The Importance Of Value

[Outlander]

What is the final determinant of value , after all the boffins and syndicates and insiders have had their say in the market , I imagine betfair closing SP is pretty accurate, that is the number that takes into account the most complete wisdom of crowds .
If you look at the data it’s pretty conclusive that the BSP correlates tightly to actual probabilities.
So if you obtain a price that beats Betfair SP you have obtained value I would say that is the best measure, but I could beat BSP all day long and lose all day long, I guess I’m a special case.

 

[Outlander]

I often feel it’s a great injustice that the more we may think we know then the harder turning a profit becomes. Fortunately i am still proving myself wrong in this respect but the feeling remains. I do not see obtaining value as the holy grail but it is a need if only to help swerve a years end loss. ! Your above beef is about not being able to back winners but a punter can back plenty and yet still show a loss, and under these circumstances obtaining value or failing to do so is prob the straightener.

What you’re saying is true , as a 18-21 yo in the early nineties , could back winner after winner. Maybe I know too much now, maybe ive gone stale.
In fairness I don’t bet U.K. anymore and little dabbles on Hong Kong races do make a small profit , so might be doing better than most.
But the last few years I was betting on U.K. racing , it was uncanny how many well thought out value bets I could make look like they were running in quicksand

 

[Outlander]

What is the final determinant of value , after all the boffins and syndicates and insiders have had their say in the market , I imagine betfair closing SP is pretty accurate, that is the number that takes into account the most complete wisdom of crowds .
If you look at the data it’s pretty conclusive that the BSP correlates tightly to actual probabilities.
So if you obtain a price that beats Betfair SP you have obtained value I would say that is the best measure, but I could beat BSP all day long and lose all day long, I guess I’m a special case.

Here’s a question if you had the Betfair SPs the day before racing but not the results and were free to bet without restrictions , would it be impossible to lose ?
I reckon it would be impossible, if you had those final probabilities you would have everything you needed to go wild, I imagine this is why people spend a lot of time and effort trying to get close to the perfect probabilities.

 

[mick]

O Outlander my own thinking is more recent with a contributing factor being the current Bookmaker attitude that customers who turn a profit or even look like they might will not be tolerated. !

 

[Outlander]

Here’s a question if you had the Betfair SPs the day before racing but not the results and were free to bet without restrictions , would it be impossible to lose ?
I reckon it would be impossible, if you had those final probabilities you would have everything you needed to go wild, I imagine this is why people spend a lot of time and effort trying to get close to the perfect probabilities.

Obviously this is a silly question as you could just trade on betfair back higher than BSP lay lower, but forgetting trading I really mean would it be impossible to lose just backing horse at any prices you could get with the bookies that bettered BSP

 

[markfinn]

Well from last night study i have 5 horses that i think have chance today and noted them down and indeed bet them.
from last night till today there odd have went like this. 11/2 now 9/2- 6/1 now 6/1- 11/4 now 7/2- 3/1 now 4/1 - 4/1 now 4/1.
Looking at it only one has came with sway of money two have drifted and two have stayed same.
So you could argue i have got value in the first one where it has been shortened since i placed bets.
But the two i personnally thought had the better chances overall are the two who stayed the same price and possibly one drifter.
So now were in the dilemma of deciding whats value some will say the one where money has come for now but i didnt know money would come for it when placed bets.
Some will say the drifters as now getting better odds.
Some will say the two who stayed the same as i already said there the two i was most hopefull of in accessing races.

but the inherent value of chance ,was your in opinion available when you bet - so drift or plunge - win or lose - you still have had your notional value - I find it hard to see any other point.

 

[Outlander]

O Outlander my own thinking is more recent with a contributing factor being the current Bookmaker attitude that customers who turn a profit or even look like they might will not be tolerated. !

Yes the bookies did me a favour I reckon, was in profit with most of them but was starting to feel it was getting harder , then one result they didn’t like and banned or restricted. It’s obvious they were all in full receipt of by betfair data by this point and knew I was in big profits on the exchanges, they made it impossible for me to even bet but I genuinely believe if they had left me alone I would have given a lot back.

 

[OnTheNod]

I think long term elements of many/several factors combined is best, and as from as many angles as possible also and never be to narrow in ones outlook, then maybe run an average price test and work from that stat to help things along.

Frankel was a one off and a few in various sports are, but it can prove hard sometimes to get things right even with these sort of events occurring i have found.

