How to Make $1 Billion Betting on Horse Racing with Machine Learning
Вставка
- Опубліковано 18 тра 2024
- Bill Benter is arguably the most successful sports bettor of all time. Its estimated he made close to $1 Billion betting on horses in Hong Kong. His story is well documented, but the models he used and the reasons for his success are often overlooked. In this video I break down the circumstances that made his his model profitable and the specifics of the exact machine learning model that he used. At the end I also analyze the feasibility of creating another model like this in 2022.
0:00 Intro
0:41 Who is Bill Benter?
1:22 Lesson 1: The Law of Large Numbers
2:30 Blackjack to Horse Racing
2:50 Bankroll Management
4:07 How did His Billon Dollar Model Work?
4:57 Secret Sauce
6:00 Improving the Billon Dollar Model
6:35 $$$$$$
7:12 The $10 Million Dollar Ticket
8:05 Could you replicate this?
National Problem Gambling Hotline: 1-800-522-4700
Articles Referenced:
- www.gwern.net/docs/statistics...
- www.tepinvestments.com/blog/b...
- en.wikipedia.org/wiki/Kelly_c...
- • How To Make $1 Billion...
- www.wired.com/2002/03/betting/
- www.wired.com/2002/03/betting/
- www.marca.com/en/lifestyle/20...
- www.bloomberg.com/news/featur...
#DataScience #KenJee
⭕ Subscribe: ua-cam.com/users/kenjee1?sub...
🎙 Listen to My Podcast: / kensnearestneighborspo...
🕸 Check out My Website - kennethjee.com/
✍️Sign up for My Newsletter - www.kennethjee.com/newsletter
📚 Books and Products I use - www.amazon.com/shop/kenjee (affiliate link)
Partners & Affiliates
🌟 365 Data Science - Courses ( 57% Annual Discount): 365datascience.pxf.io/P0jbBY
🌟 Interview Query - www.interviewquery.com/?ref=k...
MORE DATA SCIENCE CONTENT HERE:
🐤My Twitter - / kenjee_ds
👔 LinkedIn - / kenjee
📈 Kaggle - www.kaggle.com/kenjee
📑 Medium Articles - / kenneth.b.jee
💻 Github - github.com/PlayingNumbers
🏀 My Sports Blog -www.playingnumbers.com
Check These Videos Out Next!
My Leaderboard Project: • I Built the FIRST EVER...
66 Days of Data: • What is the #66DaysOfD...
How I Would Learn Data Science in 2021: • How I Would Learn Data...
My Playlists
Data Science Beginners: • Data Science Beginners
Project From Scratch: • Data Science Project f...
Kaggle Projects: • Kaggle Projects
Thanks for watching everyone! Do you like data science related stories like this? Let me know other ones you would like me to deep dive into in the comments below!
i figured after watching Id have the ultimate algo. Im gonna have to settle for my yar ovrt year negative 15 pct returns Im proud that its at house take levels let me know if theres an easier way. thx for the vid,
I really appreciate that you acknowledged gambling addictions at the end. Fun video, I learned a lot.
Thanks for watching! Yeah, didn't feel right if I didn't. I probably should have also mentioned more that the house essentially always wins still
Finding a way to beat horse racing was the reason I got into data science to begin with.
Since you aren't against a sportsbook employing Stanford alums, you just have to beat people and the vig.
Needless to say, it's far easier said than done. I knew some of this, but not nearly all. This is incredible.
Glad I could provide some additional insights!
This is a great video, please keep up the good work. Really looking forward to these Machine Learning case studies like this
Thanks Tuong!!
Really enjoyed this style of video! Such an informative and interesting breakdown, crazy to think that model was super advanced a while ago! + title and thumbnail game is on point! Nice one, Ken!
Thanks Jack! I've been enjoying your channel. Keep up the good work!
This is so. freaking. sick. Awesome vid as always my guy!
Thank you my dude! One day we can build horse racing models to compete
A quick note on Correlated/Colinear Features. These can be a huge pain for applications such as Sports Betting, where we are weighing certain factors relative to other factors. Concepts such as introducing dropout, random noise, and using clustering to engineer/select features might be helpful. Doing that would reduce the costs of running a model, improving ROIs even if drops absolute performance
Thanks for sharing!
Great video, you really explain well ! Very interesting topic, just a quick question, when you talk about better models, helping to avoid multicolinearity, could you be more specific ?
Excellent video ! Could you be more specific about the models that are better than multinomial logistic regression for this type of job ?
Very cool video. Enjoyed the video thoroughly
Thanks for watching!
