Orpheus Posted October 31, 2009 Share Posted October 31, 2009 I saw Vilmoura's thread - 'See Saw Football Betting Strategy' - and there were some similarities with my own, so thought I would post it for comments and public tracking ... Bouncebackability: the ability to be successful again after a period of failure [Macmillan English Dictionary], first coined by Iain Dowie. In other words, this is a system that expects teams to perform better after playing unexpectedly poorly, but also conversely to perform worse after playing unexpectedly well. An earlier version of this system was applied to baseball and outlined here: http://www.punterslounge.com/forum/f21/gambling-shock-baseball-system-85799. That system was too restrictive, but subsequent revisions yielded a very profitable system (proofed on sports-punter.com) until the last two weeks of the season ... in future, I will look to exclude all teams from the system once they cannot reach the post-season playoffs (i.e. the effect on incentives ... do other punters close down their systems as the season draws to a close?). This is a system that should have some applicability to football, but its primary focus is North American sports and especially baseball, basketball and ice hockey as these sports involves teams playing many times per week. Let me outline the four pillars of the system: HomeWin ... if a team wins "unexpectedly" at higher-than-benchmark odds in one game, the next time that it plays and the odds are higher-than-benchmark odds, it will tend not the win the second time. For example, visiting teams will be alerted to how the home team pulled off a surprise result previously and so will amend their tactics to counter this. AwayWin ... exactly the same as the 'home win' element, but just for away games. HomeLoss ...if a team "unexpectedly" fails to win at lower-than-benchmark odds in one game, the next time that it plays and the odds are lower-than-benchmark odds, it will tend to win the second time. For example, losses at home are felt much harder by the club/management/players than away as they lead to much wider criticism in the local press, so will lead to a much stronger effort the next time the situation reoccurs. AwayLoss ... exactly the same as the 'home loss' element, but just for away games. Here's the crucial difference between this system and many others ... the modelling is done on a team-by-team rather than all-team basis. For example, if Man Utd were to fail to win at home, there would always be a very strong expectation that they would react and that the chances of them wining their next home match would increase (in this case, Man Utd's 'HomeLoss' benchmark odds would be very high). Alternatively, if Portsmouth didn't win at home, e.g. they managed a draw, there wouldn't be the same reaction within the club or among the fans to make sure that it wasn't repeated in their next match (in this case, Portsmouth's 'HomeLoss' benchmark odds would be rather low ... only if they failed to win a game against very poor opposition would there be such a reaction). This example shows why the system needs to be team-specific rather than impose some general rules by which all teams are appraised. In terms of basketball and the NFL, I use benchmark spreads rather than odds as there is evidence that coaches use the 'covering the spread' as a measure of how well their team performed relative to expectations. I'm continuing to track this, but I'm not as convinced as with the other sports. I also add filters to limit the plays: (i) when backing a team, they must have shown a profit at the available odds over the sample period. For example, Arsenal are 1.62 against Spurs ... they would be eliminated from the above system as backing them at home at 1.65 or lower over the sample period would have resulted in a loss of 4.86pts (to 1pt stakes). (ii) generally, I only consider first-order lags (e.g. if Man Utd lose twice in a row at home of odds of 2.5 or lower, I will not consider them in the third game that they are 2.5 or lower) unless particular leagues have tendency for higher-order lags to be successful, though this is rare. (iii) I continue to track the system for all leagues, but only post plays if the sum of all plays in that league are in profit for the season (with some follow-on effect from the previous season). This, for example, knocks out leagues which just do not work at all with this system, such as the Belgian Jupiler League. The sample period for each team is either their last 100 home or away games (whichever is relevant) or their home or away games over the last five years (if this produces less than 100 home or away games). As such, the sample period is updated with each game and so the benchmark odds for each team may vary from week to week. Obviously, it is not possible to provide the finer details on a team-by-team basis for such a micro-system, but I hope that the general rules are apparent. I'll post the plays each day that qualify via this system and see if it generates some discussion. In contrast to the plays posted on sports-punter.com, all plays will be graded to the following rules: favourites will be backed to win 100pts; underdogs will be backed with a stake of 100pts. Just one qualifying play tonight. NHL: San Jose (inc OT) 1.60 Quote Link to comment Share on other sites More sharing options...
Hooloovoo Posted October 31, 2009 Share Posted October 31, 2009 Re: The 'boucebackability' system 3-1 and a nice start :clap With so many filters, how many bets do you reckon there'll be a week? Quote Link to comment Share on other sites More sharing options...
