pxb
Date Joined: 07 Jun 10
Add contact | Send message
When: 04 Feb 11 10:20
BF is an exchange that wants the profit margins of a bookmaker.


That's key. They need to focus on remaining a low cost bookie, rather than scaling up their costs as much as they can get away with. If Betfair hadn't started changing their pricing structure, I would be striking 100% of my bets here. It's now dipped below 90% and will continue to decline in line with future price increases. I'm sure they make more money from me than they would under the old structure, but this is at the expense of pushing business to their competitors, which can't be a good thing.

Regading the above, I don't really want to go into a debate about it, as this has been done to death, and hasn't really been fruitful

Betfair is best price more often than any other bookie (I both take these good price and also offer prices that are often better than any other bookie). However, bookies like pinny are a very serious competitor for match odds markets.

One thing I would like to hear opinions about though, is two specific elements of the charges which I don't really understand (especially the second):

1) No commission points for PC. This means that if you have a £100 PC bill, you would be better of giving betfair a voluntary commission payment of £100 to get your points. Some PC users (like me) will feel incentivised to trade at GP breakeven to get points (if this is very easy and not time consuming, and doesn't require cash used better elsewhere). Perhaps this is good for Betfair, as it squeezes margins. It certainly isn't good for any other winners who think the PC is a good thing though.

2) data request charges offset by a multiple of commission generated. It literally makes sense to place a bunch of random bets on a day where you hit data request charge territory, as the commission paid is much smaller than data request charges would be. This obviously encourages inefficient behaviour. In this case, I'm pretty sure it's bad for Betfair, unless enough people with DR charges don't bother to do this.

Having said that, I won't suggest any changes to the charging structure with Betfair, as they will (quite rightly) only adopt them if a change benefits them, which will normally mean it does not benefit me.

pxb, when you say the real probability doesn't exist, does that mean you hold a deterministic view? Meaning a super intelligent being with acces to all information would be able to predict with probabilities 1 and 0 who will win a two runner race?

Robot: tight margins are fine but you can't make money long term in true value markets.

Trevh: I disagree. You're only looking at one type of value, namely the odds.

Robot: When I use the term "value" I'm using it as short hand for "positive expected value" and I'm using it in the strict mathematical sense - that is to describe the relationship between the true probability of the outcome you're betting on and the odds at which you're betting. There is only one "type of value" - and that's it. When I use the term "true value" I mean "zero value" - ie exactly as many of your bets will win as the odds would lead you to expect. [snip]
So, to sum up:
All long term winners strike value bets, like it or not, know it or not. This includes traders, bracketers and everyone else.

I still disagree. I mentioned that you're only looking at one type of value, the other type is 'time', not often mentioned around here.

For example, the true price of a 0-0 is 10.0, if you back it at 10.5 you have the value, the only value there is in your opinion.

Yet I could lay that 0-0 at 10.0, the true value, knowing that in certain conditions other layers will hold the price at 10.0 for 5 minutes after kick off, when the true value may or may not still be 10.0 depending on match activity, then I can back it back at 10.0 having had 5 minutes exposure to a goal for free. That value would still be there even if the odds were poor value for me to lay and had drifted to 11.0 pre kick off.

I have just laid in a "true value market" or even a poor value one, and used a different value, called time. This also contradicts what you were saying about tight spreads being bad for profit.

That's just tit for tat. I could say odds are odds - that's why they're called "odds".

I'm not hoping the bet at 10.0 will gain value, it could still be true or poor value 5 minutes after kick off.

I decline your challenge (unsuprisingly) but I do know of bets that show a profit after commission when backing as you suggest, consistently. That's not to say that they pay a profit without hitting value, just that the MMB in that market is slightly too generous.

Are you kidding Bayes? Occasionally you can strike it lucky and win with 2 bad value bets, and of course you can open a position at poor value and close later at value in the same way that you can open at good value and close at poor to reduce variance but by definition a series of results that shows a profit MUST have shown overall value - unless it's a small series that may have got lucky on. The bottom line is that if your results don't show more wins that the odds would imply then you have lost. There are no exceptions to this.

If you place 100 back bets at 2 with level stakes you must win more than 50 to make a profit. I can't think of a simpler example than that. Anyone who can show a profit with 50 wins or fewer gets a Nobel prize.

He may not have cared whether his bet was value or not - but he was still striking value bets. Let's take his own example:


Imagine that Every football match starts with 0-0 at 10, and that it often remains at 10 for the first five minutes unless a goal is scored. Obviously this isn't reality but it illustrates the point.

If in sufficient matches goals are scored in the first five minutes to off-set those times in which the price moved against him then at least one of the following must be true:

1: More goals were scored than were expected (or implied by the odds), this means the opening bets must have shown value.

2: (this is the more likely explanation) Prices were held up when they should have decayed - Ie prices really should have dropped below 10 after play commenced if only because the end of the match was now a bit closer. This means that his closing back bets have shown value.