Thanks Ken, this is superb story to share it my brother to introduce him to applications of ML in real world before data chaos
Awesome! Thank for watching!
Thanks Ken, very good video
Thanks for watching!!
Great video Ken ! This makes me feel more excited about following your future ML courses 🌻
Thanks Ibraheem!
Mom, I'm not a degenerate gambler who had to move back home because I lost all my money, I'm a data scientist!
hahaha
Nobody has a "Gambling problem", only a LOSING $% PROBLEM!!!
That's me, but I'm pretty sure I'm on my path to millions. Horses are still super profitable, it just takes at least 5 years until you learn to beat the market. Non-gambling people have no idea how much time it takes to develop a strong winning system.
30%+ Career Win is a starting point.
Great video, although in future videos could you adjust these numbers with inflation so it matches it's current value? Would be interesting to see, cheers!
That's a great idea!
One of the better explanations of Benters work online - nice job
Thanks!
Really nice video
Do more videos of this kind
Thanks! I would love to!
incredible content thanks
Ken, have you ever attended the Sloan Analytics Conference in Boston? A professional bettor that appeared on the “Betting With an Edge.” Podcast and UA-cam channel mentioned it. He indicated it was a conference for the analytics folks from the American pro sports leagues.
Have not, but it is on my radar. Hoping to go this year!
Love this!
Thanks for watching Austin!
9:13 you misunderstood what Benter meant by "pool size limitations". In pari-mutuel wagering, the payoffs are calculated by how much money is wagered on each horse (or combination). What he meant is "how much can you wager into a pool before your own bets reduce the payoffs to a point of no longer being 'value' bets".
Thanks for the clarification!
There may also be limits put in place by the HKJC on the size of the bets you may place on a horse.
Great video Ken. I want to ask you for an opinion. I started an internship in a data virtualization project, but my goal Is to work as a data scientist or data analyst. What you think about It? Is It good starting point to enter in this area?
If you're interested in working as a Data Analyst, this guide would help you out: ua-cam.com/video/O187gv9rwzQ/v-deo.html&ab_channel=Devansh%3AMachineLearningMadeSimple
I think that is a great starting point!
Amazing story !
Thanks for watching!
Damn it. This was my backup plan if I ever decided to have my own "breaking bad" story. I've been beaten. Time to find a new market to do this in. Perhaps slap fighting. I didn't get my degrees in math and stats for nothing! R e e e e e e e.
Great video though, Ken. Hahaha, I love stories like this. I know ultimately at the end of the day you're only really supposed to make content that you enjoy and are proud of, but if that involves seeing more stuff like this, than I'm all for it!
I quite enjoy this type of content! Please make a slap fighting model lol
Hey ken, Can you check the 2nd example of Kelly criterion. It looks like, you have mistakenly put 1 in down. I am really getting confused when doing this math. Please, help..
Will it be payout=decimal odds-1 ?
Yep, made a mistake there. It is supposed to be divided by 2!
“I never dreamed about success. I worked for it.”
💪
one very important aspect that he acknowledges at the outset is that Happy Valley is a 'closed system' , meaning horses arrive on the island and stayed for the whole season (which was not that long, and they did not have races every night). This certainly would have allowed him to better establish an accurate hierarchy (ie horse A has placed higher than horse B every time) compared to the typical scenario of horses coming in, racing, and then off to another track (as happens in the US and UK). Regarding suitability to today, you forget that the UK has Exchanges which operate differently than the Tote system Benter had to bet under. The HK Jockey Club took out 18% off the top, and Benter still overcame that 'hole' and walked away a rich man. The UK exchanges take out 1% plus a commission of 3-5% depending on who you bet with. A much shallower hole to deal with.
Really interesting! I did not know that haha. I should have done a bit more research, thank you!
@@KenJee_ds your coverage of Benter was better than the other videos I've seen. The key line from his 'white paper' is this: "The easiest type of racing jurisdiction to collect data and develop a model for is one
with a closed population of horses, that is, one where horses from a single population race only
against each other at a limited number of tracks." It must have been an amazing environment, he had a bonafide professional operation with many employees, and they were betting huge amounts even by today's standards. To bet 5 figures on every race in the 90's is pretty amazing. Even the famous Blackjack team from the movie "21" wasnt betting that kind of money, except in rare situations that came up after hours of boring play.
Hi James, I see your comment is a few months old but maybe you will see this reply.
