Sgt.Sunshine Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system I think this happened to the qualifiers - quite a few times this year I've looked at a qualifier in the 1st round and been surprised at how short they are:sad IMO a major problem when it comes to backtesting systems to see whether they would have made a profit is the odds you use. Unfortunately the odds from Tennis-Data are collected from Tip-Ex, and represent the last odds available. These have been "worked on" by the betting population, and we can assume that they are close to fair odds (or can we -that's another Q). Trying to get a decent yield using fair odds is not easy :sad. (There is a similar problem with horse racing using SP). What we need is a guide of opening price, or at least an early price, maybe using Betfair historic data to use in a addition to final price? Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system Can't get that info from betfair historic data, not reliably and consistently anyhow. What i'm doing with the Betfair data is determine the odds at which X% of volume was matched. Then compare that with a sample of live prices from my actual betting. That will tell me what % i may reasonably expect to have been able to get. For example the price at which 40% total volume was matched. (it'll get clear when you look at the data) Not too difficult to create some odds-collection of our own direct from the bookies. Most have xml feeds the rest is a matter of timers really. But you do get a very big data collection so you'd need some plan of how you want to use the data. For example have a column with average odds at different timeperiods or moments during the day or relative to a match start. Have a think about it. Should be able to do some "raw" data collection of odds fairly easily. The Betfair historic data does reasonable have the moment in time the match went in-play, so odds data collected over time can be related to scheduled and actual match start. In combination with a scanner on Betfair it can get pretty real-time also. Apart from Betfair and Pinnacle who would be the most used bookies for tennis ? I.e. early prices, good prices, and open to decent stakes ! Quote Link to comment Share on other sites More sharing options...
Sgt.Sunshine Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system What i'm doing with the Betfair data is determine the odds at which X% of volume was matched. Then compare that with a sample of live prices from my actual betting. That will tell me what % i may reasonably expect to have been able to get. For example the price at which 40% total volume was matched. (it'll get clear when you look at the data) Do you mean using the historic data, or real data? If X% turns out to be high, then the odds will be similar to those at the off surely? Collecting odds data is fine, except that we can't go back in time. I wrote a prog that reads the Pinnacle XML feed for tennis and stores every odds change. I haven't run it for over a year though. Can we not get anything from the BF historic data? The Volume information for each price matched is useless, but you can get a reasonable idea of how the price changed over time, or for a time-slice, say the last hour. Quote Link to comment Share on other sites More sharing options...
Mr Intensity Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system Apart from Betfair and Pinnacle who would be the most used bookies for tennis ? I.e. early prices, good prices, and open to decent stakes ! Those are definately the 'big 2'. Apart from those I'd say Sportingbet and William Hill from my experience. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system Do you mean using the historic data, or real data? If X% turns out to be high, then the odds will be similar to those at the off surely? I mean historic data as published by betfair. The odds move over time, X% of total volume matched isa reasonable approximation of what one may expect to be able to get, in my opinion based on common sense alone. You don't know if X% is simular to the odds at the off. Some prices drift, some steam, in either case the price at the off could be substantially lower or higher than X% volume matched. My reasoning is simple that if you take the price at X% volume matched that should represent the most likely price a few hours before the match. Recoding live prices in real time can also be done of course. ( i think there's an archive somewhere, paid for, but need to do a search ) but you can get a reasonable idea of how the price changed over time, or for a time-slice, say the last hour. Don't think so. that time data simply isn't provided as such. Enough on this, i'll upload the data in the database and then you'll see for yourself. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system Right, Betfair can be accessed with the free API, Pinnacle and Sportingbet have an XML feed, William Hill doesn't have a feed but looks like the webpages can be scraped fairly easily, also they have a results service which appears to contain the odds that where available. So lets say for now that an odds archive checking odds say once every 15 - 30 minutes for those bookie is feasable. Maybe even faster by only recording actual price changes, will give it some thought. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system correction, William Hill does have an xml feed available :ok Quote Link to comment Share on other sites More sharing options...
