Inside the Shadowy World of High-Speed Tennis Betting
May 30, 2014 6:36 AM   Subscribe

Getting relevant data first is a gambling advantage. Thus the controversial practice of courtsiding.
posted by Chrysostom (28 comments total) 4 users marked this as a favorite

posted by JPD at 6:41 AM on May 30 [3 favorites]

Fascinating! I love tennis and I love gambling (or rather I love finding an edge where there isn't supposed to be one) and I had no idea.
posted by 256 at 6:55 AM on May 30

14 paragraphs in before he bothers to explain why this matters, and I'm still not clear whether it's any more profitable than just watching a major tournament on TV and entering point data directly.
posted by aaronetc at 7:17 AM on May 30 [2 favorites]

whether it's any more profitable than just watching a major tournament on TV and entering point data directly.
Yes, there's a nontrivial delay on "live" TV.
posted by katrielalex at 8:04 AM on May 30 [1 favorite]

Given how simple tennis scoring is I'm amazed that anyone ever gets caught doing this. It would be trivial to rig something so that you can transmit the scores simply by pressing buttons without looking at them.
posted by yoink at 8:22 AM on May 30

Well in the article they test it out and the writer says he can't tell that the guy is pressing any buttons. I guess it's one of those things you have to do perfectly 100% of the time and they're looking for any slip ups.
posted by Carillon at 8:36 AM on May 30

aaronetc: It's worth noting that this is from 548, which is an experiment in data-driven news and news about data. This is solidly in the latter category, basically detailing the emergence of high frequency trading techniques (as JPD notes) in gambling.

The premise here, and in HFT, is that if you have an adversarial system in which one party has more information, then they should be able to leverage that information to get ahead.

If we were talking, say, poker, we would have no doubt that knowing what the next card is a 'turn' before the other player knows should produce an advantage. Of course, poker is a 'discrete time' game (ie, turn based) so small time advantages don't really have an effect on the outcome: there's plenty of time to ponder during a turn. Betting and HFT are both 'continuous time' games, where you can make a move at any time: in this case, knowing something a little bit in advance of the other players gives you an advantage that you can exploit.

It sounds like the betting company is running a complicated model (possibly based on machine learning algorithms) to calculate the probabilities of a given match's outcome. Simultaneously, you have another model being run by the betting market, based who's bet where, creating prices on bets proportional to people's overall feeling of who's going to win the match.

When a new point is scored, these probabilities get updated and change the price of making a bet. If a match has a 50% chance of being won by player A, but then a late point shifts that probability to 75%, it's advantageous to place bets on player A in the window before the market updates the odds. You're effectively getting a discount by being faster than the market.

Note that there's two ways to make money here: First, you want to build a model which is at least as good as the market. If you can build a model that's actually better than the market, you stand to make a shit-ton of money. This is what 'sabermetrics' does for baseball, actually. The article is focusing less on this and more on how to exploit time advantage.
posted by kaibutsu at 8:37 AM on May 30 [4 favorites]

They're probably also doing some visual scans of people in the audience, and getting to know the regulars...
posted by kaibutsu at 8:37 AM on May 30

So the entire premise of this is you get a few seconds' edge with early information about each tennis point before the official sources have it. That's just nuts. In particular, why are you allowed to bet on a tennis game during the game itself? That seems like a particularly degenerate form of gambling.
posted by Nelson at 8:50 AM on May 30 [1 favorite]

if you underwrite something to be like a fair coin toss from an expected value perspective and you just make a vig, the more coin tosses you can make a market in, the better.
posted by JPD at 8:52 AM on May 30 [3 favorites]

I feel like you can add "Inside the Shadowy World of..." before anything these days.
posted by mafted jacksie at 8:53 AM on May 30 [1 favorite]

an adversarial system in which one party has more information

As it relates to HFT, speed, coupled with sophisticated algorithms, allow one side to take advantage of another. The one side, being hedge funds, have the premium (shortest length) data connections and high-throughput data connections, not to mention backend relationships (liquidity, tranche priority, etc), that allow them to complete trades in orders of magnitudes beyond what some slob could sitting there in their underwear plinking away on their cable modem-powered laptop.

Couple that with staffed levels of highly educated people and yes, one party has more information, but mainly due to the fact that the other party has none.
posted by jsavimbi at 8:54 AM on May 30

I totally agree, and don't think HFT is in any way a good thing. The lesson, I think, is that if we want humans to be competitive players in a system, we have to design systems with that in mind. One way to go about it could be to take these continuous time games and convert them into discrete time (like poker). These betting systems provide a lower-stakes model for trying things out than the full financial system, which is kinda great.

Diplomacy (the board game) offers a pretty great idea for how one might do this. In Diplomacy, everyone makes negotiations and then secretly submits orders for their armies. The orders are then evaluated simultaneously according to some rules at the end of each turn.

