Overround, OddsKey

Post by Gosen
I am new to betting and am finding the concept of overround, and the
related concept of the odds key, to be confusing. I do not see how
these concepts, which propose to express expectation (apparently), can
do so without referring to the real probabilities of the underlying
events. I am a beginner, and so the statements made below reflect
merely a transient understanding of their subject. Please feel free to
correct me as you see fit.
OddsKey = 1/odds[1] + 1/odds[X] + 1/odds[2]
where odds[x] yields the EUR-style decimal odds of outcome x. The
http://www.oddsexchange.com/servlet?cat=help#cprob
Apparently, the OddsKey gives the amount of money kept for every
dollar payed out, given an equal investment in each outcome (I'm
supposing)
OverRound = (%[1] + %[X] + %[2]) - 100%
This, as I understand, is the percentage kept of the total monies
coming in to a game.
I am confused by these two concepts because, to my understanding,
neither decimal nor percentage odds are real probabilities, and thus
cannot be treated as such. These numbers are not even assessments of
probabiltiies, but just payout ratios designed to positiviely offset
the final expectation of the bookmaker, given the REAL probabilities
of each outcome. I have been trying to understand how odds offered to
a punter have any direct relationship to the real or assessed
probability of an underlying outcome, except through a traditional
expectation formula with real probabilities explicitly included. As
stated above, I am a beginner so if any assumptions above are
incorrect, feel free to correct them.

The 'real' probabilities of the outcome of any event are pretty subjective
and have little to do with the market which will mainly reflect diverse
opinions. The overound percentage is a financial advantage based upon
market prices. On a perfect book ,and over a period of time, the bookmaker
will win because he will keep x% of the take (the overound) and is in fact
selling at a profit.
Most serious punters spend much time seeking 'value', a situation where the
price offered is better than the odds of the probable outcome, as they see
it.
This might be based on observation. With others it is based on a more
scientific approach of close form study. Whatever the method it remains a
bit
hit and miss. If anyone finds the way to calculate the true likelihood of
a
particular outcome the world is their lobster, at least, until all the
bookies have
gone broke.

Les

Gosen

2003-12-11 00:59:19 UTC

Thanks for the response. I just confused as to how a summation of odds
can have any meaning when assessments of underlying probabilities are
not included in the calculation. For instance, the definition of the
OddsKey is the total money kept for every Euro (unit) payed out,
regardless of outcome, and this number is derived simply by adding
together the odds offered for the various outcomes of the game.
Doesn't the amount of money kept by the bookmaker depend upon how
likely each given outcome is to occur, in addition to the payout odds
offered to the punter? Do you know what I mean?

Goalie of the Century

2003-12-11 07:30:57 UTC

Post by Gosen
Doesn't the amount of money kept by the bookmaker depend upon how
likely each given outcome is to occur, in addition to the payout odds
offered to the punter?

It depends on how much of each bet the bookmaker can 'sell' at what
price.

To take an example, on http://www.betfair.com , you can back
Manchester Utd at 1.42 Manchester City at 11 and the draw at 4.5 for the
match on Saturday.

That gives an overround of 1.74%. If the bookmaker takes bets of 70.42
on Man U, 9.09 on Man C and 22.22 on the draw, they are sure of a profit
of 1.74 (less Betfair's commission) whatever the outcome.

You can back or lay outcomes on Betfair and the overround on popular
events is small so their odds should reflect the view of their users of
the probability of the outcomes.

The big bookmakers have their overheads to cover but also years of
experience in knowing what the uptake will be on different offers.

On the Man U match, http://www.ladbrokes.com are currently offering
1.36, 8 and 3.75. That's 12.7% overround but the difference in the odds
on the individual outcomes are 4.4%, 37.5% and 20%. They know they have
to take a lower margin on more likely outcomes to attract enough bets.

--
Goalie of the Century

Les Corbett

2003-12-11 21:08:17 UTC

Post by Gosen
Thanks for the response. I just confused as to how a summation of odds
can have any meaning when assessments of underlying probabilities are
not included in the calculation. For instance, the definition of the
OddsKey is the total money kept for every Euro (unit) payed out,
regardless of outcome, and this number is derived simply by adding
together the odds offered for the various outcomes of the game.
Doesn't the amount of money kept by the bookmaker depend upon how
likely each given outcome is to occur, in addition to the payout odds
offered to the punter? Do you know what I mean?