You point out that Benter (my better hero) was able to do what he did due to a closed system, that was spot on, well done! That and the the form data he was able to get plus obviously knowing how to utilize the data was really how he succeeded back then. This info will be lost on most people (as shown by all the other comments) including the author of the video (no disrespect intended to him/them) as they're not really into this subject more than something of interest. Maybe you bet or not, idk, but you seem to be the sort of person that could do well with sports bettering as you noticed that simple closed system example where many haven't. Also you point out that the UK has exchanges (with dedicated API services), which is definitely a blessing. The Tote in the UK also won't shut or restrict an account when winning.
Basically I wanted to tell you that a more open environment although much harder, is still very profitable. You seem to have a better grasp of horse racing than most. Ken mentions a lot of very interesting things and is a good story teller, but this is a youtube video and not an understanding of how to make money from horse racing. For me the explanation of future possibilities at the end was like listening to a list of reasons why something can't be done based on an evaluation from a limited understanding of the subject. Having said that, it's a good thing that people only hear the stories and not understand the potential as it would cause more competition for me as I bet on horses professionally. Although I won't say how the horses are picked other than I do use customized data and software, getting winners with odds in the teens and twenties is a regular thing for me and my best this year was 334/1 with second best @ 101/1. Missed a 600/1 by a photo-finish nose hair, lol. My biggest satisfaction are bets like only one horse @ odds in the 20s something in a big Irish race and having it win because that is f-ing hard and makes betting exciting. It's happened a few times, but it's hard to do.
Anyway, I haven't been betting these last few months as much as I would like as my time is taken up working on a bot for something else. Probably that's how I found this video as it involves ML. I've probably written to much already but will leave this with saying that what Bill Benter did can be altered to suit the challenges of the wider system. Some say that whatever data is available is available for all and therefore one can't get value or an edge. That is rubbish and for people who can't or won't take the time to think outside the box or even do their homework. That one point separates the get rich quick schemers from people like me who put in more hours/years of trial and error learning how to bet the odds. That is my material goal in life, to beat the odds at whatever I do, so far it's working.
Take care mate, may the odds always be in your favour. :)
@@JohnDoe-cf8jz yes, he's my hero too. Interesting that you apparently bet some very high odds (along with I assume lower odds, perhaps the majority). In my case its all low odds, determined by an additional 4 factors. I just cant take long losing streaks, they are too psychologically negative for me to handle ( and of course the higher the odds the higher likelihood of losing ). If the past holds for the future, it should work out well, but as every financial ad in the US has to disclose "do not base expectations of future results on past performance". So any patterns I may have found could gradually revert to a mean of 1.0 ROI pre-commission. Fortunately at low odds there should be less volatility and a complete reversion to 1.0 seems unlikely. I'd love to know what odds range (if any specifically) Benter focused on. Certainly it's all made possible by Betfair and the clones, I wouldnt be trying this at US tote board odds, not with an overround of 1.20. The exchanges make it nearly a zero sum game, eminently fair. 99% of people just dont get the difference between that payout vs bookies or tote. I recently read an interesting story about dice in ancient Rome, they apparently were irregularly shaped, meaning each face was not a 5/1 proposition. It's believed that either the gamblers of that time just did not understand the concept of probability, or felt strongly that 'fate' was the over-riding factor, and you were either 'destined' to win or lose. Maybe things havent changed that much.
@@jamesrav Hi James, that's a good reply, you seem to have some method with "patterns" which sounds interesting. I use two broadly different methods to pick races and one of them is simply statistical as in I know that a certain type of race performs a certain way more often than not. Those sort of selections are always at low odds and can be auto bet with a bot. For the big prices... like Benter I make my own "tissue" which I think is called handicapping in the US. What odds Benter would have focused on is a good question for sure, I have no idea either. What I do is ignore the market odds and look for whichever runner I think will do well according to my own criteria. Sometimes those are low odds, sometimes big odds. I do get long losing streaks sometimes but I also get good winning streaks. Getting a 50/1 win one races then 16/1 the next is very satisfying. Not often, but there can also be 10 losers in a row. I usually flat bet only so don't incur big monetary losses when losing. Lastly, interesting point about the dice, makes me wonder why also.
At least in the US, data availability isn't that great unless you're willing to spend significant amounts of money. Also, most of the race tracks in the US not only don't ban the computer model betters, they actually make rakeback deals with them to give them a big advantage over the everyday better.
I didn't know this. Very interesting!
they receive a "rebate" of nearly 10%....the average "takeout" (money NOT returned) per race is on average 20%.
Correct, I was part of a syndicate that was rewarded up to 25% on exotic wagers like the superfecta here in the USA and direct computer linkups to the host tracks tote system.
This was in 2004 and still goes on to this day.
Hence, the bigger the whale, the bigger the “rebate”.
Much easier to overcome the tracks takeout obviously.
In the end, Hong Kong did the same for Woods and Benter’s teams.