 

[Outlander]

For the last few years I've been going down the rabbit hole of determining value in terms of creating a range of horse race ratings, converting them to a tissue price and comparing with the live odds on offer. The concept is a simple one, but in itself can be frustrating when an obvious winner escapes because it wasn't a value price. The flip side is that the ones that are too short, despite having a stand out chance, bet beaten and a bet is saved.

The main reason for pursuing the value concept is the satisfaction of digging out a piece of information not known or not obvious at first glance, that can skew the ratings favourably and in turn create some value.

The importance of value is what is used to crunch the data/numbers and produce an output/outcome that is ahead of the market/public opinion. If the data is useless, outdated, inaccurate, then the output will be useless, even though you might get lucky here and there.

Ego can be a issue too. It the information isn't cross checked and examined, even the winning ones, then there's a misguided belief that the groly hail has been discovered and the rest is history.

Football is a big subject and if you find a niche then there will be plenty of markets to bet on for value. I like to look at football data and various websites during the season and still get surprised at some of the upsets when an away team out of form, beat an in form home team. In essence, it's likely the away team became value due to the home team being overbet. Would I have bet the away team? No, as a football fan, yes if I was a hardened value seeker over 1000's of bets.

I also play poker. You hear terms, "betting their hand for value". Implied odds and reverse implied odds. One player betting a value hand is another players showdown hand without putting another chip in the pot and willingly give up extracting value for a safer route. In the short-term, it's probably okay, but if it's a long-term game, it will impact the bottom line of a winning player and turn them into a losing player. The same can be said of horse racing and football betting. So I suppose that's the true importance of value.

Went down the exact same rabbit hole, converting ratings to probabilities, had a excel trigger sheet that backed every horse at a certain buffer over my probabilities at a couple of times before race start with Betdaq(betfair premium charged me out of their exchange), modest stakes set it up go to a pub garden with my wife and son, come home see the nice profit every day virtually for the first month, loving it . Profits then wiped out in about the next 10 days then started to get worrying so I pulled the plug.
Obviously I had an amateurish set up, obviously my ratings or even more likely my probabilities were not good enough.
It felt like I had the holy grail in that first month, but then I was reminded i‘m just a slightly dim bloke who must have been mental to think he had the wisdom of crowds beaten.

 

[ARAZI91]

value of the holy grail of betting, perfect probabilities are impossible , the closer you can get , the closer you get to the holy grail.
Im never going to get close, most of the time when I spot value and bet (confirmed by subsequent market move) , the horse still manages to struggle home in last place. At that point I feel was it really value, doesn’t feel like it, where did I go wrong, then I realise I didn’t factor in the weight of my money on the horses back. To be honest this is the most important factor in determining any horses chance, I could get Frankel beat.

But within PariMutuel like HK or exchange based betting like Betfair you don't need to get to the holy grail , you only need to be better than the majority of your competitors - the majority of other punters - within that type of betting which is really termed PTP("peer to peer") your real competition is not the market maker, they are just the middle men taking their "cut" regardless of who wins or loses

 

[ARAZI91]

Went down the exact same rabbit hole, converting ratings to probabilities, had a excel trigger sheet that backed every horse at a certain buffer over my probabilities at a couple of times before race start with Betdaq(betfair premium charged me out of their exchange), modest stakes set it up go to a pub garden with my wife and son, come home see the nice profit every day virtually for the first month, loving it . Profits then wiped out in about the next 10 days then started to get worrying so I pulled the plug.
Obviously I had an amateurish set up, obviously my ratings or even more likely my probabilities were not good enough.
It felt like I had the holy grail in that first month, but then I was reminded i‘m just a slightly dim bloke who must have been mental to think he had the wisdom of crowds beaten.

Just using ratings won't work O Outlander you need metrics, measurements that cover the broad spectrum of racing, and usually now an ensemble of models(many)- a well calibrated final probability line is a combination of many singular probabilities derived by many different models but i admire anybody that even tries to set up something like this as most punters don't.

 

[ARAZI91]

What is the final determinant of value , after all the boffins and syndicates and insiders have had their say in the market , I imagine betfair closing SP is pretty accurate, that is the number that takes into account the most complete wisdom of crowds .
If you look at the data it’s pretty conclusive that the BSP correlates tightly to actual probabilities.
So if you obtain a price that beats Betfair SP you have obtained value I would say that is the best measure, but I could beat BSP all day long and lose all day long, I guess I’m a special case.