I think I understand your problem but think you are paying too much
attention
to this vauge concept of probable outcome which, I repeat, is something
which
cannot be specific to fine margins. If a bookmaker, offering
prices on a horserace, has this inbuilt margin of profit because of the over
round
he will profit in the long term although having to endure losses on
individual races.
In a general sense, it will make no difference to him which horse is the
most
likely to win (probability) his market is dictated by what the punters feel
to be the
likely winner, which may or may not be the same thing. The only time the
market
does not dictate is when none exists. Probability, translated into a price
is obviously
relevant to those setting the tissue, the bookies opening offers. These
opinions,
mainly based on past form , are usually pretty accurate in reflecting the
interests of
punters but once the market forms demand will again dictate.

Les

Gosen

2003-12-12 21:55:34 UTC

Allright
I'm trying to understand how the overround and S.P. works. That is,
how does it, exactly, measure the in-built profit of the bookmaker? I
understand these measurements are subject to the fluctuations of
random variables. That goes without saying. But these formulas seem to
proport to measure the expectation of the bookmaker/punter. If this is
the case, then my first motion towards understanding is to apply the
formal definition of expectation and probability. In doing so, I may
be paying too much attention to the traditional interpretation of
probability. For instance, your point about prices subsequent to the
tissue price representing much more than just a pure analysis of the
underlying event is well-taken. Synthesizing the disparate assessments
of an underlying probability from many different market participants
may not yield a valid probability, but it obviously measures
something. Hence I may even be paying too much attention to the formal
definition of probability. After all, market odds cannot routinely sum
to 1 as required by a formal definition. However, if overround
proports to measure 'expectation', but we're not using the formal
definition of 'expectation' nor 'probability', then what definition
are we using? And if the overround is not designed to measure
expectation, how can it measure the vig? I have a tentative response
to this here:

http://groups.google.com/groups?q=overround+oddskey&hl=en&lr=&ie=UTF-8&selm=6c599448.0312111748.6ae6e43%40posting.google.com&rnum=7&filter=0

But I may be using the market and 'market assumptions' as a way to
cover for misunderstanding. If this is the case, I'd appreciate a
critique.

The 'real' probabilities of the outcome of any event are pretty

subjective

Post by Les Corbett
and have little to do with the market which will mainly reflect diverse
opinions. The overound percentage is a financial advantage based upon
market prices. On a perfect book ,and over a period of time, the

bookmaker

Post by Les Corbett
will win because he will keep x% of the take (the overound) and is in

fact

Post by Les Corbett
selling at a profit.
Most serious punters spend much time seeking 'value', a situation where

the

Post by Les Corbett
price offered is better than the odds of the probable outcome, as they

see

Post by Les Corbett
it.
This might be based on observation. With others it is based on a more
scientific approach of close form study. Whatever the method it remains

Post by Les Corbett
bit
hit and miss. If anyone finds the way to calculate the true likelihood

Post by Les Corbett
a
particular outcome the world is their lobster, at least, until all the
bookies have
gone broke.
Les

Paul Robinson

2003-12-13 14:46:32 UTC

I'm new around here, so like, hello and stuff.

Post by Gosen
Allright
I'm trying to understand how the overround and S.P. works. That is,
how does it, exactly, measure the in-built profit of the bookmaker? I

Oh boy. I think you've missed the point.

OK, this is not going to be enough to define it perfectly, and for that
I would advise you get hold of this book:

http://www.amazon.co.uk/exec/obidos/ASIN/1872254063/ref=sr_aps_books_1_1/202-6015485-7367006

The short answer though is that over-round means that if a bookmaker
takes bets to a common payout and is over-round, he gets the same profit
no matter what the result. If he is betting to the same liability (not
the same as same payout) then it is likely he will make a much bigger
profit on the outsiders winning, but will lose money on a favourite
winning - this style of bookmaking is much more common in horse racing
than in football.

You must realise, in fact it is mandatory to any understanding of
betting of any sort that the prices on offer have very little to do with
a bookmaker's opinion of the chances of a particular outcome. Rather,
the bookmaker is anticipating the amount of money he can expect for a
given outcome. As a result the concept of "value" emerges. When the
bookmaker thinks he will not get a lot of money for a selection and the
price he offers as a result drifts out from the real odds, this is
considered "value".

A practical example might help explain. Let's take a game being played
today - Chelsea vs. Bolton - and the prices offered on Betfair. As with
all betting exchanges they use decimal odds which makes over-round
calculation a bit easier (divide 100 by the price).

