Forums

General Betting

Welcome to Live View – Take the tour to learn more
Start Tour
There is currently 1 person viewing this thread.
CrowsEye
21 Apr 17 15:48
Joined:
Date Joined: 08 Jan 12
| Topic/replies: 2,153 | Blogger: CrowsEye's blog
If I take a price at any time within a game (accounting sufficient liquidity e.g. a top football match), does that price generally reflect the true odds of winning at the time, in the same way that we often find the pre-game price is generally accurate?
Pause Switch to Standard View How accurate is any given in-play price?
Show More
Loading...
Report longbridge April 21, 2017 4:20 PM BST
It has to be true that if anyone knew that the price at any point was wrong, and in what direction, they'd be making themselves wealthy on that knowledge and - in the process - correcting to the price, so I'd be very very surprised if anyone were to come up with something like "the match odds market is always too short on the draw in last ten minutes of the first half" or the like.
Report ZEALOT April 21, 2017 9:31 PM BST
With sufficient liquidity you would lose at the rate of commission
Report aye robot April 24, 2017 8:21 PM BST
Assuming a tight spread between back and lay prices i.p. prices are as accurate as pre-off. When the spread widens during hectic periods in any sport value bets are more readily available. As longbridge says; money is made by pushing the prices on both sides of the book back towards true value and this mechanism keeps the prices accurate and limits the extent of any gap between true and offered prices.
Report pxb April 25, 2017 3:20 AM BST
I'd question whether there is any such thing as 'True odds', except to the extent the true odds of the winning outcome is 100% and 0% on the losing outcome(s).

The market is just the sum of current perceptions or expectations of the participants, with some factor for time. People who consistently win are proof that some people have more accurate expectations than the sum of the market has. How to get more accurate expectations is the trick.
Report blank April 25, 2017 11:35 AM BST
I agree that the market reaches an equilibrium based on weight of money, which doesn't mean they're accurate. Take horseracing, the odds could be 3.0 in the morning, 3.4 at the off and 2.8 when the race has been delayed for a few minutes, what price are we supposed to accept as accurate?
Report donny osmond April 25, 2017 11:40 AM BST
traders can cause prices to be wrong, even on slow moving long term markets.

remember spurs next manager and trapattoni price refusing to drift to anywhere near correct price.
Report aye robot April 25, 2017 12:17 PM BST
I'd question whether there is any such thing as 'True odds'

Let me have a try.....

I don't know anyone who doesn't use words like "likely" or who doesn't understand than even when future outcomes are uncertain some outcomes remain more likely than others. Further - everyone I have ever met agrees that this can be quantified to some extent - "more likely", "unlikely but not impossible", "almost certain". Probability is the numeric expression of likelihood accounting for given information. The information part is important and it's what confuses people.

I'll give an example -

I tell you that my friend is about to have a baby - what are the chances that it will be a Girl?
You'd say 50% and you'd be right, this is a true probability and decimal odds of 2 would be a true price. This is not your "opinion," you know that half of all babies born are female, this is all the available information and you're using it correctly.

I don't know the sex of the baby, but I do know that it's going to be called Leslie. Let's say 90% of people in the UK who are called Leslie are female. I know that my friend knows the sex and wouldn't deliberately confuse me so I can confidently assign a probability of 90% to the baby being a girl. This is also a true probability and would imply a true price of 1.11.

So now we have two different true prices for the same event and both are correct, that's because the two assessments are made using different information - in effect they are different events. My assignment of 90% does not make yours of 50% wrong.

When it comes to Betfair prices - 50% of bets matched at 2 do indeed win, as do 25% of those matched at 4 and so on. The data is there for all to see. If the only information you have is the matched price on Betfair then that price can be assumed to be true. If you have different information then you may reach a different assessment of true value. That doesn't make the Betfair price un-true, you're just answering a different question.

The final complication is that the information has to be used correctly. Two people with the same information may weigh it up differently and come to different assessments of the probability of the event. This is not the same as them having different information. In this scenario one of them will get closer to the true probability than the other. Most of the time this is what makes the difference between a successful gambler and a losing one.
Report aye robot April 25, 2017 12:18 PM BST
Take horseracing, the odds could be 3.0 in the morning, 3.4 at the off and 2.8 when the race has been delayed for a few minutes, what price are we supposed to accept as accurate?

