The strategy this paper seems to be describing is well known as “chasing steam”. You keep an eye on the odds at a “sharp” book, one that’s good at tracking which clients of theirs often win and adjust their odds accordingly, then find books that are slower to adjust their odds. There’s a long tail of “pay per head” bookies, basically your classic street corner bookie but with a SaaS for taking bets, and they’re often the slowest to adjust.
The problem is that identifying this is very easy for a sports book. Basically “show me all accounts that consistently make bets right before we adjust the odds” and then they ban/limit those accounts. In practice the heavy lifting for winning sports bettors is building a network of “outs” or players that will place bets on their behalf. The idea is that you have a recreational bettor (i.e. bets for fun and usually loses) that puts down a bet on your behalf with your money, and if it wins they get to keep a slice. This masks the sharp bets with their personal bad bets.
Really the strategy to beat the odds is one of the easiest parts of the equation. It’s good old fashion people management that’s the biggest part of what makes it work.
When I moved to Las Vegas in the late 90's, I became friends with someone in my new circle who was from Brooklyn, NY and a pretty active sports better. I mean this wasn't the recreational type doing for-fun $20 parlays on Sunday, he was routinely doing $500 and $1k bets ("dimes" he called them). I had been to Vegas on trips before and remember bringing $300 for gambling for a weekend and thinking that was a lot. I was in utter shock at how much he would be riding on a typical NCAA Football Saturday and NFL Sunday. It was one of my first true experiences in my life of what hardcore gambling was and a side I have never seen or known.
Anyway, I remember very clearly how serious his betting was and how he solely looked to find any edges he could. It wasn't about handicapping a game, he was a line sharp. Remember this was a time where Internet was still dial-up and wagering off-shore was still very early days. Many times he would be calling old-fashioned bookmakers back home. He was paying for a service called Don Best and at the time it was pretty expensive but was able to get line-moves almost in real time. He was "Chasing Steam" as you called it. Watch for big line moves in Best and place bets where he could that haven't caught it yet. He was pretty successful and quit his full time job while buying a house, cars, etc. Not flashy but just a living.
Eventually, as all things gambling, it started to turn bad for him. He was getting accounts suspended and getting listed as a sharp. He would call customer support playing dumb and I remember one guy at an offshore book on speaker-phone flat out told him basically no longer want his business as every time he places a bet with them the lines were moving shortly after in his favor and wanted to know what he was using.
I remember his downfall being pretty fast after that. It started with just getting blacklisted at some and the few reputable offshore books left wouldn't keep him long when they discovered he was a steam chaser. He tried changing phone numbers and fake ID's and they caught onto that. Eventually any book that did take his action he started to hear Don Best line moves in the background on the phone and they would tell him to "hold on something is happening" then just give him the new adjusted line. The edge was gone and he started to bet with shadier old-school bookies that didn't have the technology, but they just didn't pay him anyway.
He started to see the writing on the wall and the bill pressures started to effect him as he started to just push into trying to handicap and pick winners based on gut and dabble into some line making software. At one point he was in for a very large sum after a bad Sunday, something like $70k he owed. He struggled for a good 5 years after that, divorced and back to a 9-5. Last I seen him was right before Covid and he was out of that world not having placed a bet in a long while he said and was now into options trading.
I doubt betting line moves is even a thing anymore. Information just moves so fast now and books have evolved way beyond. Plus with how big sports wagering in in the US, lines probably only move now because of large public betting waves or injuries.
> he was out of that world not having placed a bet in a long while he said and was now into options trading.
There are many ways of trading options as part of an investment strategy. But it also possible to trade options in a way that is functionally gambling.
> I doubt betting line moves is even a thing anymore
It definitely is and it is lucrative if you go down the right rabbit holes. When you are able to be disciplined and completely ignore any sports fandoms you might have, you can have many "dimes" riding at a time and not even break a sweat
Thank you for sharing this. The gambling world has always been a fascination of mine and I would’ve loved to have been a bit older during the heyday of offshore bookies.