Selection Price to back Backer's Stake Payout
Chelsea 1.34 £74.62 £99.99
Bolton 11.5 £8.69 £99.93
The Draw 5.6 £17.85 £99.96

This book is very thin - just 101% over-round, but no matter who wins if
I were to lay this book, I'd make at make £1.11 (Chelsea), £1.14 (Draw)
or £1.17 (Bolton). If instead of doing a common payout of £100, I'd gone
for £1000, no matter what happens, I'd make at least a tenner. And £10k?
What about the £100ks going through the betting shops? What if instead
of a 101% book, I was able to lay it at say, 120%? Well, not suprisingly
I'd be making £20 for every £100 laid...

If I'd bet to the same liability though:

Selection Price to back Backer's Stake Liability
Chelsea 1.34 £294.11 £100
Bolton 11.5 £9.52 £99.96
The Draw 5.6 £21.73 £99.96

If Chelsea win, I lose £68.75, Bolton win I make £205.52 and if the Draw
comes up, I make £193.90 - this looks better value if you fancy Bolton
or the draw, but a wise bookmaker on this market will go to the same
payout and keep an even payout regardless...

The important thing is that I have to take bets in the proportions I'm
expecting them. If I'm not getting enough on one selection I push the
price out to attract more money. If I'm getting too much, I bring it in
so that punters are a bit more hesitant - I bet you've thought twice
about putting money on a 10/11 shot that you know should really be evens
or bigger. On a common payout, my margin on turnover will be roughly
(over-round - 100/over-round) * 100- e.g. 120% over-round is
(20/120)*100 or about 16.67% on turn-over.

In short, prices on offer have nothing at all to do with real
probability, and EVERYTHING to do with the amount of cash coming across
the counter. This is why England always look a bit thin on the prices
for internationals, even though we all know they may struggle,
bookmakers are rich, and why people who want to make money out of
backing always do so on the exchanges where books on football matches
are typically 101% or lower instead of the silly prices on offer with
traditional bookies.

--
Paul Robinson

Gosen

2003-12-19 09:35:43 UTC

:nods in assent, smiles with humility:

This seems like a far more sensible and natural way to look at the
overround. Goalie of the Century (above) was getting at this as well,
but I wasn't paying enough attention.

Your book recommendation is greatly appreciated. Does it cover
other betting structures from other countries as well? Is it technical
but accessible (i.e. technical but not just a bunch of academic
papers)? Does it emphasize analysis over history? Other book
suggestions would be great.

I can see how from the assumption that dollars bet can be
controlled by the bookmaker a constant or predictable dollar output
follows. This was your main point. And I gather this 'assumption' is
an empirical fact. However I'm not quite sure where you got these
numbers:

1.11 (chelsea)
1.17 (bolton)
1.14 (draw)

And I'm not sure what you mean by 'liability'.

Post by Paul Robinson
In short, prices on offer have nothing at all to do with real
probability, and EVERYTHING to do with the amount of cash coming across
the counter. This is why England always look a bit thin on the prices
for internationals, even though we all know they may struggle,
bookmakers are rich, and why people who want to make money out of
backing always do so on the exchanges where books on football matches
are typically 101% or lower instead of the silly prices on offer with
traditional bookies.

Do prices really have NOTHING to do with real probability? Isn't the
punter genuinely concerned with real probability and mostly
responsible for the price shifts made by the bookmaker?
Also, there does seem to be a relationship between the expectation
formula and overround. Is this just a repetition of the idea that
dollars inputted can be controlled? This wouldn't suprise me, as the
basic odds ratio (inverted) is simply being multiplied. Does it bear
any connection to traditional expectation and probability beyond mere
form? If you want to verify for yourself just make calculations based
on the assumptions that monies bet are evenly distributed and that
decimal odds directly express 'probabilities'. 'expectation' should be
proportional to moniesbet and overround. Its possible of course I've
made some banal error of calculation.

Post by Paul Robinson
I'm new around here, so like, hello and stuff.

