Feeds

back to article Number-crunching quant cooks up gambling machine, promises untold RICHES

Have you ever wanted to lay a bet down on the Las Vegas sports books? Or take a bundle of money from the suckers in your office's Final Four betting pool? Is it possible to win sports bets consistently over the long haul? According to the guys at PredictionMachine.com, the answer is yes. Their secret weapon? Numbers, baby – it’s …

COMMENTS

This topic is closed for new posts.
Anonymous Coward

Yanks

All yank stuff by the looks of it. No use for the nags.

1
0
Bronze badge

Re: Yanks

Not even the nag's head?

0
0
Anonymous Coward

hhhmmmm

"Over the last three years, the company's founders claim, Predictalator has scored wins on better than 70 per cent of their straight-up bets"

If it really did that well, why are they telling anyone? As soon as the bookies get wind of its use, they will not take your bets. And if lots of people did use it, the bookies would go out of business. Or else the bookies would use it themselves and severely nobble the odds for the forecast winners. Either way, the window of opportunity is very small. I detect a whiff of BS.

6
0

Re: hhhmmmm

Aaaah, this is 'football'. The one that doesn't involve feet.

5
0

Re: hhhmmmm

Winning 70% of the straight up (non points spread) bet is actually quite easy, bet on the favourite, but the odds will be shit - and you'll lose overall.

What would be more interesting would be to win 70% of the time on a straight up, non favourite, the odds would be significantly better (ranging between 6:5 and 4:1).

I smell BS too, as it's either useless (70% favourite) or they want to give money away (70% non favourite). It is possible of course that there's some smarter logic in there, and to be fair, why not? there's lots of clever people out there, but what certain is that it's not free money, and systems that do work (laying bets to cover all odds against multiple bookies/doubling etc.) have severe limitations (like a need for unlimited funds) or even just implementing rules "no doubling past 5 allowed".

I'm not saying that this system is fake, but if I wanted to create a system, I'd have some logic which won most of the time (regardless of odds), and if the punter won using my system, I get 5% of the win, if they lose I get nothing, that way I'd just sit back and yum up all those 5%'s - then it doesn't matter if the punters lose overall, I'd still get the 5% when they won (with absolutely no risk to me) - and I can keep saying "they win most of the time".

1
0

Re: hhhmmmm

Because:

a. if you put large bets on, somebody notices and you wake up next to a horses head.

and, more likely

b. they are looking to be acquired by a bookie

1
0

Re: hhhmmmm

"I smell BS too"

Yep, see my post above about lifting the skirts for investment or an acquisition.

The BS usually flows pretty thick and fast when looking for $.

2
0
Pirate

Re: hhhmmmm

I recall and episode of "Minder' where Arthur more or less did that with horse racing where he was providing a professional tipping service. I recall he'd stand in a public phone (this is before mobile phones) box to take the calls from his punters and give each caller a 'tip'. Simply he'd give each punter a horse from a race until all the horses in a particular race were covered, then move on to the next race and so on. Each punter would then presumably bet on their 'tip'.

If their horse won they'd pay him a percentage of their winnings if their hose did not win the tip was "free". I'm not sure how he collected I'd assume thats where Terry came in to provide the muscle to ensure payment was forthcomming!!!

0
0

Re: Re: hhhmmmm

I think they mean Hand Egg !

2
0

Predicting the future

If his system was so good, why is he (a) employed, and (b) selling it?

These people are no different from those you see in Westerns selling 'elixir' from the top of a cart. Same scam, different stage.

5
0
Silver badge
Facepalm

> Is it possible to win sports bets consistently over the long haul?

Seeing how quite a few games are rigged (and I hear you can find this out by checking how the bets are standing relative to the historical averages, in yuropean football and sumo wrestling at least), I would not put my money on it.

On the other hand, you can make money by detecting the rigged game in the first place as you then have good information from the future.

0
0
Bronze badge

Seeing how quite a few games are rigged (and I hear you can find this out by checking how the bets are standing relative to the historical averages, in yuropean football and sumo wrestling at least), I would not put my money on it.

There's an extensive discussion of the rigging of sumo in Freakonomics. Apparently it's largely due to the complexities of the payoff matrix - because of how wrestlers are ranked and how rankings affect earnings, there are various cases where the cost of throwing a fight is relatively low for one wrestler, while the benefit to his opponent is large (and so there's ample profit to be split - it's far from zero-sum).

Their model might well be able to include game-fixing, if there's sufficient bias in it (and it's unlikely to be completely, or even close to, random). If it's something like a collection of Hidden Markov Models, the pattern of fixing would also affect the internal weights.

(For determining which games are probably fixed, I suspect a Maximum-Entropy Markov Model with a long feature vector incorporating lots of heterogeneous flags would be more effective. But this is very much not my area, so this is just idle speculation.)

0
0
Bronze badge
Pint

classwork...

We did something like this in my DPhil statistics module... only we used it for predicting clinical diabetes outcomes...

Although I don't see any publications coming out of this group....;-)

"Casinos only attract 2 sorts of persons; mathematicians and suckers"

P.

0
0
Silver badge

Predictalator? Seriously?

I'd be steering clear of this, just on the name alone!

1
0
Bronze badge

Is it possible to win sports bets consistently over the long haul?

Simple: bet on the home team in Nigerian football.

You might find it hard to find someone taking the other side of the bet though...

theguardian.com/football/blog/2013/oct/29/nigeria-toughest-league-win-away

1
0
Silver badge

50,000 cycles aren't enough.

I've been involved in part of software that tested various BlackJack (the casino table card game) rules, each tied to various casinos.

To get decent _consistant_ statistical results, we needed some hundreds of thousands, perhaps a million cycles. Even on the slow machines at the time (8088 4.7Mhz), running a million iterations didn't take an obscene amount of time, so the fact we would be getting reproducable results was more important than the short time it took.

50K would get enough variations in the statistical results that you could run several, and pick the best one - just like what marketing people would do. They're not wildly different, but different enough for the best of the lot to look better on paper.

My guess is, that they're doing exactly that, running several blocks of 50K iterations, and picking the one that best fit what the customer was looking for. The customer gets what they *want* while these guys hide behind "statistical uncertainty", enough to charge for even somewhat unreliable data.

1
0
WTF?

50% against the spread?

If there wasn't a typo in the article, these guys are doing 50% against the spread-- that sounds like they could just be getting lucky since the casinos do a damn fine job of setting the spread to make those bets pan out nearly 50/50. I'm not getting the sense these are Nate Silvers here-- what am I missing?

0
0
This topic is closed for new posts.