A combination of the above is also possible of course - but overall if I were shown a profitable set of results using this strategy i'd go for reason two as the more likely explanation.

In this case I can understand why people find it difficult to see where the value is because every time he struck a value bet his profit for the match came up as zero - but the point is that if he hadn't been able to strike those bets he would have made a loss on the vast majority of matches and that loss would have wiped out his profit. Although on the face of it it looks like he's making his profit from the successful lays a better explanation is that his profit comes from the "non loss" he made by washing his face with value back bets.

Ok - let me put it another way - assume he followed this strategy over a large series of matches and made a profit.

If all of his bets were true value then his wins from when early goals were scored would have been wiped out by the losses he made in the matches where there was no early goal and the price moved against him in the first five minutes (which is what you would usually expect it to do).

The reason that his wins from early goals weren't wiped out was that he was able to strike value back bets after five minutes to prevent these losses. Therefore his profit is not really explained by his wins on the early goals, it's explained by the un-expected break evens that he was able to make by striking value back bets when no goals were scored.

where is zipper today management?

Just in case I haven't put cheese on top of my explanation of why Trevh IS betting at value - here's one final way of looking at it, we'll stay with his strategy but to make the maths easier and to show the point more clealy let's assume that 0-0 always goes off at 2.

Say Trevh bets in 110 matches. In 10 of them a goal is scored immediately, in all of the remaining 100 he manages to wash his face by backing at 2 after 5 minutes.

That means he's struck 110 lay bets at 2, and 100 back bets, also at 2.

To show that me must have hit value lets take each set of bets separately and show what happens if they are not value:

1: If all opening bets are true value - that means that out of his total 110 lay bets 55 must have won (for the layer). Well we know that 10 lays won without backs being placed against them, so out of the remaining 100 bets only 45 more can have won- therefore 55 back bets must have won. He only placed 100 back bets, and if those were true value 50 should have won. Therefore if his lay bets show zero value his closing back bets MUST have shown value.

2: let's assume all of his back bets were true value, we know that he placed 100 back bets at 2, and since they were true value 50 must have won. He also placed 110 lay bets, 10 won without a back being placed and of the other 100 matches only 50 back bets won, so 50 more lay bets must have won there. That means that out of 110 lay bets placed 60 won in total when the odds would imply only 55 should have won - so his opening lay bets MUST have shown value.

Q.E.D.

One of the ways I work out new strategies it to assume true value, work out the implications of that and then try to disprove that this is what actually happens. In other words construct a true value model and then compare it to reality - any discrepancies between the model and the reality suggest value. If I were looking at this strategy (which I'm not) I'd be trying to quantify how the odds of 0-0 should decay and then seeing if they were systematically more stable than might be expected. Frankly though - I'd be surprised if that were the case as it would have to mean that either 0-0 was systematically going off at value (probably being under-backed) or that people weren't amending their assessments of CS outcomes during the 1st 5 minutes of a match, which I also think is unlikely. After all, if 0-0 remains static in the market then usually so must all other correct scores. it's even possible that both of these things are true, but I doubt that even more. I could be wrong though - I haven't really looked into it.

If I were Trevh and I had a long set of results which showed a profit I'd be drilling into those results to see exactly where the value was coming from because I might want to look at ditching the bad bets.

Trevh - you can fluke a single event without hitting value but a large set of results can only show a profit by showing value. As before I invite you to give any example of how this can be false - there is none.

i worked as a rent boy to build my bankroll

i walk with a slight limp now but at leat I'm quids in

Trevh
Can you get set on these opportunitues at the usual buy/sell spreads that exist in these secondary mkts.?

Bayes,

That isn't inconsistent with what Aye robot was saying. His point is that on average, you have to be striking value bets.

I believe there's sometimes a difference between the world of theory and the real world we live in

In this much you are right, and I could hardly be accused of living "in the world of theory" given that I strike thousands of bets every day. The theory is used to look for points at which the reality of markets diverges from that theory and is therefore likely to show value.

holding the price at true value and denying the natural decay

This is where you're going wrong, if the price decays as you would expect then it continues to show true value, if however the price doesn't decay when it should then it starts to show positive value - which we call "value" for short. I'm not arguing that you're not striking good bets with this strategy (I have no idea whether you are or not because it's value benot my thing) - I'm just showing that if you're winning using it then it's because you're identifying ts.

...if however the price doesn't decay when it should then it starts to show positive value - which we call "value" for short.

This is where you're going wrong Robot.

Did you not read my posts above concerning match activity? You're making the theorist's mistake of thinking that a 0-0 always has to decay as soon as the clock starts ticking - not true in reality. If the price correctly stands still, my value is time. We may as well stop there though because we'll never see eye to eye. :)

Are you wasting time value Trevh by even debating this point with Robot ?
I think you're both right btw.
Time definiely has a value and you've found it in a exploitable manner, which also fits Robot's point.