Post by Gosen
Allright
I'm trying to understand how the overround and S.P. works. That is,
how does it, exactly, measure the in-built profit of the bookmaker? I

Oh boy. I think you've missed the point.
OK, this is not going to be enough to define it perfectly, and for that
http://www.amazon.co.uk/exec/obidos/ASIN/1872254063/ref=sr_aps_books_1_1/202-6015485-7367006
The short answer though is that over-round means that if a bookmaker
takes bets to a common payout and is over-round, he gets the same profit
no matter what the result. If he is betting to the same liability (not
the same as same payout) then it is likely he will make a much bigger
profit on the outsiders winning, but will lose money on a favourite
winning - this style of bookmaking is much more common in horse racing
than in football.
You must realise, in fact it is mandatory to any understanding of
betting of any sort that the prices on offer have very little to do with
a bookmaker's opinion of the chances of a particular outcome. Rather,
the bookmaker is anticipating the amount of money he can expect for a
given outcome. As a result the concept of "value" emerges. When the
bookmaker thinks he will not get a lot of money for a selection and the
price he offers as a result drifts out from the real odds, this is
considered "value".
A practical example might help explain. Let's take a game being played
today - Chelsea vs. Bolton - and the prices offered on Betfair. As with
all betting exchanges they use decimal odds which makes over-round
calculation a bit easier (divide 100 by the price).
Selection Price to back Backer's Stake Payout
Chelsea 1.34 £74.62 £99.99
Bolton 11.5 £8.69 £99.93
The Draw 5.6 £17.85 £99.96
This book is very thin - just 101% over-round, but no matter who wins if
I were to lay this book, I'd make at make £1.11 (Chelsea), £1.14 (Draw)
or £1.17 (Bolton). If instead of doing a common payout of £100, I'd gone
for £1000, no matter what happens, I'd make at least a tenner. And £10k?
What about the £100ks going through the betting shops? What if instead
of a 101% book, I was able to lay it at say, 120%? Well, not suprisingly
I'd be making £20 for every £100 laid...
Selection Price to back Backer's Stake Liability
Chelsea 1.34 £294.11 £100
Bolton 11.5 £9.52 £99.96
The Draw 5.6 £21.73 £99.96
If Chelsea win, I lose £68.75, Bolton win I make £205.52 and if the Draw
comes up, I make £193.90 - this looks better value if you fancy Bolton
or the draw, but a wise bookmaker on this market will go to the same
payout and keep an even payout regardless...
The important thing is that I have to take bets in the proportions I'm
expecting them. If I'm not getting enough on one selection I push the
price out to attract more money. If I'm getting too much, I bring it in
so that punters are a bit more hesitant - I bet you've thought twice
about putting money on a 10/11 shot that you know should really be evens
or bigger. On a common payout, my margin on turnover will be roughly
(over-round - 100/over-round) * 100- e.g. 120% over-round is
(20/120)*100 or about 16.67% on turn-over.
In short, prices on offer have nothing at all to do with real
probability, and EVERYTHING to do with the amount of cash coming across
the counter. This is why England always look a bit thin on the prices
for internationals, even though we all know they may struggle,
bookmakers are rich, and why people who want to make money out of
backing always do so on the exchanges where books on football matches
are typically 101% or lower instead of the silly prices on offer with
traditional bookies.

Les Corbett

2003-12-19 23:05:47 UTC

Post by Gosen
This seems like a far more sensible and natural way to look at the
overround. Goalie of the Century (above) was getting at this as well,
but I wasn't paying enough attention.

Do prices really have NOTHING to do with real probability? Isn't the
punter genuinely concerned with real probability and mostly
responsible for the price shifts made by the bookmaker?

As I said way back, only in a very general sense is there such a thing
as probability. We know that Manchester United will win most of their
home games and therefore will 'probably' win any named home game.
But the science of probability ,which might work perfectly for roulette,
plays little part in competitive events. What are the probable chances
of Man U beating this particular team if offered as a price? We are
now in the area of opinion, the exact area where all betting takes place.
Probablity is a subjective assessment of what might happen but is
obviously meaningless, in a gambling sense, unless related to price.
Your question about "real probability" seems to suggest you still feel that
such a thing exists and can be calculated. It doesn't, unless consensus
(opinion) and probability are the same thing?
Probability, or one persons concept of it, comes into the play only on
opening prices offered before a market has formed, and even then (the
England team price was a fine example) might well reflect expected
demand rather than likely outcome.
I feel you are getting bogged down in trying to make sense of something
which is, and always will be, inexact. Competitive betting sports have too
many variables and unknown factors to be approached this way. Even the
bookies overound margin works in their favour only in a general way over
a period. They will seldom have a perfect book where their liabilities (the
amount they have to pay out on any given result ) will give them a profit
regardless. Most on course bookies would not want such a thing ,
prefering to lay some to be big losers, therefore a bigger book win iwhen
beaten.

If you want to know what the "real" probability of Man U beating Spurs this
week
are, translated into odds, a crystal ball maybe?

