The insight about starting from the public odds is so important. It's like running from a polar bear: you don't have to outrun the bear, only your buddy. Not counting the track take, any deviation from the public odds in the correct direction makes you money.
(And the public is often reliably biased -- this is why they say sports betting is more about betting on gamblers than on games.)
A smaller deviation makes you less money that one time, of course, but it also saves you lots of money if you're wrong, ensuring you live to bet another day.
Also in real life: people in aggregate aren't dumb (just a bit biased), so using the public opinion as your prior and then adjusting it Bayesically with private information gets you a more reliable edge than guessing wildly.
The dynamics of odds movements are interesting in their own right imo. Some sports, with many days between matches (nfl, college football) lines can move drastically between when odds are published to match start. Others sports that play daily or by-daily often don’t see the swings.
And then what facotors influence the line movement? People with predictive models that see great value in opening lines can move the line a lot. Things like injuries, starting lineup, breaking news before the match can certainly
move the line. Casual bettors and fans betting on their team (the public) move the line but usually day of the match.
The key insight is the books are trying to balance winners/losers to maximize their profit, that’s why the line moves. It’s also why historically spread betting is nearly 50% odds.
The key insight is the books are trying to balance winners/losers to maximize their profit, that’s why the line moves.
In two-way markets (e.g. a tennis match) maybe. With more outcomes, it's a rare market that ends up with the bookmakers locking in a profit regardless of the outcome. For horse racing, the favourite in the race is most often a loser in the book for the betting firm.
Even for football (soccer) betting, i.e. team A/draw/team B, generally a bookie is very happy for a draw, because people just can't get very enthusiastic about betting on a tied result. Raising the odds on offer for a draw won't attract many extra regular punters, it only risks luring professionals who are more sensitive to the odds - and they aren't the customers you really want!
In horse racing, at least in the US, betting is done via a pari-mutuel system where the house pools all the money, takes a 20% cut, and then pays the remainder split between winning tickets proportional to the bet size.
> Things like injuries, starting lineup, breaking news before the match can certainly move the line.
But! At least until recently, people overreact to events like these. If you see the line moving strongly at once, it's more often an overreaction than an appropriate adjustment.
Along those lines it used to be a very good strategy to bet against popular clubs/players. As fans like to bet on their favorite teams/players there was too much money on that side so the bookies moved the line to balance.
I don't follow sports betting anymore and my guess is that there is enough smart money out there these days to close that market inefficiency but it's an interesting mechanism.
It's not just a 'wisdom of the crowds' thing where you are using people's betting in aggregate. The odds also reflect private information that other in-the-know people may also have.
The problem with pool betting (like in HK) is that the odds only become significant right before the start, when everyone has placed their bets. This means that there is a strong motivation for everyone to wait until the last possible moment to bet - meaning that the odds then only become useful even later!
This is a good book that discusses how an individual may be wildly wrong, but in aggregate, groups are remarkably good at estimating many common things. Although with horse racing, the betting favorite only typically wins ~30% of the time or places (1st or 2nd) ~50% of the time. Given that this is with average field sizes of probably 6-8 horses, it is materially better than blindly betting.
>>> (And the public is often reliably biased -- this is why they say sports betting is more about betting on gamblers than on games.)
One mistake here. It's the house that sets the odd, not the public.
When the house sets the favorite to 1:3 odd and the second favorite is 1:4, they better be damn right about that proportion. At the end of the run, the money on the losing side will need to cover the winning side, and if it doesn't the company must pay the difference out of pocket.
The direct effect is that there is an enormous spread, the article mentioned about 17%, for the exchange to cover high uncertainty and make some profit.
I doubt it's parimutuel per the strict definition. It's subject to extreme variations due to liquidity and doesn't work well in practice, especially for quant trading and at the volumes involved here. (If it were, the article certainly did a terrible job of showing it, where's the part where odds moved last minute and massive losses ensued, or where both rivals took the same large position causing massive losses to one another).
Anyway, the deeper point still stands, the house takes an ENORMOUS margin, could be in the order of 25% (even nowadays). This changes everything.
From my understanding[1][2], the Honk Kong Jockey Club does indeed follows a strict parimutuel pool betting system, like the PMU in France (at least years ago).
Though you are right, the house have historically taken massive cuts per pool. Eapecially as for years they were the only allowed gambling arenas.
About a year ago I read this and attempted to build a similar model (with more layers ;)) using data I scraped from Hong Kong Jockey Club’s website. Although I used much fewer features, it still produced profit in held-out races: https://teddykoker.com/2019/12/beating-the-odds-machine-lear.... Obviously there are many caveats when backtesting like this but I thought it was a fun project!
I had a coworker who would prepare for weeks for stakes races and follow a few second tier horses as well. Twitter made some aspects easier as there is a racetrack Twitter community. His specialty was identifying exactas where a long shot would place or win with a favorite.
He’d pay people to film workout at Belmont and Saratoga and tweak his model (an Excel spreadsheet) based on what he saw. He would have a sense based on the workouts, weather, etc and would pick 4-10 races a week.
