Bill Benter did the impossible: He wrote an algorithm that couldn’t lose at the track. Close to a billion dollars later, he tells his story for the first time.
No, he told this story decades ago, and published papers detailing his algorithms, e.g.
Benter, W. (1994).“Computer Based Horserace Handicapping and Wagering Systems: A Report.” In L. Hausch and Ziemba (eds.), Efficiency of Racetrack Betting Markets. San Diego, CA: Academic Press, pp. 183–198.
Benter, W. (2003). “Advances in the Mathematical Modeling of Horse Race Outcomes.” 12th International Conference on Gambling and Risk-Taking, Vancouver, BC, Canada (May 2003)
There are videos of him (just search youtube) presenting at conferences detailing some of his algorithms too... there are no secrets being revealed for the first time here!
>there are no secrets being revealed for the first time here
I think this is the first time Benter has admitted winning the Triple Trio. The only mention of it I could find just said it was suspected he had won the unclaimed prize.
To be fair, I just got to the end of the article and discovered that it mentioned this paper too...
One aspect that isn't mentioned, is that Benter (and Zeljko Ranogajec, mentioned at the very end) got a large edge over over punters by negotiating large discounts/rebates from the betting pools by providing such large volumes. It's easier to make a profit when you are getting (say) 10% back on your losing bets...
If you're interested in those two papers, the earlier paper 'Searching for Positive Returns at the Track: A Multinomial Logit Model for Handicapping Horse Races' (Bolton, Ruth N. and Randall G. Chapman) is also worth reading:
I've found that many conference "papers" were never actually written as papers at all, they were just oral presentations at a conference. Someone who was there cites it in a paper of their own, and then a bunch of other people copy the citation without actually having seen/read the "paper".
I don't believe that's a paywall, just Researchgate being scummy like usual and using a citation while pretending to have a copy - all that would do is ping Benter to ask for a copy. I can't find any evidence that that conference presentation was published or that the conference proceedings as a whole were ever published.
It's no secret that it's possible to figure out which horse is going to win a race with reasonable statistical confidence.
The trick - and always has been - is making sure you can make a profit from it.
If a horse has a 30% chance of winning a race, and you're being offered odds as if it only has a 10% chance (i.e. 9/1), you have identified "value" or what Kelly Criterion would describe as "edge". You can mathematically in the long term make a profit, and Kelly can even tell you what %age of your bank to wager to optimise your expected returns: https://en.wikipedia.org/wiki/Kelly_criterion
Identifying the "true" odds of an outcome and then identifying what odds you're prepared to take (adding an over-round), is a key piece to what bookmakers do when creating what they call "the tissue" - the opening odds they offer that is then adjusted to manage liabilities in line with incoming weight of money (WoM) from bettors.
Kelly was a colleague of Claude Shannon. His papers have stood up to the rigour of mathematical analysis and also make intuitive sense to those who spend enough time thinking about it.
It has successfully been applied in casinos (Blackjack system pioneer Edward Thorpe is/was a fan), bookmakers, hedge funds (Thorpe again, but many other funds apply Kelly too), and everything in between.
But here's the rub: you won't get to keep your edge for very long. Bookmakers, betting exchanges and parimutuel systems all adapt to deal with long-term winners. Your window of opportunity is limited. Already bookmakers have good systems for identifying true odds for English Premier League games and make a tidy sum on over-rounds (profit margins) as thin as 3% because they can eliminate value.
That said, as any fan of Sabremetrics will tell you, statistical analysis of a sport you love can be very rewarding and I would not put anybody off this if they already love the horses (as I still do).
Just tread carefully, and to make a long-term killing you're going to have to treat your setup like a hedge fund and always be moving around finding new edges.
I'm @p7r on Twitter if anybody wants to talk about this - I could talk for hours and hours and hours... :-)
Have you thought about just recording yourself talking with a few friends for hours and hours about it and uploading online (e.g. Youtube)?
(I don't gamble and never learned statistics, but I am interested occasionally and digging in more details "behind the headlines", so what's why I wouldn't mind listening to this.)
To add to the complications of picking the size of bets, you will always find that you can get more money matched on a loser than you can on a winner... so even after identifying 'value' bets and using Kelly to size your stakes, you may find out that you never quite get enough money on the winners and that will limit your profits too!
(Edit: getting the money matched only applies to bookies / betting exchanges, the equivalent problem for tote betting is that the price of winners will often be worse than your algorithms predicted)
Exactly - getting matched is a big problem, and is the main way that bookmakers are able to prohibit winners. Most people in the UK have had their stakes limited after a couple of winners, and it's almost become a joke.
Betfair and other exchanges provide more opportunity, but if your Kelly output is telling you to put £5k on a horse on a Tuesday evening AW track race, you are going to have to spend some time thinking about how you _actually_ do that without having the odds spiral in: markets respond to liquidity, and you are liquidity.