Les

Post by Gosen
Also, there does seem to be a relationship between the expectation
formula and overround. Is this just a repetition of the idea that
dollars inputted can be controlled? This wouldn't suprise me, as the
basic odds ratio (inverted) is simply being multiplied. Does it bear
any connection to traditional expectation and probability beyond mere
form? If you want to verify for yourself just make calculations based
on the assumptions that monies bet are evenly distributed and that
decimal odds directly express 'probabilities'. 'expectation' should be
proportional to moniesbet and overround. Its possible of course I've
made some banal error of calculation.

Post by Paul Robinson
I'm new around here, so like, hello and stuff.

Post by Gosen
Allright
I'm trying to understand how the overround and S.P. works. That is,
how does it, exactly, measure the in-built profit of the bookmaker? I

Oh boy. I think you've missed the point.
OK, this is not going to be enough to define it perfectly, and for that

http://www.amazon.co.uk/exec/obidos/ASIN/1872254063/ref=sr_aps_books_1_1/202
-6015485-7367006

Post by Paul Robinson
The short answer though is that over-round means that if a bookmaker
takes bets to a common payout and is over-round, he gets the same profit
no matter what the result. If he is betting to the same liability (not
the same as same payout) then it is likely he will make a much bigger
profit on the outsiders winning, but will lose money on a favourite
winning - this style of bookmaking is much more common in horse racing
than in football.
You must realise, in fact it is mandatory to any understanding of
betting of any sort that the prices on offer have very little to do with
a bookmaker's opinion of the chances of a particular outcome. Rather,
the bookmaker is anticipating the amount of money he can expect for a
given outcome. As a result the concept of "value" emerges. When the
bookmaker thinks he will not get a lot of money for a selection and the
price he offers as a result drifts out from the real odds, this is
considered "value".
A practical example might help explain. Let's take a game being played
today - Chelsea vs. Bolton - and the prices offered on Betfair. As with
all betting exchanges they use decimal odds which makes over-round
calculation a bit easier (divide 100 by the price).
Selection Price to back Backer's Stake Payout
Chelsea 1.34 £74.62 £99.99
Bolton 11.5 £8.69 £99.93
The Draw 5.6 £17.85 £99.96
This book is very thin - just 101% over-round, but no matter who wins if
I were to lay this book, I'd make at make £1.11 (Chelsea), £1.14 (Draw)
or £1.17 (Bolton). If instead of doing a common payout of £100, I'd gone
for £1000, no matter what happens, I'd make at least a tenner. And £10k?
What about the £100ks going through the betting shops? What if instead
of a 101% book, I was able to lay it at say, 120%? Well, not suprisingly
I'd be making £20 for every £100 laid...
Selection Price to back Backer's Stake Liability
Chelsea 1.34 £294.11 £100
Bolton 11.5 £9.52 £99.96
The Draw 5.6 £21.73 £99.96
If Chelsea win, I lose £68.75, Bolton win I make £205.52 and if the Draw
comes up, I make £193.90 - this looks better value if you fancy Bolton
or the draw, but a wise bookmaker on this market will go to the same
payout and keep an even payout regardless...
The important thing is that I have to take bets in the proportions I'm
expecting them. If I'm not getting enough on one selection I push the
price out to attract more money. If I'm getting too much, I bring it in
so that punters are a bit more hesitant - I bet you've thought twice
about putting money on a 10/11 shot that you know should really be evens
or bigger. On a common payout, my margin on turnover will be roughly
(over-round - 100/over-round) * 100- e.g. 120% over-round is
(20/120)*100 or about 16.67% on turn-over.
In short, prices on offer have nothing at all to do with real
probability, and EVERYTHING to do with the amount of cash coming across
the counter. This is why England always look a bit thin on the prices
for internationals, even though we all know they may struggle,
bookmakers are rich, and why people who want to make money out of
backing always do so on the exchanges where books on football matches
are typically 101% or lower instead of the silly prices on offer with
traditional bookies.

Paul Robinson

2003-12-21 13:11:01 UTC

Post by Gosen
This seems like a far more sensible and natural way to look at the
overround. Goalie of the Century (above) was getting at this as well,
but I wasn't paying enough attention.

Easily done. :-)

Post by Gosen
Your book recommendation is greatly appreciated. Does it cover
other betting structures from other countries as well? Is it technical
but accessible (i.e. technical but not just a bunch of academic
papers)? Does it emphasize analysis over history? Other book
suggestions would be great.