I agee - Betfair Sp converted to a probability and then multiplied by a 100 = win % - those win %'s will be very very close to outcomes at win% when looked at over the whole spectrum of incremental odds, ie Decimal 3 shots will win around 32.82% of the time - BUT that's on a macro level - nobody i know bets every horse in every race , so on a micro level there is value in nearly every race as the laws of bookmaking still apply - if one horse in a race is overpriced (and their is virtually no margin - 100% book) there MUST be horses in that same race that are underpriced. All of these "efficient market" studies are done lumping all the data in together -punters don't even think like that never mind bet like that.

 

[ARAZI91]

Went down the exact same rabbit hole, converting ratings to probabilities, had a excel trigger sheet that backed every horse at a certain buffer over my probabilities at a couple of times before race start with Betdaq(betfair premium charged me out of their exchange), modest stakes set it up go to a pub garden with my wife and son, come home see the nice profit every day virtually for the first month, loving it . Profits then wiped out in about the next 10 days then started to get worrying so I pulled the plug.
Obviously I had an amateurish set up, obviously my ratings or even more likely my probabilities were not good enough.
It felt like I had the holy grail in that first month, but then I was reminded i‘m just a slightly dim bloke who must have been mental to think he had the wisdom of crowds beaten.

O Outlander - You need to stop putting yourself down Sir, looking over some of your past posts the other night and i see an intelligent punter not afraid to be a contrarian within the domain of time and speed figures and you had some tremendous ideas - found an old copy of your old standard times and they were excellent, from a proportional point of view most correlated to around 0.98 + - which is top notch.

 

[DuckandDive]

Went down the exact same rabbit hole, converting ratings to probabilities, had a excel trigger sheet that backed every horse at a certain buffer over my probabilities at a couple of times before race start with Betdaq(betfair premium charged me out of their exchange), modest stakes set it up go to a pub garden with my wife and son, come home see the nice profit every day virtually for the first month, loving it . Profits then wiped out in about the next 10 days then started to get worrying so I pulled the plug.
Obviously I had an amateurish set up, obviously my ratings or even more likely my probabilities were not good enough.
It felt like I had the holy grail in that first month, but then I was reminded i‘m just a slightly dim bloke who must have been mental to think he had the wisdom of crowds beaten.

You've likely heard of Bill Benter. He started off with 16 variables within his betting/syndicate model, after early promise, blew the bank. His second attempt, took in 120+ variables. (are there that many?) The main difference was the market (ISP) was not previously factored into the first model.

Unless you're super connected to all the stables and key information at hand at the click of a button before the race (some bookies think they do) the extra info won't be known until the result is in and accounted for. BB would re-calculated and measure the % of difference and use this as a guide going forward. A market move steam/drift pre-off would help, but it could be based on a false flag and skew the model.

In the example you gave, did you spend the time reviewing the winning results and comparing them with the losing results as a whole?

 

[DuckandDive]

Here’s a question if you had the Betfair SPs the day before racing but not the results and were free to bet without restrictions , would it be impossible to lose ?
I reckon it would be impossible, if you had those final probabilities you would have everything you needed to go wild, I imagine this is why people spend a lot of time and effort trying to get close to the perfect probabilities.

The Betfair SP for the whole book is usually around 102% compared to the ISP of 115% or so. (more in bigger fields)

If I knew the BSP pre-off and say is was around a 110% overround, knowing it was a truer reflection of the wisdom of the crowds, by narrowing the field and creating a book within a book, yes it's possible to win long-term. Factor in compounding the bank from profits only, rodders next year....

 

[ARAZI91]

You've likely heard of Bill Benter. He started off with 16 variables within his betting/syndicate model, after early promise, blew the bank. His second attempt, took in 120+ variables. (are there that many?) The main difference was the market (ISP) was not previously factored into the first model.

Unless you're super connected to all the stables and key information at hand at the click of a button before the race (some bookies think they do) the extra info won't be known until the result is in and accounted for. BB would re-calculated and measure the % of difference and use this as a guide going forward. A market move steam/drift pre-off would help, but it could be based on a false flag and skew the model.

In the example you gave, did you spend the time reviewing the winning results and comparing them with the losing results as a whole?

Very easy to get 120 variables - start with say 20 variables and factors, then layer the data , LR,L5R,L10R ,Career, Split-Half Career1 and Split-Half Career 2 - that's 120 numerations right away. Add in Recency weighted LR,L5R,L10R,Career, Split-Half Career1 and Split-Half Career 2 and your up to 240.
If you want to know a little about how Benter worked see some of my input in the "Bookies and Team Ratings" thread under the "football" section - i have corresponded with ex-Benter and ex-Woods employees.