Such a great story. Something like this could have only been pulled off in that era.
I'm not sure if you're aware of this, but this is literally how the market becomes more efficient: by allowing participants with more information and reducing the arbitrage opportunities.
The bookmakers don't set the odds based on their own expectations, they set the odds compared to how the market bets and then add their own margin on top (the "vig"). Just like a stockbroker would.
In sports betting, compared to financial markets, the general reason why people aren't wildly successful is that if they have too much of an edge the bookmakers find some way to limit their succes (limited bet sizes, banning, etc)
A caveat here with the stock market comparison, although there are some betting exchanges like Matchbook, most betting is done with fixed odds. That is, the book isn’t letting the market set the odds in the sense of trying to balance their action and add vig, they’re trying to get the most “correct” odds. Often this involves tracking the sharp bettors within the market, but not the market as a whole.
The sharp odds is another factor that makes it impossible to be wildly successful (in addition to your callout on banning/limiting). Like NFL lines as an extreme example are so sharp that I don’t think anyone could reliably beat them.
Prediction markets like Polymarket solve this by not having bookies. Just bet against other players.
Why bet against a house who takes an enormous cut even when you win when you could bet in an efficient market and keep 100% of your winnings?
The reason is fact that bookies are odds makers while bettors are odds takers. Theoretically you can't win if a market is perfectly efficient. When market depth can be in the hundreds for the less popular things like e-sports, someone betting a few hundred will make the odds converge to true probability.
Assume bettor's model and the bookie's model represent the true probability of an event with added random noise, neither model is perfect but one is not better than the other. For any bet, the true probability will on average be between the bettor and the bookie's implied probability. However, the bettor bets only when their model shows a higher probability than the bookmaker's, thus they gain positive value over time. The bookie is disadvantaged because they can't use the bettor's information to set odds, while the bettor can use the bookmaker's odds and make a decision to bet. Remember here that they are running the same model, the bettor doesn't even have an advantage. Because of this, there is no one on Polymarket running a model offering odds on all possible markets (acting like a bookmaker), it will just lose money. A bookie probably has to offer all odds on all games, in order to compete with other bookmakers. The idea is from William Benter's paper[1], the guy who made a 1bn betting on horses. The books are probably well aware of this, which is why they will ban anyone who wins consistently with a good enough model, they can only ban the person or stop making odds on that market which obviously is not a good idea
[1] https://gwern.net/doc/statistics/decision/1994-benter.pdf
Why make bets as the house when you could just take a cut risk free?
so how do you decide where the odds should start? I think they still need a bookmaker for this, I think. What happens when the first person bets? I have never read up on how poly market works but I believe they must act as an Automated Market Maker as they provide instant liquidity.
you don't really care... you just adjust the odds you are offering to balance the books with each bet and until you get good or get a nut you put limits on the bet sizes so it's easier to bin pack the bets. This was super common with sports bookies pre just holding the vegas line.
Get a huge bet on paul that you don't have offset? offer better odds on tyson till it balances. as a bookie your goal at the end of the day is to have the most total amount of bets on the book (say 1M in total wagers) where the money placed at the odds completely evens out so that at the end of the day you take 10% of 1 million for providing the market (the vig) and pay one set of betters with the other's money. In fact as the house you're just building creating your own arbitrage by inventing currency exchange rates.
Back in the 1980s I did some work on a computerised tabulation system for a large race course: my part was programming EPROMS for the displays, but I learned a lot about the system from that time.
Basically, the house takes their cut from all the bets. The house cannot lose. Then they share whatever's left amongst the winners.
The returns for a $1 bet on a horse are calculated in real time and displayed on the screens. THESE ARE NOT THE FINAL ODDS, they change as bets are placed right up until the race starts, which is the starting price (and where the term SP bookmakers comes from).
Contrast this with bookmakers at the track who offer fixed odds when the bets are placed. They stand a chance of losing if they get their numbers wrong.
You're not betting against the house. The house always takes opposing bets so that it's fully hedged. The players are betting against each other.