I wish I'd quoted this in my reply - the answer is that all are accurate as explained above.
Report blank April 25, 2017 12:59 PM BST
If I decide to lump on a bet for some excitement and force the price down, is the new price accurate?

When it comes to Betfair prices - 50% of bets matched at 2 do indeed win, as do 25% of those matched at 4 and so on. The data is there for all to see.

That's irrelevant for individual markets, the odds could be up to +/- 20% off the true probabilities but if you're just as likely to matched higher than the true odds as you're lower then the overall average will fall in line.
Report aye robot April 25, 2017 1:59 PM BST
If I decide to lump on a bet for some excitement and force the price down, is the new price accurate?

Simply - yes if the only information  you have is the price.

As I have tried to explain above - all probabilities are contingent on information. A probability is not just the probability of an event, but the probability of an event GIVEN SPECIFIC INFORMATION. If the only information you have is the price then the price on Betfair is true.

If you have information other than just the price then you're answering a different question so you get a different answer.
Report pxb April 26, 2017 2:53 AM BST
If the only information you have is the matched price on Betfair then that price can be assumed to be true. If you have different information then you may reach a different assessment of true value. That doesn't make the Betfair price un-true, you're just answering a different question.

I essentially said there is no such thing as 'true value', and you respond with the definition can vary.

In the scientific world, there is a saying, 'If you can't measure it, it doesn't exist.' How do you measure true value?

If we could measure true value, then there would be no gambling. As gambling does exist, it follows true value can't be measured (at least currently), and I'd argue doesn't therefore exist (at least currently in a practical sense).

However, there are factors (in most sports many factors) that point to one outcome over another. Knowing some of these can give you a predictive edge, and a few percentage points are all you need to make money.
Report aye robot April 26, 2017 10:18 AM BST
Of course you can measure value, it's the work of a moment and I do it all the time - how else would I develop my strategies? You measure the value of a group of bets - it wouldn't be meaningful to try to do it with one. That's because probability is not really a property of a single datum but the average of a theoretical infinite group. When I say that my friend's baby has a 50% chance of being a girl what I really mean is that if he has infinitely many Babies half will be girls. My friend's baby is a single datum in that group so we assign it the average probability of the group. After the baby is born I can't say that my assessment that it had a 50% chance of being a girl was wrong because it's not half female.

We assume each bet in a group to have the average value of the group regardless of it's outcome just as we ascribed the average probability of the infinite group to each of its members. It wouldn't make sense to evaluate the bet as part of a group before the outcome is known and then try to evaluate it as in individual datum afterwards. You have to compare apples to apples - groups to groups.

Given that a data set is never infinite measurements of value are always approximate but so are measurements of distance. As a data set reaches an appropriate size we are able to measure value with great confidence.
Report aye robot April 26, 2017 10:30 AM BST
Knowing some of these can give you a predictive edge

Edge is expected value, they are the same thing.
Report Cardinal Scott April 27, 2017 2:11 PM BST
This is a throwback to when this subsection of the forum was full of interesting debate on all things sports trading, all we need now is Trevh and Smithy to drop by!
Report nathanrh April 29, 2017 8:18 PM BST
In answer to the original post and probably agreeing somewhat with others who have posted - I believe the prices on BF are generally very accurate, and probably even more so when there is a significant amount of liquidity.

The market-makers on here are working with very complex but well understood models, be that supremacy ratings, goal expectancies, player specific models etc. etc. The models are constantly evolving and being adjusted real-time. I remember a while back someone quoting the market-maker profit was somewhere between 0.5-1%.

This doesn't mean the models can't be beaten but with a couple of % over-round & a commission rate to overcome you will need something rather special to beat them long term.

The price you 'take' is everything; those offering prices are likely the ones making a profit, anyone who blindly takes the market offer has a significant disadvantage to overcome.

Regards.
Report screaming from beneaththewaves May 5, 2017 10:31 AM BST
Out of every 206 births 106 are male. The proportion varies according to racial and environmental factors, but that's the overall average.