Sgt.Sunshine Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system I can program the API (BUT, not much time at the moment AND I already use up my allowance recording racing data so someone else would have to run the application using their account). I have to say that I disagree in part with what you say re BF historic data DP. Yes, taking 40% vol would tell you the price you might reasonably expect to match, but to say that it represents any particular time would be wrong, especially a few hours before a match starts (still think it is most likely to be towards the off as most volume is traded then). Also, the time info is useful to an extent. It tells you when a price was first matched and last matched. It doesn't tell you when it was matched in between, but if you plot time vs odds you get a decent idea of the change in odds over time. I would do this manually by making columns like: Col1=Time, Col2=Price FirstMatched1 Odds1 FirstMatched2 Odds2 FirstMatched3 Odds3 FirstMatched4 Odds4 FirstMatchedn Oddsn LastMatched1 Odds1 LastMatched2 Odds2 LastMatched3 Odds3 LastMatched4 Odds4 LastMatchedn Oddsn Then SORT for Time column (note that the odds data are repeated for FirstMatched and LastMatched times). Info you can get is: last price matched (by checking when market when IP) linear gradient to represent change over time (for any time period) odds at certain times before the start (eg 1hour, 15mins) etc. I agree that there is no guarantee that for any given calculation you would get that price, but this info might be useful (e.g. identify steamers/drifters etc.). Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system I have to say that I disagree in part with what you say re BF historic data DP. Yes, taking 40% vol would tell you the price you might reasonably expect to match, but to say that it represents any particular time would be wrong, especially a few hours before a match starts (still think it is most likely to be towards the off as most volume is traded then). Subject to "personal" testing :lol but yeah, time will have to tell how usefull and applicable this is. ( I may be misguided by horse racing prices where the odds ranges are much wider. ) I still think that apart from clear drifters or steamers no sensible conclusions can be made from the historic data in terms of odds movements over time. Especially if the macthed volume is not taken into consideration. Look at odds drifting from 3.00 to 2.5 then remaining fairly stable then moving back up to 2.8 Imagine the matched volume in the move from 3.00 to 2.5 is 5% of the total and the other 95% is matched around 2.7-2.9 . How significant is that early price move ? Pretty irrelevant i'd say. And this is why the X% volume matched is personal. It will be different for someone taking early prices and someone with the time to watch the markets. Should be interesting to see these different approaches ;) Quote Link to comment Share on other sites More sharing options...
Sgt.Sunshine Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system I still think that apart from clear drifters or steamers no sensible conclusions can be made from the historic data in terms of odds movements over time. Especially if the macthed volume is not taken into consideration. Look at odds drifting from 3.00 to 2.5 then remaining fairly stable then moving back up to 2.8 Imagine the matched volume in the move from 3.00 to 2.5 is 5% of the total and the other 95% is matched around 2.7-2.9 . How significant is that early price move ? Pretty irrelevant i'd say. Very true. I still don't see how you can use matched volume though, as the First and Last matched times relate to the Price not the volume, and it is impossible to partition the volume between the two (although it is fairly safe to assume that most would be matched when the price was matched Last). Anyway .... Quote Link to comment Share on other sites More sharing options...
Smokin Joe Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system On the subject of Betfair historic data, I presume that you guys are aware that with the advent of the "keep" bet you can no longer tell what the last price matched pre event was. That might not bother you, but I think it is a real shame that their data now cannot give that information. Quote Link to comment Share on other sites More sharing options...
Sgt.Sunshine Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system I haven't looked at BF historic for a while now, so I wasn't aware of the change - what do you mean exactly? Quote Link to comment Share on other sites More sharing options...
Smokin Joe Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system I haven't looked at BF historic for a while now' date=' so I wasn't aware of the change - what do you mean exactly?[/quote'] Betfair flag their bets either PE or IP (pre event or in play), this flag relates to when the bet was placed. Prior to the introduction of keep bet you could strip out all the IP lines and you were left with pre event bets. All you had to do was look at the price with the latest time matched, and that was the last price matched pre event. Now however, a bet can be submitted pre event, but it is a keep bet. This means it was appear with a PE flag. Let's assume that the price a selection was last matched pre event was 2.0, but there will also be unmatched bets at 2.0 in the system that are keep bets. It is feasible that one of these bets could be matched at any stage during the game. This means that the price of 2.0 for flag PE will have a last matched time of some stage after kick off - therefore you only know that 2.0 was first matched 3 days before kick off and last matched (say) 20 minutes into the match. Now just imagine that the same happens 1.94, 1.95, 1.96, 1.97, 1.98, 1.99, 2.02 and 2.04 etc. When you look at the raw data by just looking at lines flagged PE and you search for the last time matched (but pre event time start) you could well think that 1.93 was the last price matched pre event start. This is immensely sloppy and I have informed the powers that be at Betfair. The problem can be easily overcome by changing the IP or PE flag to relate to when the bet was matched and not when it was placed. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system haven't looked in detail yet but, the data also contains the scheduled off and the actual off, so any PE bet/odds matched after the actual off was placed before the off and matched in-play, right ? Quote Link to comment Share on other sites More sharing options...