Orpheus Posted October 31, 2009 Author Share Posted October 31, 2009 Re: The 'boucebackability' system Thanks :) Normally 1-2 NHL plays per day and it would be the same with the NBA once that season gets fully underway and the system starts to shows profits this season. In terms of football, I'd expect 12-15 plays to qualify today. Quote Link to comment Share on other sites More sharing options...
Vilamoura04 Posted October 31, 2009 Share Posted October 31, 2009 Re: The 'boucebackability' system Sounds good Orpheus good luck. The book I took the see saw strategy from also describes something very similar to what you are describing - its pretty brief, theres not a huge amount on it - so i could email you the page if you are interested. In summary it backs up pretty much exactly what you are saying in English league games - teams have a 'pain threshold' of number defeats at which point they bounce back. For a Man U this would be one game, for Portsmouth this would be three games. One thing I have considered doing with the see saw strategy is to calculate the value in the selections and only bet when odds>true odds. This is the fundamental basis for the systems 'edge' and so is important. The big problem with this is that there wouldnt be enough selections to keep me interested :$ Quote Link to comment Share on other sites More sharing options...
Orpheus Posted October 31, 2009 Author Share Posted October 31, 2009 Re: The 'boucebackability' system Thanks Vilamoura04 :) Could you send a scan of that page via pm? It would be interesting to see it. The idea behind my system was William Mallios' book The Analysis of Sports Forecasting: Modelling Parallels between Sports Gambling and Financial Markets which introduced the concept of statistical vs. gambling shocks (explained in a little detail in my original thread - link in first post). This is a book for statisticians, but while the theory sounded interesting the empirical testing was incredibly weak. I've seen similar systems, but not exactly the same, so thought I'd test it myself ;) Today's 3pm football plays: lay Hartlepool 3.25 lay Leyton Orient 4.60 Bury 2.75 lay Hereford 1.94 lay Shrewsbury 5.80 Rotherham 1.76 Heerenveen 1.67 lay Naval 2.46 Celtic 1.27 lay Motherwell 2.62 Airdrie 2.53 lay Dundee 2.44 lay Ross County 3.75 lay Dumbarton 2.72 Quote Link to comment Share on other sites More sharing options...
Orpheus Posted October 31, 2009 Author Share Posted October 31, 2009 Re: The 'boucebackability' system 1-0; +100pts Later football plays: Athletic Bilbao 2.82 lay Osasuna 7.80 Sevilla 1.62 Quote Link to comment Share on other sites More sharing options...
bogdancavnic Posted October 31, 2009 Share Posted October 31, 2009 Re: The 'boucebackability' system not to bad tonight:ok Quote Link to comment Share on other sites More sharing options...
Orpheus Posted October 31, 2009 Author Share Posted October 31, 2009 Re: The 'boucebackability' system Thanks, though I did dodge a couple of bullets with late equalisers in the League 1 lays :eek Afternoon NHL play: New Jersey 2.02 Quote Link to comment Share on other sites More sharing options...
Orpheus Posted October 31, 2009 Author Share Posted October 31, 2009 Re: The 'boucebackability' system Evening NHL plays: Buffalo 1.89 St Louis 1.56 Nashville 1.88 FWIW All NHL plays include both overtime and shootouts. Quote Link to comment Share on other sites More sharing options...
bogdancavnic Posted November 1, 2009 Share Posted November 1, 2009 Re: The 'boucebackability' system Your strategy is working orpheus. Yesterday, in soccer, only 5 losses from 17 games. Good work!! I folow you with atention. Thank you for your posts. Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 1, 2009 Author Share Posted November 1, 2009 Re: The 'boucebackability' system Thanks bogdancavnic. It's always a relief that once you post a system that it starts with a profit on day one ;) Football: 13-4; +695 NHL: 3-2; +11 Today's football plays: lay Sochaux 2.78 Cagliari 2.26 Fiorentina 1.48 lay Liverno 6.40 lay Palermo 2.12 Sparta Rotterdam 2.46 Pacos Ferrerira 2.32 Sporting Lisbon 1.49 lay Dundee Utd 4.30 lay Espanol 2.10 Quote Link to comment Share on other sites More sharing options...
glory6 Posted November 1, 2009 Share Posted November 1, 2009 Re: The 'boucebackability' system In contrast to the plays posted on sports-punter.com, all plays will be graded to the following rules: favourites will be backed to win 100pts; underdogs will be backed with a stake of 100pts. Hi interesting system an especially in the football field the start was amazing great. In your stake rules you are talking about backing favs and underdogs. But how is this rule working in case of a lay? I guess everything Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 1, 2009 Author Share Posted November 1, 2009 Re: The 'boucebackability' system Hi glory6, Yes, you are correct ... "lay Sochaux 2.78" means having a liability of 178pts in order to win 100pts; yesterday's "lay Hereford 1.94" meant having a liability of 100pts that could win 106pts. Quote Link to comment Share on other sites More sharing options...
