Efficiency of Horse Racing Markets on Betfair

[Kafkaesque]

I agree with weemac, the Betfair market nowhere near reflects true probability but accurately reflects the probabilities that people perceive. If a horse has a precise probability of 50% and is trading at 2.1, it's quite possible it might drift to 3.5 - it's chances are irrelevant, it's down to punter's perceptions which are often wrong. I've often seen big drifters hack-up, proving the rumours or whatever caused the drift to be wrong.

The market is volatile enough when a horses chances of winning doesn't change. Occasionally they can change massively half-an-hour before the race. For example, a change in the going, the revelation of draw bias, a sole front runner being pulled out of the race which would advantage/disadvantage some horses substantially.

What I was referring to was the price at the off (or close to it) being a reflection of true chance rather than perception, which I might not have been clear enough about. Obviously if a 50% shot trades between 2.10 and 3.50 - without it being down to new information - then the true chance reflected in both those extremes cannot be true. So yes, perception plays a role in the pricing as the market finds its level.

Also, it is an observation for most markets and the average of all markets. There will clearly be outliers, with markets that isn't very efficient, but on average is another matter. It's semantics in the end, and we might actually be as to what role perception. The outliers occur imho, because there'll always be markets here and there, where more bad judgements enter the market than the other way. Is that perception? Perhaps. But on average the sum perception and smart money will be close to correct and push the price towards true value.

I should really have mentioned that my understanding is based on sports other than horses. So I might have jumped the gun And the efficiency is different for horses, which is why it's such good trading opportunities?

FWIW:

https://www.pinnacle.com/en/betting-art ... 8P6NRD6ZYK

[Derek27]

I should really have mentioned that my understanding is based on sports other than horses. So I might have jumped the gun And the efficiency is different for horses, which is why it's such good trading opportunities?

My brother is a statistician who use to work on the statistical modules that calculate probabilities for the bookies to compile their odds. From what he tells me it is possible to quite accurately calculate probabilities for the premiership title (not sure about individual matches), and perhaps other sports. As you suggested it's not possible for horse racing, the size of the field being just one of many factors.

[LucyLoo]

Today we had the lowest weight horse heavily gambled. It seemed to have a 20% better chance of winning than when the market was formed, but it didn't and it never really came into contention. That's not very efficient.

Did you trade the steam and the reversal?

[marksmeets302]

One definition of an efficient market is one in which no one can structurally can get a better return than the risk free interest rate. This forum is full of people that make a living operating on the betfair markets.

I do think the horse racing markets are becoming more efficient. Strategies that once worked have stopped to perform well.

[Crazyskier]

I do think the horse racing markets are becoming more efficient. Strategies that once worked have stopped to perform well.

Me too

CS

[Derek27]

If by efficient you mean reflecting the true probabilities of horses winning, I don't think they are getting more efficient or ever will be.

I have a live spreadsheet that calculates the overrounds for the highest best prices available to back. If I start the spreadsheet just five minutes before the off, in the space of five minutes the market fluctuations give a back overround of around 90% or less, for most markets.

It goes without saying that when a horse is backed from 4.0 > 3.0 in just two minutes it can't have a 3.0 chance and a 4.0 chance. Whatever its chances, somewhere along the spectrum there is outstanding value.

[spreadbetting]

I have a live spreadsheet that calculates the overrounds for the highest best prices available to back. If I start the spreadsheet just five minutes before the off, in the space of five minutes the market fluctuations give a back overround of around 90% or less, for most markets.

It goes without saying that when a horse is backed from 4.0 > 3.0 in just two minutes it can't have a 3.0 chance and a 4.0 chance. Whatever its chances, somewhere along the spectrum there is outstanding value.

A lot can happen in 2 minutes , it's perfectly possible for it to have been a 3 or a 4 shot within that time frame and I'm sure if you stuck you spreadsheet on a football match you'd also get a degree of movement. For us to consider whether markets are efficient we'd simply have to look at snapshots of set times or simply check odds traded against win rates to see if the probability of winning tallied with the odds matched. If we consider in running markets, which should in theory throw up lots of errors, you'll be hard pushed to find any odds increment that is out of line with the win rate enough to profit from them, that should give you an indication of how 'efficient' markets can be as a whole. Much the same as quantum physics markets can be efficient and inefficent at the same time , as traders we can profit when we spot those inefficiencies, especially if we understand why they are occuring.

[Derek27]

I have a live spreadsheet that calculates the overrounds for the highest best prices available to back. If I start the spreadsheet just five minutes before the off, in the space of five minutes the market fluctuations give a back overround of around 90% or less, for most markets.