There are some papers on how to deal with this, but none of them have a solid answer. My solution was to put in the amount Kelly advised and if I didn't get fully matched, c'est la vie!
Would love to hear about your process for working out probabilities, etc. I've had a horse racing project kicking around ever since I was at uni but that's mostly around stats gathering vs prediction models
I started after the dot.com crash in 2001/2002 when work was hard to find.
It sounds like a dream job, but actually, it's not. You start to tire of a sport you loved, and each morning you wake up with a pile of cash that you want to make a few percent bigger by the end of the day.
You're not adding anything to society, you have no colleagues, there's little to talk about with friends and family, and most people think you're a lazy bum.
Slowly your life becomes consumed - I was doing 16-17 hour days just moving money around and trading - your diet turns to crap and you start to resent what you're doing.
It's at that point you are vulnerable to "breaking the rules" and chasing losses, over-betting or doing other stupid things. Back then I did not have the resources to automate my strategies, so I lost £5k in 2-3 days about 9 months in and thought "time to get a proper job before I do some serious damage", and got back to working for a living.
I have since then automated many of my strategies and it's my favourite hobby, but I am under-capitalised and not interested in using other people's money: it will remain a hobby for some time yet, but I can get 10% RoI a month for some strategies, which definitely beats savings. Just need to wait for the bank to grow, and grow and grow... Maybe one day I will be a "professional gambler" again, but really it's just data science and software engineering for a very strange product.
Well done for making it work! In those pre-Betfair years, the logistics of actually getting your bets on must have been painful. We are a bit spoilt nowadays when the betting can be done through an API and everything can be automated (the downside is seeing a large % of profits going to the exchange...)
The commission structure is tricky, no doubt about it. By the time I hit premium charges (which I expect to), I'm hoping competing exchanges with lower commissions will have got a bit more liquidity in them.
I've talked to finance shops (I live in London, so there are plenty about), but they worry about my lack of formal mathematical training and of course they don't want to hire "gamblers". I've got some calls with a few people lined up that might mean I get to make the jump at some point in the next few months.
I am enjoying it being a hobby right now, and I seem to be more comfortable calculating and assessing risk and pay-off than others which can be useful in an engineering team with an operations component.
Is this a reprint? I swear I've read this article before.
I spent some time doing handicapping algorithms, if anyone is coming from other types of sports betting, the biggest challenge with predicting horse racing is the lack of fixed odds in a US-style parimutuel system. No matter how good your handicapping model is, you also have to build an additional model to predict how much of the pool will end up on particular horses.
The price changes constantly and you have no ability to lock in a price or withdraw your bet if you made it at a certain price and the price changes. It wasn't uncommon to see a horse at 3/1 when the race starts which could seem like a good value and attract a well-placed bet, only to see a huge volume of money come in with milliseconds left that only hit the board after betting is closed to push the price down to 8/5 and make it a very very poor value. As a parimutuel bettor you are stuck with whatever the final price is.
It becomes a vicious circle because the only way to be more certain about the odds is to place your bets as late as possible. Which means that fewer and fewer bets are struck earlier on, making the 'current' odds even more inaccurate. Which motivates people to wait even longer before betting...
Eventually it became almost an HFT-type thing like in the financial markets, the biggest advantage comes to the person able to put a bet in at the last possible nanosecond because they will have the most accurate prices.
Which would be fine (or at least comparable), except the pool software is way slower at updating when new bets come in and the tracks take a 10-17% commission on every bet, which is a drag that is just really difficult to overcome.
Salient item from the article - theres tons of opportunity out there so the quant funds get involved:
In 2016, Susquehanna International Group LLP, an American quantitative trading company, started an Ireland-based operation called Nellie Analytics Inc., targeting basketball, American football, soccer, and tennis. Phoenix, a proprietary sports-betting company with headquarters in Malta and data-mining operations in the Philippines, won a £9 million ($13 million) investment in 2010 from a unit of RIT Capital Partners Plc, the £3 billion trust chaired by Lord Jacob Rothschild of the global banking dynasty. (RIT sold its stake in 2016 to a private buyer, quadrupling its money.) What isn’t widely known is that Phoenix was founded by former employees of Woods, including his protégé Paul Longmuir.
I know that algorithmic betting on horse racing is popular in the UK via exchanges like Betfair. Does anyone know if there is an exchange/API for Hong Kong horse racing?
No, he told this story decades ago, and published papers detailing his algorithms, e.g.
Benter, W. (1994).“Computer Based Horserace Handicapping and Wagering Systems: A Report.” In L. Hausch and Ziemba (eds.), Efficiency of Racetrack Betting Markets. San Diego, CA: Academic Press, pp. 183–198.
Benter, W. (2003). “Advances in the Mathematical Modeling of Horse Race Outcomes.” 12th International Conference on Gambling and Risk-Taking, Vancouver, BC, Canada (May 2003)
There are videos of him (just search youtube) presenting at conferences detailing some of his algorithms too... there are no secrets being revealed for the first time here!