@@garettoverstreet I had seen another person post this observation, it certainly does help explain their success. But I've only seen small rebate amounts in the US, 2 or 3%, even for fairly large bet amounts. So unless HK gave them extraordinary rebates on losses (6-8%), what they did is astounding. But clearly reducing the track take of 18% down to 12% is a huge help. It begins to make things viable. There's one greyhound track that has an 11% take, I keep telling my greyhound betting friend he should be focusing on that one and skipping the 18-20% tracks. He doesn't get it.
Great content! Very helpful insights. Just wondering though if same model could be followed for state horse racing lotteries?
I honestly don't know enough about them to be able to answer that question, sorry!
lol I was just reading the Bloomberg article this week.
It was a good video! Hopefully I differentiated!
man people back in the days are amazing. how do they even collect data back then? all on paper?
pretty much lol
great content!
saudações from brazil o/
Thanks for watching!!
Ken, we need to make a ML model to predict Papaya futures!!
#PapaysIncToTheMoon 🍈🚀
Gotta up my papaya game!
Was he using software for this?
So much big brain talk, finance is one of the sector I really want to excel in professionaly and personally, nonetheless great video!
lol thanks for watchin!
This is awesome. Honestly, I've always wanted to try sports betting lol (but to anyone reading, I didn't say that :P)
Really glad you enjoyed it Kylie!
Bookies have full DS teams these days...
I tried my hand at poker for about a year. I am not good at it. I think I'll try writing a Q-learning trading bot to make my side money instead.
my goal is to become a professional gambler. I told my guidance counselor I wanted to be one and she said that there is no such thing as being a pro gambler. I showed her this documentary on bill benter but she still thinks pro gambling isnt a good career. How can I get started being a pro gambler?
I’m not a pro gambler by any means haha. From talking to people, it seems like you really need a solid bankroll to do this effectively. If you are serious about going down that route, I recommend getting a analytics job in the gambling space (at a sports book or casino). That will teach you how the industry / system works, and it will give you the skills to capitalize on it. I encourage technical skills, like coding or statistics, these are really important for getting an edge and maintaining it
If you need to ask how to become one, you will never become one, because you are not an independent thinker.
Your IQ is also probably not high enough.
pretty much impossible, its not even a zero sum game and you are competing against everyone else. My advice, if I could do everything over, would be save,save,save and then when a great opportunity arises, deploy your money. The Covid pandemic and the meme stocks and easy $$ led to an incredible opportunity. Even idiots were making $$ in the penny market just because there was so much $$ sloshing around. Of course you had to know when to call it quits. But sports betting is a nearly impossible way to make $$.
@@jamesrav
It is possible if your IQ is high enough.
But the stock market if much easier thanks to two factors :
- long term economic growth of the economy and as a consequence long term increase of stock prices
- smaller commission on your bets/investments
@@miyamotomasao3636 totally agree on that, I have a friend who typically loses in the Penny stock market who made a life-changing amount in a few months during the pandemic. He said during a stretch of time everyone was making money , which seems impossible, yet with all the stimulus money being given away and the Fed essentially handing out $$ like candy, it was possible. He keeps saying he needs another pandemic, so it is possible to make a lot at the cost of several million people dying haha. Disasters are always an opportunity. But Stocks, bonds, even options and futures, are pretty boring compared to sports betting. Not much happens day to day in general. A huge part of betting of course is the adrenalin rush if you win. But 99% or more lose at sports betting, because you can't beat The Crowd (+ the vig or track take) in the long run.
Daily horse racing tips, avg +20U / month
Wow he lived out my dreams 😂
Lol
Nobody has a "Gambling Problem", only a "LOSING $% PROBLEM"!!!
I think Is hard to beat the Booker
I would agree!
I can't wait to make a video in 5 years called "How @Ken Jee Made $1,000,000 by Playing Zed Run"
Lol that would be something!
You talk too fast. Please slow down.
Right speed with the right knowledge
He did all this in HK at the time There were only 2 tracks , now there are 3 All horses at the time and possibly still now ? , were domestic to HK as in no horse travelled in or out , much easier having the exact same horses running on the same 2 tracks once/ twice a week ,Data wise .He also threw stupid money at every race trying to hit a trifecta of trifectas ,as in the correct position of 123 horses in 3 separate races .My point is you could not replicate this anywhere else ,HK has a small pool of horses and no horses in or out of the country like say the UK . If I must take the shine off it , it pretty much had nothing to do with the actual horses IMO it was the combination of numbers and the sheer volume of tickets that got him his wins . Just like me Playing max entries in a DFS contest week in week out your chances increase the more you play .