How do you compile odds for a tennis match?

[loco23]

Hi Guys, I'm hoping i can get a good discussion started about this subject, and at the same time bring my own ideas of wisdom. Compiling odds by yourself is obviously a great thing to do, especially if you can do it accurate enough to find value. I realise there are people who have been ex-odds compilers at UK bookies and then left their jobs to become full time punters (why not heh?), most likely because they have the right probabilities and techniques learned from working at the bookie to compile and look for weaknesses in bookmakers themselves. To skip to the chase here i'm on a voyage of trying to find out the best way to price up a tennis match (pre-match), and although i have had moderate success in doing so, i believe the hardest thing to price up is the rankings between the players, and the advantage the ranking has on the matchup. Shown below is a spreadsheet i have worked on over the past week. It's fairly simple, and after many attempts it's the best one i have managed to make. To use the spreadsheet you simply enter your data into the blue boxes, everything else is simply calculated for you. The way the spreadsheet works is for career, surface, head to head and last six matches, it takes the wins and losses and simply finds percentage of wins for that criteria. For instance, if one player had 100 wins and 20 losses his win percentage would be 80% in that particular criteria. At the end of entering all of the data into the boxes, the spreadsheet simply tots up all the percentages and brings out the probability and odds for both players. When using the spreadsheet you should start with the favourite player (or the highest ranked) and put that at the top half of the spreadsheet and the underdog at the bottom half. The reason it does this is because my formula (or math) for the spreadsheet simply finds the % difference in ranking and times it by 0.1. Again this is probably the weakest part of the spreadsheet, and although most of the time it does price up matches fairly accurately, it's probably not as accurate as i'd like. The main reason being for this is that it manages to price up closely ranked players well, but the wider the margin between the players, the less accurate the odds are, but maybe someone can help with this. If anyone has any questions feel free to post here, but the main aim of this thread is for me to get other ideas on how i can improve the compilation of odds, especially when taking ranking into consideration and how it can effect the pricing up of a match. The file was created using open office but might open in excel too. http://uploaded.net/file/em81dj9r

[loco23]

Re: How do you compile odds for a tennis match? Does anyone have an answer for this? what is a good basic way to compile odds for a tennis match?

[CzechPunter]

Re: How do you compile odds for a tennis match? I've never liked statistical approaches to individual sports, so I can't help you out here mate. I am not saying it can't work, but I think that you are prone to missing the most important details (injuries, motivations, etc.) with this particular approach.

[loco23]

Re: How do you compile odds for a tennis match? The thing is all bookmakers will use stats to create the starting prices, unless they know of an injury days or even weeks before the match is about to start. I'm just wondering how they implement ranking into the equation so to speak? I can compile odds fine between two evenly matched players, ie Tsonga and Murray, but when it comes to pricing someone who is ranked 237 and someone who is 783 it gets a bit more tricky. I'm just really confused. For instance, take these two examples. One player the other day was ranked 119 and the other 720, yet the price was 1.80 and 1.90 for both players, meaning each have an even chance. Looking at the stats it appeared that they were fairly even but the rank difference is massive. Surely this bigger difference in rank would mean that one player would be priced more along the side of 1.10 to 1.20 rather than evens? This is what confuses me.

[CzechPunter]

Re: How do you compile odds for a tennis match? The simple answer seems to be that they don't go according to rankings ONLY, but that they are more knowledgeable than that.

[TopJimmy]

Re: How do you compile odds for a tennis match? Out of interest, what were the two players you mentioned (ranked 119 and 720)? Compilers will vary in how mathematical their approach is. I know some who don't take a statistical approach at all; like a lot of guys on this forum they watch a lot of tennis, know most of the players pretty well (certainly top 150 or so of both sexes) but even the players they don't get to see much they are following their results week in and week out. So they can price up most matches reasonably well just on feel considering ranking, recent results and head to heads. If other firms have gone out with their prices they would also then check to see what the market consensus is to make sure they offer no large ricks, and decide which matches they want to be top or bottom price on. Once prices go out weight of money coming in and market movements will then lead to tweeking the odds up or down. Personally I like a more mathematical approach and so consider the game in terms of serve percentages i.e.calculating the expected percentage of points each player will win on serve. With these numbers you can then build spreadsheets to predict odds of winning games, sets etc. This is also the basis of all in-play tennis algorithms used by online bookmakers.

[TopJimmy]

Re: How do you compile odds for a tennis match? I would also add that I think you're on shaky ground if you try to use rankings alone, particularly once you get down into the hundreds. First of all rankings down at that level can be so volatile, players can move hundreds of places through periods of inactivity/injury so the number alone isn't always a true reflection of ability. Secondly you get a lot of surface specialists, some of the guys down there play exclusively clay courts and wouldn't know their way round a hardcourt so you need to be aware of that. Also, the difference in standard becomes less as you move down the rankings e.g. I would give the world number 600 a better chance of beating the 200 than I would the world number 50 beating Novak Djokovic. Don't ever catch yourself rationalising a bet on a short odds on favourite simply because he's ranked 100 places higher than his opponent, because if we're talking 400 and 500 there's likely nothing between them!

[loco23]

Re: How do you compile odds for a tennis match? Yeh i agree about ranking, and have discovered it usually means little in terms of pricing up a match. I can't remember what the players were but it was two females competing in an ITF match. I see you mention about pricing up a match using serve percentages etc; could you clarify exactly how you do that at all please?

[milkovich]

Re: How do you compile odds for a tennis match? First of all you dont have to consider stats, also the odds are adjusted to players performance in the past months ( also they add more things, like grand slam winner, multi-tournament winner etc), so for example if Ferrer recently won a tournament and played 3 matches in the next and Dolgopolov has 4-5 odd you absolutly dont think that taking Ferrer is a sure bet, you have to consider his age, how many hours played in last 7 days, these are the important things when predicting a tennis match.

[Apple_Pie]

Re: How do you compile odds for a tennis match? how I get the odds? I do some researches+ I always bet on teams that I know. I'd say informing yourself a lot and comparing a lot of facts and predictions is the main ingredient.