Smokin Joe Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system haven't looked in detail yet but, the data also contains the scheduled off and the actual off, so any PE bet/odds matched after the actual off was placed before the off and matched in-play, right ? Correct. But the problem is that each price only contains 2 pieces of information - time first matched and time last matched. So as soon as a PE bet is matched after the event actual start time (ie in running) you have no idea what time it was last matched before the event started, as the time last matched will now be the after event time. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system you have no idea what time it was last matched before the event started, as the time last matched will now be the after event time Aha :loon Have you an example with PE bets showing as matched after the off ? Quote Link to comment Share on other sites More sharing options...
Smokin Joe Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system Aha :loon Have you an example with PE bets showing as matched after the off ? Yeah. Look at the data for any event that went in running - there will be plenty of such bets. Quote Link to comment Share on other sites More sharing options...
Sgt.Sunshine Posted August 14, 2008 Share Posted August 14, 2008 Re: Tennis rating/info/assessment system Bummer. Thanks for pointing that out Joe. Hopefully BF will listen to your comments; I'm sure they will, they seem pretty on the ball with stuff like that, even if it takes a while. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system 2,21096400,22-07-2008 20:35:29,Group A/Rogers Cup/First Round Matches/Canas v Soderling,22-07-2008 17:30,Match Odds,22-07-2008 18:51:12,2263597,Robin Soderling,1.08,28,13772.22,22-07-2008 20:17:30,22-07-2008 20:16:11,1,IP 2,21096400,22-07-2008 20:35:29,Group A/Rogers Cup/First Round Matches/Canas v Soderling,22-07-2008 17:30,Match Odds,22-07-2008 18:51:12,2263597,Robin Soderling,1.08,1,39.7,22-07-2008 20:16:11,22-07-2008 20:16:11,1,PE both 1.08 , second one PE but last matched after the off, so the bet was placed before the off with the "inplay keep option", so you can distinguish between in-play odds and pre-play odds, but indeed you don't know when the last match BEFORE in-play was. i can't seem to find one that was first matched before the off and last after the off, you got ? Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 14, 2008 Author Share Posted August 14, 2008 Re: Tennis rating/info/assessment system so i'm basically wondering if the occurences of first matched before off and last matched after off are maybe rare situations ? Quote Link to comment Share on other sites More sharing options...
Smokin Joe Posted August 15, 2008 Share Posted August 15, 2008 Re: Tennis rating/info/assessment system so i'm basically wondering if the occurences of first matched before off and last matched after off are maybe rare situations ? I wouldn't have said they were rare, they are common enough to give me a headache. I just pulled up a match at random and there were plenty of PE prices that were matched pre event time and after event start time. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 15, 2008 Author Share Posted August 15, 2008 Re: Tennis rating/info/assessment system which match ? so we're looking at the same thing. Quote Link to comment Share on other sites More sharing options...
mcgin Posted August 15, 2008 Share Posted August 15, 2008 Re: Tennis rating/info/assessment system Hey guys, just stumbled on this looks v. interesting would be interested in taking part. DP did you ever get that common DB up and running? If not give me a shout, I got a server you could use. Have you gotten round to collecting realtime data as the matches are in play (stats/scores/odds)? I've always considered inplay betting (on any sport) an area that is ripe for exploitation since it is relatively new. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 15, 2008 Author Share Posted August 15, 2008 Re: Tennis rating/info/assessment system Working on the DB, just need to find a bit of time to fit it in. Got a price recorder on racing that i could expand for Tennis but needs some work. Was thinking of getting into that in a few months. Haven't really got anything yet towards a pre-match odds archive if you're up for that ? Basically scanning xml feeds on timers and recording price movements. Up for that ? Quote Link to comment Share on other sites More sharing options...