 

[Outlander]

You've likely heard of Bill Benter. He started off with 16 variables within his betting/syndicate model, after early promise, blew the bank. His second attempt, took in 120+ variables. (are there that many?) The main difference was the market (ISP) was not previously factored into the first model.

Unless you're super connected to all the stables and key information at hand at the click of a button before the race (some bookies think they do) the extra info won't be known until the result is in and accounted for. BB would re-calculated and measure the % of difference and use this as a guide going forward. A market move steam/drift pre-off would help, but it could be based on a false flag and skew the model.

In the example you gave, did you spend the time reviewing the winning results and comparing them with the losing results as a whole?

I did spend a bit of time trying to work out why it wasn’t working how I hoped, there seemed a few races where one horse was just a lot lower in the final odds than my ratings probabilities, forcing my trigger sheet to back multiple horses in opposition. In these instances the market seemed to know something my ratings could not predict. I was concentrating on handicaps in the hope there would be less of these scenarios, but there where simply to many races where the market knew more, not surprising really, I must have been over optimistic in thinking I could beat the market with such a basic set up.
I guess I could have spent time mitigating against this, but I was back betting full time in-running and didn’t have the time to devote to it, as ARAZI91 says my approach was too simplistic so it almost certainly would have ended badly anyway. At the time I didnt think to factor in the market, as I thought that was defeating the objective of finding those horses underestimated in the market.

 

[DuckandDive]

Very easy to get 120 variables - start with say 20 , then layer the data , LR,L5R,L10R ,Career, Split-Half Career1 and Split-Half Career 2 - that's 120 numerations right away. Add in Recency weighted LR,L5R,L10R,Career, Split-Half Career1 and Split-Half Career 2 and your up to 240.
If you want to know a little about how Benter worked see some of my input in the "Bookies and Team Ratings" thread under the "football" section - i have corresponded with ex-Benter and ex-Woods employees.

ARAZI91

Will take a look.

 

[Outlander]

T

I did spend a bit of time trying to work out why it wasn’t working how I hoped, there seemed a few races where one horse was just a lot lower in the final odds than my ratings probabilities, forcing my trigger sheet to back multiple horses in opposition. In these instances the market seemed to know something my ratings could not predict. I was concentrating on handicaps in the hope there would be less of these scenarios, but there where simply to many races where the market knew more, not surprising really, I must have been over optimistic in thinking I could beat the market with such a basic set up.
I guess I could have spent time mitigating against this, but I was back betting full time in-running and didn’t have the time to devote to it, as ARAZI91 says my approach was too simplistic so it almost certainly would have ended badly anyway. At the time I didnt think to factor in the market, as I thought that was defeating the objective of finding those horses underestimated in the market.

Looking back the only surprise was how well it did in that first month, I think I only came back to about smallish 3 losing days and won a nice amount of money nearly every day, it changed almost overnight and started losing every day.
Was playing with real money so stopped before it became a problem, maybe it was just statistical noise and if I had bigger balls would have been ok if I had carried on, but I pretty much know that wasn’t going to be the outcome.

 

[DuckandDive]

I did spend a bit of time trying to work out why it wasn’t working how I hoped, there seemed a few races where one horse was just a lot lower in the final odds than my ratings probabilities, forcing my trigger sheet to back multiple horses in opposition. In these instances the market seemed to know something my ratings could not predict. I was concentrating on handicaps in the hope there would be less of these scenarios, but there where simply to many races where the market knew more, not surprising really, I must have been over optimistic in thinking I could beat the market with such a basic set up.
I guess I could have spent time mitigating against this, but I was back betting full time in-running and didn’t have the time to devote to it, as ARAZI91 says my approach was too simplistic so it almost certainly would have ended badly anyway. At the time I didnt think to factor in the market, as I thought that was defeating the objective of finding those horses underestimated in the market.

Depends what you mean by too simplistic.

ARAZI91 just posted up ways that by layering the data, 120+ variables can soon multiply. I'll take a look at the thread he suggested, but my initial view is that some will contradict or maybe not have enough data sets, I'll keep an open mind. Bill Benter was/is different gravy, maybe ARAZI91 is the next generation.

I've found the best approach (maybe simplistic) is one where there are little or no conflicts when grouping data and producing a ratings subset. ie: Weight/Speed. Going/Time..etc What the excel sheet triggers from this information, although widely known information, can be illuminating. I then step outside the ratings and see if there are other factors not accounted for. ie: Headgear, New trainer..etc

I reckon if you revisited your excel sheet, with data that's freely available and a forum like this, you might be nearer than you think.