Getting back to the original question, the biggest in-running market, i.e. for cricket matches televised in India, has odds set according to the liabilities of the illegal Indian bookmakers hedging into the Betfair market. This money dwarfs that of punters with an opinion betting directly into the same market. As far as I can see its not uncommon for prices to remain out of kilter with "true odds" for very long periods in these markets, as the weight of money from people hedging liabilities from vast numbers of uninformed punters betting illegally can dwarf that of those of us who might be better able to judge the true odds.
Report screaming from beneaththewaves May 5, 2017 10:44 AM BST
What I'm trying to say is that the odds on in-running cricket markets on Betfair don't necessarily reflect the views of people trying to make money out of those markets themselves. They instead often reflect the views of people who would be better off losing on Betfair, because that would mean that they were at the same time winning bigger sums on the illegal Indian market.

In other words, true odds get distorted by hedging, and on markets of this size there simply isn't always enough money available to take advantage fully of those distortions.
Report stu May 18, 2017 11:25 AM BST
donny osmond 25 Apr 17 11:40

traders can cause prices to be wrong, even on slow moving long term markets.

remember spurs next manager and trapattoni price refusing to drift to anywhere near correct price.


I've often thought this, and still believe it to be true - it even forms one of my own specific strategies in part.
Report stu May 18, 2017 11:26 AM BST
Traders want a specific price often, regardless of fundamentals, the gap is there to exploit.
Report pxb May 18, 2017 11:52 AM BST
The market-makers on here are working with very complex but well understood models, be that supremacy ratings, goal expectancies, player specific models etc. etc. The models are constantly evolving and being adjusted real-time. I remember a while back someone quoting the market-maker profit was somewhere between 0.5-1%.


The myth that there is a correct price is remarkably persistent.

If sports were that predictable, there wouldn't be any point to sports.
Report stu May 18, 2017 1:04 PM BST
pbx, wouldn't you agree there are always fundamentals in any event - just like any trading. This suggests that they can provide an approx true value (with of course some unpredictability)?
Report pxb May 18, 2017 1:29 PM BST
Of course there are fundamentals. Name one fundamental you can accurately measure in any sport. Stats aren't fundamentals.
Report aye robot May 18, 2017 4:17 PM BST
Anyone who can show that probability is a 'myth' shouldn't be wasting their time here - they should be winning a Fields medal and changing the world  by overturning a foundational concept in mathematics. The whole of statistics and most modern physics will crash to the ground and you will be rich beyond the dreams of avarice.

If you feel able to use a phrase like 'Arsenal will probably beat Leicester at the weekend' then you have grasped the basic concept of probability - the rest is just maths that follows from that. It's not something you 'believe in,' you just either understand it or you don't.
Report Fatslogger May 19, 2017 8:00 AM BST
I think pxb is not really suggesting that probability is a defunct concept, rather that the complexities of live sport make it very difficult to assign correct probabilities to. My view is that this is only true up to a point and that in some situations the markets will be be pretty much spot on. I'm not nearly a good enough mathematician to do this myself, but it certainly is possible to do prospective testing of predictive models (including Betfair markets) against outcomes to see how good they are. How much literature is out there on Betfair I'm not sure but there certainly is some.

I don't bet much on football match odds because I can't really see that I've got any additional insight but I do win money on side markets where sentiment certainly outweighs calculation and on cricket, where modelling is pretty tough. It would be an entirely different and rather long thread needed to detail why, for example, the WinViz predictive tool is poor but a simple example from the last test played, between WI and Pakistan, had it thinking the game was a certain draw when draw odds on Betfair were approaching double figures when the 8th and 9th wickets fell, I think because it didn't know how many overs were left.
Report pxb May 19, 2017 8:45 AM BST
No sporting event is ever repeated with the same starting conditions, in the way a coin toss would be (although a couple come close like snooker). Conventional probability doesn't apply to the outcome. Hence, the 'myth'. Which is not to say, some causative factors can't be approximated by statistical analysis, but some can't, and some causative factors are indistinguishable from chance. Although some causal factors can be directly quantified (the weather for example).

