Quantitative Finance Stack Exchange is a question and answer site for finance professionals and academics. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

On a betting exchange the price (the odds that an event will happen expressed as a decimal, 1/(percentage chance event occurring) of a runner can experience a great deal of volatility before the event in question begins. Since very little about the event in question actually changes before it starts the movement in price must be down to pure market forces. This is especially true in the minutes leading up to the start of the event. A prime examples of this are the ten minutes before the start of a horse race or two minutes from the start of a greyhound race.

It is possible to monitor the market in real time (including the currently best available back and lay prices, and the amount of money it is possible to back or lay). Given all of this, what devices from quantitative finance could best be used to predict the immediate movement of the odds of a runner? As with a predictive model of price movements it is possible to trade on a betting exchange as you would a normal stock exchange.

share|improve this question
My suspicion is that some bettors intentionally wait until the "last minute" to bet, and the volatility change might be primarily due to this. Do you have any real-time volume-of-bet numbers? – barrycarter Mar 13 '11 at 2:07
I don't have those numbers, but I do know that betting markets are more volatile just before an event than at any other time pre-play. I also know that people trade these markets just like a stock exchange. Take a look at one of today's horse races 10mins before the off on betfair/betdaq to see just how much the market moves. – Peter Mar 13 '11 at 10:31
Couldn't you use a GARCH-model to predict volatility? Is there a way to trade "straddles"? – Owe Jessen Mar 17 '11 at 10:01

Have you looked into "noise trader" models? This seems like a market that is mostly noise. A few betters may have some information on or real knowledge of who might win, but certainly nothing like equity markets where there are a lot of people who really know the ins and outs of the firms they're trading.

The classic model is Pete Kyle's, which should give you some of the intuition, although here I don't think there are an "liquidity" traders -- just noise and informed. This model isn't very applied, but the conclusions should help you develop your picture and you can probably find a lot of working papers in SSRN.com that apply the model in a way that you can use...

I found the Kyle paper, but it's Econometrica, so it may be a little dense for an introduction. This textbook chapter may be a better introduction to see if it's something you'd like to pursue.

share|improve this answer
Do you mean Albert Kyle? – shabbychef Apr 28 '11 at 23:23
I think "Albert Kyle" is his given name, but people always refer to him as "Pete Kyle" (I certainly don't know the guy personally). – Richard Herron Apr 29 '11 at 1:36
Kyle's bio at UMD – Richard Herron Apr 29 '11 at 1:37

Increased volatility towards the event start is definitely from increased order flow. There are some papers specifically on "prediction markets", the ones with practical applications are on market making which I suspect is generally a loss-making operation conducted by the exchanges themselves when a market is opened. Given the short-time periods and small ranges you probably want to be looking at market microstructure and HFT.

share|improve this answer
Could you provides links to the papers you refer to (at least some of them)? – SRKX Sep 28 '12 at 15:22

I'm not familiar with betting exchanges but what you're describing reminds me a lot of binary options, where the payout is 1 if something occurs (ie APPL trades above 1000$) and 0 if the event doesn't occur. So at any time before expiration, the option is worth the probability that the even will indeed occur, so the options is always worth between [0, 1]. Although, option pricing has nothing to do with path predictions, it has everything to do with volatility pricing, which to me seems perfect for what you're describing.

On a betting exchange the price of a runner can experience a great deal of volatility before the event in question begins

share|improve this answer
Ok, so would it be more appropriate for me to ask another question about volatility pricing? Also, I'm not sure if betting exchanges are comparable to binary options (I could be totally wrong, I have no experience here :)) as your payout is either x * odds of event or -x depending on what occurs. – Peter Mar 15 '11 at 9:30
The outcome is binary, and you can interpret the odds of the bet as the implied probability (1/decimal odds). Don't think volatility pricing is useful here, if you want to price the outcome you'd have to compile statistics and model. But I suspect that's not what you're after. – jokerjoe Sep 28 '12 at 4:31

Timeserieses of pre-play betting markets look quite different than classical stock markets or forex-like stuff. Monitoring exchanges like betfair, the volume kind of skyrockets in the last 15 minutes before the start. The long hours before the race start there are basically no bets matched and the bid-ask-spread is enormous. In the last minutes there is ten fold the volume traded then in the whole hours before. Price fluctuation increases in amplitude and frequency.

Given that, applying classical WMAs for example won't work. To make it more comparable with its predecessing timeline, probably some kind of constantly increasing sample ratio could help.

Since very little about the event in question actually changes before it starts the movement in price must be down to pure market forces.

True. Nevertheless, statistic shows, that the final market price reflects the win probabilities much better than the price the hours before. So if you want to predict short term market movements, you should also consider some final estimate for the odds and the estimated winning probability to give you some idea, where the travel will go in the long run.

Besides that - it's close to a random walk. Trying to find an edge, it helps, that the fees for trading are calculated on the total result of your bets on a race market. Nevertheless, it's a lot, so at least we couldn't find any useful edge there. Good luck on that!

share|improve this answer

I have studied what happens in the Betfair horse racing markets in the few minutes prior to the off and without knowing anything about racing or markets for that matter I trade buying and selling fast and making small profits/losses as I go. The key is discipline, accept small losses when they occur and NEVER follow a trade into play. Although I trade only on instinct I make more correct decisions than not however the move from smaller stakes to larger stakes is very daunting as I am sure emotions will start to creep in as the amounts of MY money become larger and therefore the risks greater.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.