It goes without saying that when a horse is backed from 4.0 > 3.0 in just two minutes it can't have a 3.0 chance and a 4.0 chance. Whatever its chances, somewhere along the spectrum there is outstanding value.

A lot can happen in 2 minutes , it's perfectly possible for it to have been a 3 or a 4 shot within that time frame and I'm sure if you stuck you spreadsheet on a football match you'd also get a degree of movement. For us to consider whether markets are efficient we'd simply have to look at snapshots of set times or simply check odds traded against win rates to see if the probability of winning tallied with the odds matched. If we consider in running markets, which should in theory throw up lots of errors, you'll be hard pushed to find any odds increment that is out of line with the win rate enough to profit from them, that should give you an indication of how 'efficient' markets can be as a whole. Much the same as quantum physics markets can be efficient and inefficent at the same time , as traders we can profit when we spot those inefficiencies, especially if we understand why they are occuring.

I'd have to disagree with that. Not much can happen in two minutes in a pre-race horse racing market. If you can back a horse at 4 and lay it two minutes later at 3, in the absence of any information or anything obvious that could have changed its chances of winning, it's proof that the markets are inefficient.

I don't look at football markets but it's not a question of degree of movement, but how much ?

It obviously wouldn't move as much as a horse race.

[Dublin_Flyer]

My brother is a statistician who use to work on the statistical modules that calculate probabilities for the bookies to compile their odds. From what he tells me it is possible to quite accurately calculate probabilities for the premiership title (not sure about individual matches), and perhaps other sports. As you suggested it's not possible for horse racing, the size of the field being just one of many factors.

Euler had a youtube video on the BetAngel channel about Wage Bill v League position. General overview was the more you pay, the better you do unless you're Newcastle or Villa I think, was the nuts and bolts of the probability predictions along the same lines?

EDIT: sorry for sidetracking initial topic

[Derek27]

My brother is a statistician who use to work on the statistical modules that calculate probabilities for the bookies to compile their odds. From what he tells me it is possible to quite accurately calculate probabilities for the premiership title (not sure about individual matches), and perhaps other sports. As you suggested it's not possible for horse racing, the size of the field being just one of many factors.

Euler had a youtube video on the BetAngel channel about Wage Bill v League position. General overview was the more you pay, the better you do unless you're Newcastle or Villa I think, was the nuts and bolts of the probability predictions along the same lines?

EDIT: sorry for sidetracking initial topic

My brother's the statistician, not me, so I wouldn't have a clue how these guys calculate probabilities.

What I can say thought, is that if a horse is purchased for a massive price (similar to football club spending), this information is only useful for two-year-olds or maiden races where there isn't much form to go on, as expensive horses obviously average better than cheap horses. But for horses with proven form, the price tag becomes irrelevant as you can see how good they are. I would imagine it's the same for football. A club spending money is likely to produce better results but you would evaluate their capabilities and probabilities on the results.

[Euler]

I've measured various metrics over the years and the book percentage is shrinking, so the market is getting more 'efficient'. But the volatility is getting higher so it's less efficient. It's a sort of inefficient efficiency.

[Wolf1877]

I've measured various metrics over the years and the book percentage is shrinking, so the market is getting more 'efficient'. But the volatility is getting higher so it's less efficient. It's a sort of inefficient efficiency.

My take on that is that the "efficiency" is created by greater competition in arb-ing whilst the volatility is created by the markets being manipulated by traders and their bots.

As a newbie I've been very surprised at just how much volatility there is in the racing markets. The frequent seemingly artificial volatility makes the markets very difficult to read visually and trade manually for me anyway. The opportunity for anyone who can read/predict the likely outcomes from the volatility are great though. I do wonder how much attention I should start giving to traditional racing fundamentals (i.e. developing analysis methods of pre race form/handicaps etc) to gain a better idea of fair price for a horse so that I can profit when volatility pushes prices to unrealistic or unjustified levels.

[Euler]

You often see two equally weighted horses in a handicap with the same rating, where there is no difference between them in any respect but the prices vary widely, despite starting at the 'correct' price. Makes no sense.

[ShaunWhite]

You often see two equally weighted horses in a handicap with the same rating, where there is no difference between them in any respect but the prices vary widely, despite starting at the 'correct' price. Makes no sense.

The handicapping system isn't especially hard to fiddle, horses with the same weighting and rating are often of quite different ability.

[Derek27]

Two horses can have the same handicap rating, one could be out of form and due to draw his pension, the other could be a lightly raced sort full of potential and a stone better than his rating.