It covers the history of bookmaking in the UK and in addition discusses
the mathematics involved. Some parts are very technical and require a
good grounding in algebra, but other bits are much more accessible. I
enjoyed it, and it's about the only book I know on the subject.

US bookies (aka "criminals" :-) ) tend to bet to even money bets and
handicap the teams to make it all even and make their money by getting
the vig off the losing bets. As for other countries, I don't know of
many but this is quite a light-hearted look at the US game:

http://www.amazon.co.uk/exec/obidos/ASIN/1581600704/202-6015485-7367006

You'll notice I've already provided a comprehensive review there
already. :-)

Other than that, although dated now (it concentrates on UK betting in
the early 1990's which pre-dates the exchanges) this is quite a nice
little book if you like horse racing and covers some scams:

http://www.amazon.co.uk/exec/obidos/ASIN/1851588167/qid=1072011545/sr=1-1/ref=sr_1_16_1/202-6015485-7367006

And if you want to read up on how the odds are often put up against you
and you just want to look at ways to cheat (!) then this is pretty good
reading:

http://www.amazon.co.uk/exec/obidos/ASIN/0818405295/qid=1072011591/sr=1-1/ref=sr_1_2_1/202-6015485-7367006

Post by Gosen
I can see how from the assumption that dollars bet can be
controlled by the bookmaker a constant or predictable dollar output
follows. This was your main point. And I gather this 'assumption' is
an empirical fact. However I'm not quite sure where you got these
1.11 (chelsea)
1.17 (bolton)
1.14 (draw)
And I'm not sure what you mean by 'liability'.

Ok, by far the best thing to do right now is to go an play with a
betting exchange with small money. Betfair is the biggest. Learn digital
odds, go to a market you're interested in with just a few items, above
the market make sure "View P&L" is ticked, click on "Lay All", click on
"Payout" and put in a common takeout. Play with the "Liability" button
too. It all starts to make sense after a while, as does the over-round.

Alternatively, you can wait three or four months when I get all my notes
on this written up and publish them for free on the web. The problem is,
this area is VAST. It looks so simple but the more you scrape away, the
more you realise there are caveats that need to be understood
mathematically if you want to make money...

Post by Gosen
Do prices really have NOTHING to do with real probability? Isn't the
punter genuinely concerned with real probability and mostly
responsible for the price shifts made by the bookmaker?

Punters can be classified in two categories:

1. Followers who will bet no matter what price is offered
2. Value seekers

It's the latter that control the markets and the ones the bookies need
to watch out for. They cause market moves, and without them, bookies
would quite possibly be even richer than they already are. The question
is, does "value" have anything to do with probability. Well, yes,
obviously. If I think the chance of something occuring is greater than
the price suggests, the price is good value. Likewise, if I offered you
10/11 that a coin I toss comes up heads, you'll see that as poor value
because you know the probability is evens.

The way sports betting works however is more complicated because we
don't know specific probabilities. A vague sense of chance yes, but
that's not the same as probability. As we all have a different
perception of what is going to happen in a given event, bookies just
need to work out not the probability, but the proportion of stakes he is
likely to have to take bets in. If we know in a game between Man Utd and
Leeds that 10 time more people will bet on MUFC than LU, and only 1 in
100 will bet the draw, we can form a book with a common takeout, and
thus, a profit!

As such, prices are not to with probability, but instead with the
proportions of perceptions the bookmaker predicts.

Post by Gosen
on the assumptions that monies bet are evenly distributed and that
decimal odds directly express 'probabilities'. 'expectation' should be
proportional to moniesbet and overround. Its possible of course I've
made some banal error of calculation.

Almost. Think of it like this - prices don't correspond to
probabilities, but instead are the lowest prices possible a bookmaker
can expect people to bet in at appropriate proportions. Does that make
it easier to grasp at what I'm trying to point out?

--
Paul Robinson

Gosen

2003-12-24 22:39:49 UTC

Forgive the Latency...(holidays)...for those that care I shall make a
committment to responding more quickly henceforth. I have responded
below to Paul,Les and John...

Post by Paul Robinson
Ok, by far the best thing to do right now is to go an play with a
betting exchange with small money. Betfair is the biggest. Learn digital
odds, go to a market you're interested in with just a few items, above
the market make sure "View P&L" is ticked, click on "Lay All", click on
"Payout" and put in a common takeout. Play with the "Liability" button
too. It all starts to make sense after a while, as does the over-round.
Alternatively, you can wait three or four months when I get all my notes
on this written up and publish them for free on the web. The problem is,
this area is VAST. It looks so simple but the more you scrape away, the
more you realise there are caveats that need to be understood
mathematically if you want to make money...

