How Bookmakers Create their Odds, from a Former Odds Compiler
This post originally appeared at www.daily25.com and has been reposted here with the permission of Steve from the Daily25 blog. It is written by Matthew Trenhaile who has worked as an odds compiler for many years and is now out on his own taking on the bookies. You can follow Matthew on twitter @CrazedAlchemist and he also has his own blog. Over to Matthew.
Matthew’s Background as an Odds Compiler
My name is Matthew Trenhaile and I worked as an odds compiler for six years at the UK Sport Spread Betting division of IG Index. Steve has very kindly allowed me to use his blog to showcase some of my writing and I hope to do that in a series of articles from the perspective of someone who has worked in the betting industry. Firstly I would like to go over some of the phrases I will use in my articles as the names for various things change from country to country and era to era. When referring to a favourite I will use the word “Jolly” and “Rag” for underdog, these are very much UK phrases to my mind. To describe smart punters I use “Sharp” and “Square” for losing punters. These are very much Vegas words and will also be used to describe bookmakers which I may abbreviate to “Books” on occasion. For accounts in another name used for putting on bets I rather like the term “Bowler” I assume this is an Australian phrase and I rather like it. When describing odds I will use decimal odds and when talking money it will be in UK pounds and pence just for my own ease. I am seeing more and more articles by industry insiders not to mention television documentaries and they all seem to generate a lot of heated debate. I am all for this if it is good natured but do please remember these are just my opinions and I happily respect those belonging to other people.
An Evolving Betting Industry
In each of the following articles I hope to tackle the subjects from a past, present and future perspective. First I will look at odds compiling and how it has changed over the past fifteen years and also try to tackle the subject of bookmakers copying prices off each other, which seems to be a hot topic. My perspective on odds compiling comes from spread betting and this is important because the spread betting industry is responsible for every significant change in odds creation of the last 25 years. For those of you unfamiliar with spread betting please go to Sporting Index’s website and go to their training section to get an idea of what is involved. You could oppose an outcome in sports spread betting before you could lay on Betfair and you could bet in-running online on the spreads before any of the fixed odds bookmakers. It was also IG Index who first invented the dreaded close out button on the doomed to failure fixed odds betting product Extrabet. While none of this makes me particularly good at odds compiling compared to another it does mean I was constantly at the forefront of technological advancements in the industry. This was true then and the sports spread betting industry still leads the way now either through employees who have gone on to devise the models of other bookmakers or through companies such as Sporting Solutions, a spin off from Sporting Index where they sell in-running prices to other bookmakers. It is with good reason that many fixed odds bettors take a look at the spread betting firms prices as a secondary check before placing their bets.
How Odds Compiling Works
Odds compiling has steadily become more and more about databases, statistics and mathematical models and less about personal experience, intuition and feel. I was fortunate enough to work with men who had watched thousands of hours of racing and could pick out the smallest of nuances about how a horse was ridden or the strategy of a given trainer however, I was also fortunate enough to work with people who were able to break down sports into their fundamental inputs and turn those inputs into probabilities both before and during events. Odds compilers were initially split on the prospects of Betfair and certainly on its uses with regards to compiling prices. I started working at a time when Betfair had just reached significant liquidity levels and could not possibly be ignored if only because of the ever increasing numbers of arbitrage bettors. Within 6 years we used our Betfair API to price up all our horse racing products with a human simply to oversee the process. Contrast this to when I started and we had a trader for each horse racing meeting and a room of 40 traders in a time when the number of sporting events priced was less than a tenth of the number now. The luxury of being a subsidiary of a large financial firm was that we were better paid than the rest of the industry and had greater resources at our disposal whether it be staff or IT support. All resources were increasingly ploughed into trading more events in-running and developing more and more complex models for generating odds in-running. These odds were generated for our clients but also so that we could provide liquidity to the betting exchanges which enabled us to generate a significant secondary revenue stream.