To get the machines out of your tennis betting pool, you could aggregate lots of orders (bets) and then simultaneously evaluate them. You then figure out the 'market' odds once you have lots of orders, and then setting the prices for the bets retroactively to match the market mood. If your time windows are matched to the duration of a point, then submitting orders anywhere in that point would have no effect on the cost of your bet, thereby nullifying the discount for early information. The downside is that you don't necessarily know what you'll be paying to bet at the time you place the bet.
posted by kaibutsu at 9:30 AM on May 30

It always astonishes me the lengths to which people will go to make money without adding any value to anything. HFT and its ilk make me sick.
posted by Ickster at 9:33 AM on May 30 [2 favorites]

Eh. Very few people in relative terms actually add value to anything. And its not like they actually go to great lengths. Most of them trained to do something else and end up doing that instead.

A know a guy who is very successful at it who will flat out tell you he moved into the field because he didn't think he'd make it as an academic.
posted by JPD at 9:39 AM on May 30 [3 favorites]

I mean and what makes you sick about it? They make a lot of money?
posted by JPD at 9:39 AM on May 30 [2 favorites]

if we want humans to be competitive players in a system

That would be uncompetitive. How would I convert people to my model if I offered no advantage over the competition?

Online poker is one of the worst activities a person can engage in, imho. I've never met a professional gambler, an addict, who was mentally and financially sound. They're mechanical turks handing their money over to whomever the current receiver is.
posted by jsavimbi at 10:12 AM on May 30

JPD: “I mean and what makes you sick about it? They make a lot of money?”
It seems like cheating.
posted by ob1quixote at 10:15 AM on May 30

It seems like cheating.

You and Doyle Lonnegan can sit over there. I'll be having drinks with Mr. Hooker and Mr. Gondorff.
posted by aureliobuendia at 11:23 AM on May 30 [1 favorite]

I've got a very basic question, knowing nothing of sports betting. This only seems to be sensible if they accept bets while the game is still going on. I had assumed that once a game starts, betting stops. Which one am I wrong about?
posted by benito.strauss at 11:45 AM on May 30

in this case the game is each point.
posted by JPD at 11:49 AM on May 30 [1 favorite]

It seems like cheating.

fundamentally a misunderstanding of what they do. Most of it is market making.
posted by JPD at 11:50 AM on May 30 [2 favorites]

I made a living as a poker player for a while and I absolutely struggled with the "not adding value" issue. Every other job I've held in my life, even if I didn't think what I was doing was especially worthwhile, I've at least been contributing to some constructive endeavour.

As a poker player, I was just taking money from people in exchange for giving them the possibility to take my money. For a while I tried to justify it as an entertainment service, and certainly in some cases it was, particularly on friday nights in Atlantic city when the NYC professional crowd shows up. They all hope to win, but many expect to lose and are paying for the experience of playing at the table.

But mostly, I was winning by being a little better than the people who thought they were a little better than me. And when I would lose, I certainly didn't feel that I was paying the other players a wage for entertaining me. I wasn't being entertained; I was working.

I still enjoy gambling, and now that it is not my career I feel a lot less conflicted about it. But I totally get where people who think it's a dishonorable way to spend your life are coming from.
posted by 256 at 12:06 PM on May 30 [2 favorites]

Yes, there's a nontrivial delay on "live" TV.

But there's also a nontrivial, and less predictable, delay on intercontinental Internet communication, plus the expense of sending someone (someones presumably, since there are many simultaneous matches in the early rounds) to Melbourne for two weeks.
posted by aaronetc at 1:21 PM on May 30 [1 favorite]

JPD: “It seems like cheating.

fundamentally a misunderstanding of what they do. Most of it is market making.”
Allow me to restate: Frontrunning seems like cheating.
posted by ob1quixote at 1:35 PM on May 30

Market making often is front running
posted by JPD at 2:36 PM on May 30

Very few people in relative terms actually add value to anything

Teachers, doctors, nurses, therapists, lawyers, plumbers, carpenters, bakers, farmers, postal workers?


knowing about just one more point than your betting counterpart often provides only a minuscule edge

I read the article and the comments, and maybe I'm slow, but can someone provide a real-world example of how someone uses his/her knowledge that the game is 40-15 vs. the rest of the world at 30-15 to make enough money to justify the expenses (tickets/labor)? I am totally lost. I haven't bet on sports in a while, but are we talking mid-game wagers or is it a more complicated projection.

I guess I was hoping for someone to explain that how that "minuscule edge" turns into profit (i.e. real numbers.) Maybe I missed it. If it's in TFA, just tell me. ;)
posted by mrgrimm at 3:49 PM on May 30

JPD: “Market making often is front running”
You are, of course, much more versed in these matters than I am; that's why there's a star next to your name on my screen and I look for your comments in finance and market threads.

Still, to my eye there's legitimate front running/market making based on anticipation and prediction and then there's being on the route over the public network to the next exchange, seeing an order cross the wire, and beating it there over your own private connection so you can buy before the order arrives and then fill it yourself, skimming a penny in the process.
posted by ob1quixote at 4:53 PM on May 30

« Older Racism Award: $2B   |   Of Course Newer »

This thread has been archived and is closed to new comments