Haha whenever people tell me they own NVDA or whatever I tell them that the four hundred basis points more than pays for free drinks.
Can you please explain a bit more?
Is "NVDA" the NVIDIA ticker?
What do the 400 basis points mean?
Who is paying for, and who is getting free drinks?
If you play perfect basic strategy on a six card shoe at the Wynn your odds are a matter of some dispute but a reasonable consensus figure might be on the order of 1% to the house, or a hundred basis points.
Only professionals can maintain perfect basic 21 for any serious stretch, a tourist who keeps their wits is probably at something like a 4% disadvantage, or 400 bps.
I was making fun of people who sit down at a table with Citadel as their counterparty and don’t consider it gambling.
The bid ask spread is not even 10 basispoints for your regular stocks out there. In fact for NVDA right now it's less than 1 basispoint.
Anyhow the comparison doesnt make a whole lot of sense
There are a lot of different forms of Fama EMH, but even under very weak versions of it an uninformed speculator has the uniform geometric Brownian motion of holding an equity pretty much pegged at 0.5, expected value is par in the absence of an arbitrage opportunity. This is eliding fees and other friction, and eliding the market microstructure in a stand price-time precedence setting.
People (uninformed speculators) routinely think they have alpha by virtue of knowledge that was priced into a Citadel regression days, weeks, or months before they obtained it.
Uninformed speculators are almost by definition gamblers, and it’s pretty much a set cover between gamblers and arbitrageurs.
I wasn’t making fun of arbitrageurs.
P.S. Equities on Island and ARCA are decimalized on the consolidated tape, lit venues can’t have a spread less than one penny.
Anyone holding NVDA or any tech stock for more than 10 years would have had massively beat the market.
I assume by speculators short term holders and traders. Sure these people wont have alpha over entities that take it seriously.
Anyway I don't think EMH holds at most a very very weak version of it. It's quite obvious that smallcaps, micro- and nanocaps are from time to time mispriced. The big traders are also not around to mess with those.
> Anyone holding NVDA or any tech stock for more than 10 years would have had massively beat the market.
Right, so anyone with a time machine can beat the market? That's not how this works.
Can you confidently point to a stock today that will have out-earned the market by next decade? That is the question facing investors and it's a much harder one.
A single stock no (usually 1 in 3 perform disappointingly) but a collection of stocks sure. They are all companies you never heard of and has worked out just fine for me.
Usually opportunities come from time to time. For example us met coal mines during covid. ESG boomed and funds dropping them, banks stopped giving new loans, supply chain messed up, priced for bankruptcy. Plenty of people treated met coal the same as thermal coal. Clearly steel is was not going away so a couple hand picked west virginian and alabama met coal mines did wonders. I spend a lot of time researching calling emailing companies there is no free lunch. Its fun and makes me very good money.
Anyway Im not going to peddle my small caps here. Indexing is still the best strategy in general.
I'm surprised you're getting downvoted. You're absolutely right -- the expectation less fees from holding a stock a short period is equal to the risk-free interest rate.
Over longer horizons it gets more complicated because compounding, but it's not obviously in favour of the speculator, even though it has historically turned out that way.
I’ve been told that I tend to “TED Talk” / lecture, and on re-reading I can see why someone might get that vibe.
It isn’t my intention to condescend, I know a bit about this stuff and try to make informative comments when the opportunity arises.
Thank you for having my back on the substantial accuracy of the comment.
Post/post script.
It is in fact a time when HODL the Magnificent Seven has been a great trade for years: it hasn’t been this easy to turn money into more money since the 1920s.
But that’s a combination of sanctioned monopoly fraud and wild levels of stimulus: in the last decade we basically transferred half of all wealth from working people to Boomer 401ks, and the New Wall St guys got a pretty hefty cut.
NVIDIA has some very smart people, but they don’t fab EUV wafers and they don’t do whatever it is where it’s ok for a dorky guy in a leather jacket to sign a girl’s breasts.
The price of the stock is “supported”.