mcgin Posted August 16, 2008 Share Posted August 16, 2008 Re: Tennis rating/info/assessment system Could do. I've been taking a dump of Pinnacles prices every day at about 11am since the start of June. Just need to parse it for tennis and we can use that for starters. Can extend it to be a bit more intelligent about the times it grabs, what sort of time intervals were you thinking of? Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 16, 2008 Author Share Posted August 16, 2008 Re: Tennis rating/info/assessment system Well there are 2 considerations, one is bandwidth as most hosting companies count incoming as well as outgoing bandwidth so if you poll prices every second it really adds up. But this can be solved by extending the poll time to 5 min for Pinnacle and maybe 10-15 for other bookies depending on the size of xml feed to read. The other is to keep the database managable. This we can do by only registering actual price movements. In case of very volatile movements like Betfair we can keep it managable by only noting price movements above a certain %. Like a movement from 1.25 to 1.26 we consider that same, but 1.25 to 1.28 get's registered. Bit of testing we'll be fine. I'll make a skeleton application so it can be tied in to the central database easily, got some routines from football i can re-use. If you can have a look at adding some bookies that would be great :ok Quote Link to comment Share on other sites More sharing options...
mcgin Posted August 16, 2008 Share Posted August 16, 2008 Re: Tennis rating/info/assessment system Well I've got a perl script running now that is pulling the pinnacle tennis prices every 5 minutes through the below URL. Not sure if it is the correct one to use though as it doesn't seem to be giving just Tennis, easy to change if I can figure out whats wrong tho ($last is the appropriate value from the previous request so we get incremental updates): http://xml.pinnaclesports.com/pinnacleFeed.asp?sportType=Tennis&last=$last Once you have your app up and running we can plug that in. I'll take a look at the other bookies later on. Quote Link to comment Share on other sites More sharing options...
Matthew Posted August 20, 2008 Share Posted August 20, 2008 Re: Tennis rating/info/assessment system Hi Guys, How are we looking? Any significant progress. I'm conscious that I'm not currently contributing a great deal. It's led me to think about what exactly people think this might turn in to. I know the long term aim is to come about some sort of system / ratings type for Tennis with a hope of making profit. But what will it look like? Will it be web based. Will the database calculate all the facts we need and pump out the picks along with the current best odds available? Anybody thought this far ahead? Quote Link to comment Share on other sites More sharing options...
Datapunter Posted August 21, 2008 Author Share Posted August 21, 2008 Re: Tennis rating/info/assessment system Matthew, I'm working on the central DB, should have it on-line by the end of the weekend. This is something i would be outside the scope of this thread anyhow. Basically it will be a collection of data that can be used for research and from that i'm hoping for some joint research projects. The database will be controlled by a number of reference tables. This will allow individuals to add tables of their own that are linked to the central DB. So the whole thing will allow for joint research as well as individual projects. Things will get more clear once it's up and running and i'll be providing a description how you can plug-in data of your own. This is what the central DB will contain and i will building the data collection routines for this over the coming 2-3 months. So that means all the data will be collected and updated automatically as much as possible. - Player and tournament profiles with the basic data, name,age,right/left hand,etc... name,location,surface,prizemoney,timezone,etc... - Tennis matches, fixtures, set results, match statistics - Matches odds archive, tennis-data,betfair historic, as well as live odds collecting The first version that i'm hoping to complete this weekend will have: - Player names and player naming reference table, - Tournament names and naming reference table - Match set results, updated every few hours - Match fixtures, with odds from at least 1 source updated every few hours Once this is done you can have a look at how it works and see if you can plug-in any data of your own. Then we can look how to proceed. Quote Link to comment Share on other sites More sharing options...
Matthew Posted August 21, 2008 Share Posted August 21, 2008 Re: Tennis rating/info/assessment system Sounds brilliant!!!!! :notworthy:ok:notworthy:nana Good work :cheers Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.