Creating Statistical Models for Sports
Most statistical odds compiling originated with a simple counting of how often an event happened. Two football teams, count how often the home team won at home in the last 20 games, how often the away team lost away in the last 20 games etc. This was well before my time of course and what gave bookmakers an edge back then was studying the sports more than the punter and comparative odds knowledge. What I mean by that is if a bookmaker calculated a team should win 50% of its home games in football and last time these two teams played you put up 1.83 odds and the punters backed it like crazy the bookmaker then records or remembers this and puts out at 1.75 next time and when the punters come back it again and the bookmaker has extracted extra value from his margin simply by knowing what the punters did and what they were willing to pay for it before. Odds compiling was more about knowing the punters and their habits than the actual percentage probability of outcomes. Even now this particular area is the difference between bookmaking and punting. Bookmakers have to anticipate money flow whereas the punter has to determine the probability of outcomes and find value. In an online world the odds have come to reflect real probabilities more and more and less public opinion or the narrative of the event. The emergence of in-running betting is what really drove odds compilation to mathematical modelling, it became too hard for a human to quote prices in multiple markets for multiple events in-running all with pen and paper. Bookmakers needed automation, which meant models.
The Poisson Distribution for Modelling Sports
Most sports betting models are easily found online and have been around for a while but with many tweaks and refinements over the years and ever improving levels of data. Poisson distribution won out as the way of modelling football not only because with some refinements it can be very accurate but also because it was easy to add time decay to the inputs.You have goal inputs for each team and as the match progresses those goals input in to the model slowly decrease meaning as the simulation is recalculated each second it produces a slightly different set of odds. This went from Poisson distribution to custom distributions per league and we went from working out goals for and against to shots for and against and then turning those in to goals. Now compilers are measuring the quality of the shots and the impact of individual players on the shots to create advanced player based shots models. Or are they? All sports have their equivalent mathematical model and all can be tweaked further as more data is recorded and made available. The question for bookmakers is how far do they go down that route? Do you pay to have people maintain huge databases? Do you pay someone else to provide prices? There are already several firms using the same prices from the same company for in-running football for example. Even if you did the odds in house just how good do they need to be to beat the average punter and can good (using sharp bettors positively) risk management hide a multitude of sins simply by letting your good punters move the prices in to shape for you? Sadly few bookmakers allow for that style of risk management or for the investment in better pricing. Odds compilers are finding out that one pound spent on promotions and marketing can generate a greater increase in profit than one pound spent on their salaries or in software development. Bookmakers want to spend just enough to beat the 98% of punters they need to with the minimum number of staff. Analysing in greater detail actually shows them that paying for decent profiling software and young no experience employees to operate it is even more cost effective than training or acquiring odds compilers of almost any skill level.
How Bookmaker’s Business Models and Views on In-House Odds Compiling has Changed
Maybe you are thinking that surely odds are a bookmaker’s product and surely they would be better served by developing that in the long run. This is an incorrect assumption for modern bookmaking. You do not choose which hotel to stay in by comparing the cost of a beer in the minibar. And bookmakers see the odds as about as important as the price of that beer. Their product is entertainment and not the selling of an intellectual contest between punter and bookmaker. It is foolish to think this has ever been the product that bookmakers have sold. They sell an adrenaline rush and anyone who thinks great characters pitting themselves against the punters and taking anyone on in horse racing betting rings is what betting used to be about is kidding himself or herself. Everyone has a story of how great bookmaking used to be. The people I speak to remember betting tax in the UK, huge margins, little choice between bookmakers to use, being refused payment and in fact being threatened with violence when trying to get paid out. I have no idea where this image of the gentlemen bookmaker comes from. Throughout time some bookmakers take your bets and some don’t and some allow you big bets and some don’t and some go bankrupt on you and some don’t. This has been true for both sharp and square punters. All bookmakers expect to win almost regardless of the quality of their odds (not a view held by their compilers necessarily) and better to channel their efforts in to getting punters through the door than lowering the price of the beer in the minibar. I do not like this model but I do not see this model going away.
