apparently, "unbeatable". I am aware of very simple models, similar to DC, that could beat retail bookies on Brier for EPL as recently as 2017. And, ofc, smaller markets are far softer (or other sports). At the cutting edge, yes...it probably is complex (I have no idea). But a surprising amount is actually just doing the simple things well.
https://www.academia.edu/37585525/Flexible_Regression_Models...
I do generally find these top-down team-strength models quite primitive though. Teams are made up of players, plans, all sorts of under-studied interactions. I would also say that there are enough quant jobs in football now that you don't really need to gamble to make a living. :)
I would also think carefully about what you are doing and why. Most people fail because they try to bet on markets that are, for them, unbeatable (for example, football data is expensive). It is far easier to pick off obscure markets (I did not do this because I had a junior high school maths education and needed some guidance).
I have heard that the largest syndicate has groups that only cover one football team. I have no idea whether this is true or how/where you get the liquidity to justify this.
Then you have to deal with all the corrupt behaviours the sports internet bookies will deploy to limit their exposure to you, which is the other reason why you won’t win.
No. You win by being the bookie or better yet, by not gambling. Sports betting is worse than a zero sum game because you gotta pay the bookie too. I'm betting that 99.99999% of everyone who bets on sports loses money.