The former makes a great deal of sense. This way I won't have to ask
dull questions. The second makes even more sense. Finding good
teachers is not an easy task. As for the area being vast, it seems to
me that mystery (uncertainty), among other things, keeps the market
alive.

By the way, do you believe its possible for someone without luck (the
kind that characterizes entire lifetimes), brilliance and inside
information to make money in the gambling markets merely as a punter
(see also my comments on John Gee's below)?

On to books (momentarily)...I think I'll take a look at your first
suggestion (Art of Legging etc.) because right now I'm in search of
substantive and technical works that will lead me towards a profession
(or a pipe-dream I suppose). I don't intend this to be a criminal one
(i.e. rice paper and buckets)! You said in your review for "The Art of
Legging" that the American betting system is technologically
backwards, which I find to be suspect. I'm interested in your argument
for this. I'd like to try to provide a refutation or be bested. I
might argue that by placing contraints on the content of the game
itself the bookie in fact yields more control over the outcome of the
proposition (thereby reducing variability), while at the same time
clarifying for the punter the precise cost incurred (reducing
ambiguity).

I also noticed the following books you might interested in at least
commenting on:
http://www.amazon.co.uk/exec/obidos/ASIN/0415260914/qid=1072254352/sr=1-1/ref=sr_1_2_1/202-9496537-6833426
http://www.amazon.co.uk/exec/obidos/ASIN/1843440091/ref=pd_ecc_rvi_f/202-9496537-6833426

this one might carry some interest:

http://www.amazon.co.uk/exec/obidos/ASIN/0942828321/qid=1072254352/sr=1-2/ref=sr_1_2_2/202-9496537-6833426

Post by Paul Robinson
1. Followers who will bet no matter what price is offered
2. Value seekers
It's the latter that control the markets and the ones the bookies need
to watch out for. They cause market moves, and without them, bookies
would quite possibly be even richer than they already are. The question
is, does "value" have anything to do with probability. Well, yes,
obviously. If I think the chance of something occuring is greater than
the price suggests, the price is good value. Likewise, if I offered you
10/11 that a coin I toss comes up heads, you'll see that as poor value
because you know the probability is evens.

The fact that value seekers have 'control', as opposed to influence,
is for me rather dissappointing, for it makes the likelihood that
betting markets, especially online exchanges, are or may become
efficient (I shall leave the term open now). Perhaps I should leave
the question very open: under what circumstances can or can't a person
make a LIVING through gambling investment, given modest intitial
starting cash? Or is such a pursuit a fool's errand? Perhaps I'm being
too frank.

For the purpose of discussion, I'd like to float the idea that in a
negative-sum market (like gambling), an efficient market is not a
profitable one. My understanding of these terms is not precise and so
I expect much room for disagreement.

Post by Paul Robinson
The way sports betting works however is more complicated because we
don't know specific probabilities. A vague sense of chance yes, but
that's not the same as probability. As we all have a different
perception of what is going to happen in a given event, bookies just
need to work out not the probability, but the proportion of stakes he is
likely to have to take bets in. If we know in a game between Man Utd and
Leeds that 10 time more people will bet on MUFC than LU, and only 1 in
100 will bet the draw, we can form a book with a common takeout, and
thus, a profit!

Perhaps as a collective, punters assessing through fuzzy means the
'real' probabilities of given outcomes yield, through a market
process, an increasingly accurate measurement of those 'real'
probabilities.

Post by Paul Robinson

My only point about the relationship I posited here:
http://groups.google.com/groups?q=overround+oddskey&hl=en&lr=&ie=UTF-8&selm=6c599448.0312111748.6ae6e43%40posting.google.com&rnum=7&filter=0

is that overround does seem to be precisely related to the calculation
of expectation made given the assumptions I mentioned. I'm only
wondering if this relationship is a mere accident, expresses nothing
new or interesting, or perhaps reveals some connection between odds,
odds-setting and expectation. I understand that perhaps there is no
such connection and price setting by the bookmaker is only concerned
with evaluating the collective investment of punters in each outcome.

In Response To Les Corbett:

My own, somewhat ignorant position on the matter, although I am
informed by readings from different economic texts and academic
papers, is that probabiliy is frequently used as a device for
describing even very complex statistical phenomena. To give two
examples:

Paul A. Samuelson in "Proof That Properly Anticipated Prices
Fluctuate Randomly"

He uses a price model for describing the price stream of a single
good which has at its heart a conditional probability function

The Researchers Ali (1977), Asch,Malkiel,Quandt[1982] and Snyder
[1978]

These researchers used a technique wherein the objective
probability of Horses were assessed by grouping them according to
their odds (assuming the resultant group is comparable) and measuring
their actual performance.