There can only be one Pnnacle
Now for the obligatory Pinnacle paragraph. Why can’t all bookmakers be like Pinnacle? The answer to this is simple and it is that there can only be one Pinnacle business model at a time. Others can take bits of it here and there but the exact Pinnacle model can’t be easily duplicated. Pinnacle’s tagline is that they welcome winners and simply put that is because they want to beat all the square books with the greatest sports proprietary trading scheme in the world. By employing the best odds compilers, betting to tiny margins, allowing all arbitrage and sharp bettors and moving your odds in accordance with how sharp they are your odds become the truest representation of the probability of sporting outcomes. You start limits low at your most vulnerable and increase them as your price improves. You spend nothing on marketing but allow absolutely anyone to use affiliate banners and give your API out to anyone you can especially arbitrage and odds comparison services. Your low margins mean you are the other side of arbitrage trades all day and all night and the square book your price is against is so bad at pricing you must be getting the value side of the arbitrage trade. In fact you are getting it over and over again but always moving your price to capture that small bit of value every time. You become the single largest punter that Bet365, Ladbrokes, William Hill etc has and all without opening a single account. Your risk management must be largely automated and you must have the servers of an investment bank. Anyone else trying to setup this model and succeed must do so with smaller margins and better risk management and not to mention a lot of capital. The only one I see really trying this is Marathon and they have not got to the point where they will not close accounts, in fact far from it. What about all those Asian bookmakers accepting winners? Well they are happy to take them on events where they see a huge volume of square money, which are of course all football matches and some US sports and maybe decent level tennis matches. The Asian Handicap model is designed to be low margin, fast moving and high volume and the sharpness of the money once close to the off is almost an irrelevance in some cases and early on limits are low like Pinnacle while the market forms so the risk is low. They are not however running a Pinnacle model but simply have got a large enough square volume to adopt certain aspects of it and are more comfortable to have loss making areas as long as the book as a whole is profitable.
The Future of Odds Compiling
So where do I see the future of odds compiling? Ultimately I believe that odds compilation will become an entirely outsourced function for bookmakers (are they still bookmakers if they don’t make the book?). There will be specialised companies whose primary function is providing odds both pre-match and in-running. There are already several companies that do this and there have long been companies who made it their business to provide liquidity to betting exchanges. You are going to see even more homogenised pricing than you do now. The reason that you see so much “copying” of prices is that the liquidity is such now in certain sports betting markets that there is now what can be deemed a market price much like in the financial markets. In top-level football there is very little incentive for bookmakers to deviate heavily from each other. Believe me when I say that extensive studies have been done that show trading against the market in liquid sports is a loss making proposition in the long term as bookmaker, not necessarily as a punter but when you have to provide thousands of prices across so many events you are better served by following the market. There are increasingly companies that will look at the market and say that a combined price from Pinnacle, IBC, SBO, Betfair (when liquid) and Bet365 for football for example can easily be concocted from API and will be more than solid enough. These operations aren’t going anywhere and see a staggering amount of bets, which keep the prices true. So why do they put arbitrage opportunities up all the time even in liquid markets I hear you cry. The answer is we are still in a period of transition in the industry and some are still coming to terms with the idea of a market price that should just move on weight of money and not a trader’s opinion.
I actually envisage growing confidence from bookmakers in these consolidated feeds. They eliminate arbitrage or at least with any sharp sources and they are cheap to use. I think they will also outsource all risk management as well. There will always be exotic markets attached to the core liquid ones and if bookmakers have any sense they will raise limits on the core and reduce the limits on exotic stuff to 100 pound takeout which is fine for the average “amusement” bettor. As for horse racing I have no idea where that will end up, SP only maybe? I actually think that pari-mutuel betting with very low margins would be the way forward, a sort of universal Betfair SP. In Asia this kind of pool betting has not deterred punters of either the recreational or professional type. The problem with the UK is the amount of terrible racing that is on but who knows if every bookmaker pooled tickets, and Betfair and the Tote and shared the profits then maybe something of worth could emerge and it would also put pay to some of the rancid SP rigging that goes on. Punting will become about beating the market as a whole and not picking off the sick and the weak prices from bad books. This will require punters and tipsters to understand the strengths and weaknesses of the markets they tip in as much as the sports themselves. Ultimately do not concern yourself with how the price is made or whether all bookmakers are showing the same price, only concern yourself with beating it, as there will always be flaws and edges to be found even if they become smaller and harder to spot.
Written by Matthew Trenhaile
You can follow Matthew on twitter @CrazedAlchemist and he also has his own blog.