> NVIDIA has some very smart people, but they don’t fab EUV wafers and they don’t do whatever it is where it’s ok for a dorky guy in a leather jacket to sign a girl’s breasts.
Why does it matter whether NVDA fabricates wafers or not? It's not their business model.
The dorky guy sells something all of the mag7 and more want desperately.
> The price of the stock is “supported”.
Supported by gigantic growing cashflows. Did you ever valuate/analyze a business? You know people don't put random numbers on the stock price (looking at longer timewindows obviously).
A basis point is 1% of 1%, so 400bps is 4%.
It's a joke. Claude will explain it to you.
Are prediction markets essentially zero sum where every outcome is fully covered by all the bets?
Minus fees I’m guessing.
Yep. For the ones I've looked into, orders match when they sum to $1. Eg, I bid $0.75 on yes and you bid $0.25 on no. The contracts are very carefully written such that they are mutually exclusive, only one of us gets $1 when the contract resolves. And like you suggest, you pay a fee on the way out.
If this is the case, it kills me that online poker is still illegal federally. Zero sum skill game minus fees.
Is it technically possible to encode a whole game of hold’em using one of these markets?
The issue here is that "skill game" is something that lawyers would have to prove to judges.
That's called parimutuel being and the broker just takes a vig (vigorish). It's analogous to agency based trading in finance.
The broker can also set the odds and act as a market maker, like principal based trading in finance.
I didn't understand, how would you use the information given by the bookie ( the prices) to find inefficiencies in that same information. Unless the bookie made arithmetical mistakes?
From the abstract it sounds like maybe they are finding differences across different sportsbooks, i.e. arbitrage?
No, not arbitrage. The "insight" was that the odds offered by bookies are remarkably accurate predictors of the outcome. Averaged across a number of bookies the small variations cancel out. In other words, you can't beat the house.
However, there are two additional details to consider. 1: Bookies may incorporate information at different rates and delays, so every now and then you can have a couple of outfits who are behind the trend, and where the odds for a given outcome are better than the industry consensus. And 2: sometimes a bookie may choose to offer over-the-odds opportunities to invite bets on one side of the market to balance out their net exposure for a lopsided game.
If we ignore the bookie's commission structure, let's say that a game has seen lopsided betting activity. Out of 1 million as placed bets 80% have been placed for the home team, so their exposure from home win is at least 800k. If they calculate that the away team has 13% probability of winning, the bookies may want to offer odds above that (ie. the payout of a bet is above than its fair value). That will invite some sharper bettors to put their money on the away team win, and the bookie gets more money on both sides of the book.
In other words, they try to reduce their net exposure to the game. In an ideal world for a bookie, the bets placed across all three outcomes cancel each other out, and over time they make their money from the commission.
In the real world, the bookies will identify sharp clients and prevent them from playing. They will also use the winning clients' trades as a pricing signal for themselves - and will hedge out accordingly on the exchanges when the same mispricing opportunity presents itself.
I don't think it counts as arbitrage, strictly speaking, because from what I can tell, the authors are not making offsetting bets with different bookies to expose themselves only to the difference in odds.
They are, however, finding opportunities where one bookie has odds much longer than others for the same outcome, and betting on that. This means they expose themselves both to the difference in odds and the outcome of the event, but opportunities are plentiful enough that on average it is possible to show a reliable profit in a few months, apparently.
What surprises me about this is that some bookies still offer odds out of line with others. The article mentions that they might do this to counterbalance the risk of large amounts of bets on another outcome, which makes sense. But! I was under the impression bookies placed bets with each other to even out those risks, specifically to avoid creating arb opportunities.
I'd say it's a type of arbitrage.
The typical arbitrage in bookies (scalping I think they call it), hedges one bet against the other.
But in the case of the paper, as I understand it, they just take the bet unhedged for some reason. Which unlike most other forms of arbitrage is sustainable.
If you were buying commodities you would also need to sell the commodities, but since these pseudo-securities coalesce into money, they don't need to make the inverse trade to sustain their alpha-seeking fund.