I do not doubt that the device of probability bears fruit even when
put to use for complex phenomena like the stock market or horce
racing. I do however take your point that most punters, even
successful ones, probably do not go to such lengths to evaluate
probability and simply use broad strokes to yield 'chances' for
themselves when making investment decision. In light of this, I wish
to defer to you as you probably have more to teach me than vice versa.
Thus, the question:

what methods do you use to choose good deals and how successful are
you?
(please be as detailed as you feel comfortable in answering this
question, if you wish to do so)

In response to John Gee:

The fact that quoted odds may closely correspond to real
probabilities is troubling. This correspondance is in some academic
studies (I will find references for those interested) used to measure
the effiency of a betting market. As an example, one such study found
a close relationship between actual price spreads and those predicted
by Vegas Betting lines in NFL games. My understanding of 'efficiency'
(albeit very limited) would suggest that these are not good markets in
which profiteers should participate. Do you John, have references to
support your claims? These would be very helpful.

Post by Paul Robinson
Hi Gosen,
In soccer betting at least the quoted odds is a very good indicator of
the true odds of a teams chances.If you gather up all the 6/4 home
chances over a decent period you find they win about 38% of the
time,which is close enough to the "expected" 40% to show that the
books know what they're doing when it comes to setting prices.
Even money shots win 47% of the time,4/7 shots 58% & so on & so on.
You could I suppose use the difference between the actual & observed
to calculate an overround of sort for certain price bands.
It's not really surprising that initial soccer odds are so
accurate.There's loads of data going back seasons,teams are always
trying unlike horses & there's very little inside information known
only to a few people.You can even get a pretty good idea about how
much effect a star players absence is going to have on a result.
With 4% overround per outcome in 112% books that's a fairly large
comfort zone.And don't forget many of the popular selections are often
tied up in accumulators(& that multiplies the book's advantage),so
balancing the book on a single game isn't really a priority.You might
even offer a relatively generous price on Man U beause you know you'll
get lot of failed accas that include the reds as a starting point.
At the end of the day they're still offering 6/4 about a 38% chance
even if now one backs any of the other two outcomes.
J

john gee

2003-12-27 21:24:45 UTC

On 24 Dec 2003 14:39:49 -0800, ***@aussiemail.com.au (Gosen) wrote:

Hi Goosen,
most betting markets are fairly efficient.

Over a dozen or so seasons of English soccer if you bundle the odds
together then longterm,groups of odds win slightly less than their
odds imply they should.And as soccer odds have historically(pre
internet anyway) very rarely been changed once issued,it would seem
that bookies are very good at knowing what the 12X odds should be.

Same with horseracing,group enough even money shots together to get a
decent sample size & their win percentage tends towards a bookie
friendly 48 or so percent.

And as you point out the NFL spread quotes tally well with actual
results in their usually 10/11 each of two market.

That's not to say individual prices carn't be out of line.Soccer's
probably the least likely candidate.There's full media coverage on
virtually all teams(lower league Scottish games are least covered),the
books have a better idea than most punters about how significant is an
absence of a star player & team's are always playing to win.

Of the three mentioned,racing's the most likely to find value.Horses
aren't always trying & inside info is at least a possibility.

Well exposed horses in high class events are probably racing's
equivalent of a soccer game.No surprises,reasonable overround per
runner.

Large fields of unraced/under exposed horses is where the bookies will
get twitchy.Depending on their attitude to risk verses reward these
are the races where you might see the outsiders prices shaved(25/1
shots that should be 100/1 shots in case one of them turns out to be a
real 4/1 shot).The option then is to increase the overround per runner
& risk taking very few bets or push out the prices of the shorter
priced runners(but not hopefully to a value price).

Every horse race can be subtly different,but the clues are there.

NFL's different again.Because the tendancy is to change the spread as
opposed to the price,then longterm the spreads have to be correct.Move
spreads too often & too far & if the margin of victory ends up in the
middle of the opening & closing spreads then the books can lose twice.

I'd say the factors that influence pricing of most betting events are
in order of importance,true odds,weight of public money,with shrewd
money a poor third.

Fascinating subject.

J.

john gee

2003-12-23 20:09:35 UTC