His company is http://www.ave4.com/
I meet him a few times, interviewed to work there (didn't get the job.) I mostly talked to a lawyer who was very nice, but reminded me of Kobayashi from the Usual Suspects. The offices are a bit extravagant with some amazing original artwork on the walls, https://www.jendoco.com/portfolio/fourth-avenue-analytics/
I've been spending almost every weekend for the past few months trying to break a crypto crash game (called bustabit) using ML. While I'm also dealing with probabilities, my challenge has to do with the limited data I'm dealt with even though their API is public.
In horse racing, you have access to so much shit that you can feed into the AI, like rider statistics, horse performance, horse attributes, etc. In my case, I only have the game/round number and the bust result (red or green). From thereon, I have to create many other variables (aka featuring engineering) like round win/lose streaks, moving averages, MACD, etc. I can also feed in what and if other players are currently betting for the round or not.
Currently, the house always wins. And their game is rigged so they have that 1% advantage over you. However, so far with my AI, I've came with enough confidence score to win as long as my bot plays 0.001% of the games.. which isn't declared victory since those blackswan games happen once every 3 weeks.
There's also another way to beat it without using any ML, just plain old fibbonaci martingale - requires around 10k starting money for error margin but it ends up always ahead of the house.. That's just a math way to beat it and not very fun :)
It's probably crazy of me to even attempt to predict randomness, but there's something intriguing when you mix big data and ML with forced randomness. Again, they have to keep forcing the odds to be 49% vs 51%..
Every professional gambler to whom I've spoken or have known easily puts in 60+ hours a week... some of them, to make no more than $40,000 - $50,000 a year.
I struggle with understanding that mentality, since just about anywhere, doing just about anything, if you're working 60+ hours, you can easily make double those numbers, but hey... I'm glad people like that exist.
I've worked in betting and I've seen people consistently printing millions of pounds a year with remarkable stability over the years (I was myself running an exchange).
However they may not fit the popular image of an older man addicted to horses and drinking. They're professional working at (small) companies doing that activity full time.
It's a major national event for Hong Kong, like the guy winning a $1B national lottery. The winner will be shown on TV and newspapers, a random citizen who accidentally hits the jackpot and got lifted to being a king.
Except in that case, it's some foreign math computer nerds, who have already profited billions. Billions of dollars taken off the cumulative bets of Hong Kong citizens. It's not acceptable socially or politically.
The guy was already cashing hundreds of millions anyway, leaving a few millions on the table doesn't make a difference to him. Better stay anonymous and safe.
> The winner will be shown on TV and newspapers, a random citizen who accidentally hits the jackpot and got lifted to being a king.
I’m not aware of this happening in HK. Mark six (lottery) winners are not shown on TV or in newspapers. Winners are announced, yes, but it doesn’t include their identities.
I’m not sure about the process when you win millions of dollars, but you can buy tickets at the track, which are anonymous, and you just cash them in later. Of course, my experience is only winning a few $100s.
> It's not acceptable socially or politically.
Socially, I more expect it to be the other way around. Other gamblers will probably applaud them for finally doing the impossible. This was also in 2001, nobody would even consider this to be politically unacceptable. Even now, that wouldn’t be the case.
"Other gamblers will probably applaud them for finally doing the impossible."
I suspect not. Some people have the idea that "luck" is intrinsic to yourself, that it is somehow a measure of your own worth, like karmic rewards. This view seems prevalent in China.
Thus systematically beating the odds will seem like cheating.
This is Hong Kong in 2001, quite different from China. Horse racing is also taken very serious in HK, so when you’re able to beat the odds, it wouldn’t be seen as luck, but rather all your hard work finally paying off.
If you feel that way, then why even place the bets? I mean, you could generate a text file that said "I would have placed the following bets" and then see if you would have won...
Sometimes, you don't realize the magnitude of something you're about to do until it's done. Even if you setup to buy all the wagers, so you know you'll win, you might not understand the consequences of winning until it happens.
I suspect they never thought they would win. And when they did win, and saw the amount they won, they somehow didn’t think it would be fair if they took it.
This censorship mentality is pervasive. I also hate it.
For example, even in something as important as a recent US Supreme Court case, the Washington Post reported it this way: “F--- school, f--- softball, f--- cheer, f--- everything,” 14-year-old Brandi Levy typed into Snapchat one spring Saturday.https://www.washingtonpost.com/politics/courts_law/supreme-c...
No, Brandi didn't actually type all those dashes.
Fortunately, the court documents have the actual words in them.
(And the public is often reliably biased -- this is why they say sports betting is more about betting on gamblers than on games.)
A smaller deviation makes you less money that one time, of course, but it also saves you lots of money if you're wrong, ensuring you live to bet another day.
Also in real life: people in aggregate aren't dumb (just a bit biased), so using the public opinion as your prior and then adjusting it Bayesically with private information gets you a more reliable edge than guessing wildly.