The people who do statistical analysis are basically looking to approximate enough of the causative factors to overcome the factors that can't be approximated and chance of course. And with more precision than the next guy doing the same.

I don't bother with any of that. Except directly quantifiable factors.
Report aye robot May 19, 2017 10:06 AM BST
Conventional probability doesn't apply to the outcome.

Probability still applies. You don't need successive identical events for probability to apply. As I explained above probability is a property of information not of an event.

Probability theory models non-random events as if they were random as a way of dealing with imperfect information. With perfect information a coin toss coming out heads would have a probability of either 0 or 1 but we assign it 0.5 because with the information we have that's the best we can do. It's exactly the same with a sporting event or a weather forecast.

Like anything else you can get a probability wrong or you might not know what it is, but the fact that you don't (or even can't) know it doesn't mean it doesn't exist.
Report nathanrh May 20, 2017 12:09 AM BST
I agree to a certain extent with pxb, there isn't going to be enough (historical and current) statistical data to precisely define an exact 'true price' or probability of an outcome like a football match, the data-set is too small and chance plays a much greater part in the overall equation.

However, the point I am making is that the market making models and the weight of money are incredibly good performers when measured long term. This doesn't of course mean there aren't pricing inaccuracies that can't be exploited, but that is when the Betfair commission rate, over-round and lack of liquidity will bite, it is fantasy to think that the market will consistently be giving away value of that magnitude.

And there in is the rub; If you had a consistent edge that could overcome the commission rate by 1-2% you would be a millionaire, and if your edge was 1-2% on markets with masses of liquidity you'd be a multi-millionaire, either way you wouldn't be messing around on the forum debating semantics.

But of course this is the Betfair Forum, and we are all consistent winners here ;-).....
Report henryluca May 20, 2017 7:34 AM BST
If I take a price at any time within a game (accounting sufficient liquidity e.g. a top football match), does that price generally reflect the true odds of winning at the time,

Absolutely not---sporting events are not random roulette wheel type events-

There is often  player manipulation- team manipulation- late come backs from the better team - all every day occurrences.

All the jargon about probabilities etc doesn't apply when their is gambling and humans
Report aye robot May 20, 2017 8:44 AM BST
All the jargon about probabilities etc doesn't apply when their is gambling and humans

It's not jargon, it's maths and it applies to EVERYTHING.

Coin flipping and roulette are page one in the ladybird book of probability. Probability goes far beyond that. People don't like it so they choose not to believe it - what can you do about that?
Report Coachbuster May 20, 2017 3:36 PM BST
what Henryluca probably means is the emotional side to gambling -  sentiment betting . It's  all related to supply and demand ,where the odds aren't true .

Most of us have  knowingly bought poor value to reduce risk
Report henryluca May 21, 2017 4:26 AM BST
Coachbuster

In part yes---but also  the commonly occurring rorting at some level --micro bets or otherwise

Coachbuster and I are playing tennis and we privately collude that I will give him a headstart in first half of game and he will let me make a comeback in the second half----

Aye--this is human nature and greed and the only maths is the number of  big TVs , expensive cars and penthouse units around the world that we both want

We are talking about sport , humans and gambling---

The real fallacy trap many fall into is the 'certainty that a second half of a game will always be a mirror image of the first half'


Like anything else you can get a probability wrong or you might not know what it is, but the fact that you don't (or even can't) know it doesn't mean it doesn't exist.

Pure probability is not guess work---it should be the outcome of considering the  the ratio of the favourable cases to the whole number of cases possible.

So unless there exists  the 2 elements of this then there is no probability
Report henryluca May 21, 2017 4:38 AM BST
Like anything else y[u]ou can get a probability wrong or you might not know what it is, but the fact that you don't (or even can't) know it doesn't mean it doesn't exist.
[/u]
Pure probability is not guess work---it should be the outcome of considering the  the ratio of the favourable cases to the whole number of cases possible.

So unless there exists  the 2 elements of this then there is no probability




To clarify--I mean that probability is calculable and therefore capable of accurate calculation if the 2 elements are present----save for clerical error you cant get the probability calculation wrong--
Post Your Reply
<CTRL+Enter> to submit
Please login to post a reply.

Wonder

Instance ID: 13539
www.betfair.com