With that difference aside, I'd say it shares many of the same properties of arbitraging, I don't know how translateable it would be to stocks, but many of those assets coalesce into cash as well, especially fixed income (at a much slower rate), and options (albeit in more complex manners).
I imagine that with the amount of new currencies that exist, a big part of the job is going to be moving currency of one form into another and essentially forming a new circuit, for example buying crypto with fiat to bet in a crypto sportsbook, and the selling that crypto to buy more fiat.
In essence there's also the risk of becoming a money launderer doing this, whether trading stocks, currencies or betting, it's a form of arbitrage and you are getting money from ???.
> I was under the impression bookies placed bets with each other to even out those risks, specifically to avoid creating arb opportunities.
Doesn't that just mean they don't lose money? You can still make money as a punter in that scenario?
My intuition is that it would mean these underpriced outcomes would go away.
Not quite arbitrage; sometimes bookies adjust prices depending on how bettors are placing their bets to avoid over exposure, etc. When this happens the price is no longer reflective of the estimated probability and that's where our savvy players come in to try and take advantage of that..
Interestingly, the gambling platforms restricted them after they got too successful.
This is from 2017, so I am wondering if the strategy would still work (and how much faster they would be blocked).
In a recent interview with Nate Silver, he talks about how it’s basically impossible to make sustained edge from online sports books. They just ban you if you consistently make large, winning bets.
Anyone who makes a living off of sports betting is paying other people to put the bets down for them. Even the in person casinos will ban you if you win too much.
How has this survived legally? Only non-winners are allowed to gamble?
Yeah it’s extremely sketchy but legal. Sports betting is treated as a game offered on the casino’s terms and they get to choose who and for how much they’ll play with. You can see it in the financials of the big companies. Where the theoretical win for the casino on most sports bets is around 5%, DraftKings and the others are regularly hitting 3x that much. They ban/limit winners and push everyone else toward bad bets (e.g. same game parlays).
This has always been the case throughout history. The difficult part of winning big or consistently is not doing the probability judgments or expected value computations -- it's getting the counterparty to pay you.
The actual long-term winners have to win small and inconspicuously.
> This is from 2017, so I am wondering if the strategy would still work
Knowing the gambling industry, I bet it doesn’t work now
Knowing the gambling industry, it likely does still work, they just ban you if you use it.
The only gambling strategies that work are the ones nobody tells you about.
You can tell people the 30k ft version but you wont get the lines of code, or the smart operators needed to keep the system running for that matter. Data science, dev, ops etc.
Given that the sports-betting population is not entirely rational, and perhaps in predictable ways, would it be surprising if the odds that maximize profit in the real world differ from those which would be optimal given only rational and informed punters?
Here in Australia the average punter on horse-racing is normally competing against his fellow rational and informed punters.
But every November we have a huge Melbourne Cup race where millions of people make their selection because they like the jockeys' colours or the sound of the horse's name. So the serious punters are competing against uninformed punters and on that day they are in a much better position to make some $$$.
> We provide a detailed description of our betting experience to illustrate how the sports gambling industry compensates these market inefficiencies with discriminatory practices against successful clients.
Correct. I have written that code in a past life. Some clients get rate limited. Keep their bets as a nudge to check your odds. Poor big betters (these days called whales) get taken out to dinner.
Also worked on the other side and you can avoid being shutdown by using 2 sided markets. Pools, betfair, crypto, for example.
>> While several betting strategies have been proposed to beat bookmakers, from expert prediction models and arbitrage strategies to odds bias exploitation, their returns have been inconsistent and it remains to be shown that a betting strategy can outperform the online sports betting market.
This is the most academic sentence I’ve ever read. There’s always been an industry of professional gamblers with various successful strategies. Some have become billionaires.
> There’s always been an industry of professional gamblers with various successful strategies. Some have become billionaires.
Be the house.
[flagged]
The bio of this account is lorem Ipsum text and this comment is a copy paste of the first 2 sentences of the paper abstract. Feels like this could be spam.