lawalahmed Posted November 1, 2009 Share Posted November 1, 2009 Re: The 'boucebackability' system Hello Friend, i thnk Dundee shud be a scotish team but no bet under bet365 4 scotish league 2day. Goodluck Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 1, 2009 Author Share Posted November 1, 2009 Re: The 'boucebackability' system Yes, that was disappointing as Rangers were leading at the time of the abandonment :( Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 1, 2009 Author Share Posted November 1, 2009 Re: The 'boucebackability' system NHL play: Columbus 2.60 Quote Link to comment Share on other sites More sharing options...
greatred Posted November 2, 2009 Share Posted November 2, 2009 Re: The 'boucebackability' system "Really impressive is this stuff. Could you please help me understand two things: what are the benchamrk odds in soccer (home+away FAV) and how Cagliari HW qualified for this system yesterday? I would appreciate any answer." EDIT: nevermind sorry for this question as your sportpunter analysis covers my question. I do although have a question regarding the odds you take as an benchamark. I wont bother you with them yet will wait for your football plays for today and having allready analysed today games hope to understand the concept :) Quote Link to comment Share on other sites More sharing options...
Vilamoura04 Posted November 2, 2009 Share Posted November 2, 2009 Re: The 'boucebackability' system Orpheus - looking good so far - how long does it take you to run your models for so many sports/fixtures every day? Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 2, 2009 Author Share Posted November 2, 2009 Re: The 'boucebackability' system greatred, It certainly was a good weekend ... but like any model that uses odds, I'm merely trying to seek an edge rather than trying to model the true probability of game results. As such, it will unfortunately be the exception that the system hits over 75% of plays over a weekend, but it has been producing long-term profits. I'm not sure that I'll have any plays today as there are so few games in the leagues that I track (thanks to the data from football-data.co.uk). I'll look at them later this afternoon. Vilamoura04, The football data for my model is taken from football-data.co.uk, so I use their bi-weekly odds & results updates, which means that I can perform all my league computations on Friday afternoon for Friday-Monday games and again on Tuesday afternoon for Tuesday-Thursday games. It is all done via macros in Excel and each league would take 5-10 minutes for a full programme of fixtures. It's a lot more work to run a model on a team-by-team basis rather than for teams as a whole and it means that it's not something that I can describe here and others easily replicate. I'll try my best to answer any questions though ;) Quote Link to comment Share on other sites More sharing options...
greatred Posted November 2, 2009 Share Posted November 2, 2009 Re: The 'boucebackability' system Orpheus, thanks for your reply. By "seek the edge" you mean taking an avarage of home or away games and particullar odds? Please help me understand this by correcting following analysis: Hansa Rostock due a WIN at home @2.6 after floosing previous home game beeing an 1.8 favorite. Quote Link to comment Share on other sites More sharing options...
bogdancavnic Posted November 2, 2009 Share Posted November 2, 2009 Re: The 'boucebackability' system Orpheus, in order to answer all the questions that appeared or could appear, I think it would be batter to demonstrate how to analyze a game from the beginning to the end. In this way everyone would understand, including me;) Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 2, 2009 Author Share Posted November 2, 2009 Re: The 'boucebackability' system greatred, that's not what I mean by "seeking an edge". IMO there are two primary routes for systems handicapping. The first is to develop a full model of sports forecasting that predicts the 'true' probability of team x winning. This probability is then compared to the implied probability given by the bookies' odds and if there is a significant difference, you make the bet. The second, and this is my adopted approach, is to take the bookies' odds as an approximation for the 'true' probability of team x winning. This takes out the problem of trying to model the 'true' result better than the bookies and the resources that they have at their disposal for this task (and, yes, I know that there are other rationales behind odds compiling). The 'edge' can now simply be earned by finding an important aspect of sport that the odds compilers to not take into account. For example, imagine if the bookies didn't take into account home advantage. You wouldn't need to develop a full model to better predict game results; simply allow for home advantage when appraising odds and you would make profits. Obviously, that 'extra element' won't be as simple as home advantage, but if you do find one, it should provide a profitable edge over the bookies. Vilamoura04, that's a good idea ... I'll do that in a new post ... probably tonight as there are no football plays to be posted. Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 2, 2009 Author Share Posted November 2, 2009 Re: The 'boucebackability' system This is the step-by-step approach for one team ... in this example, Celtic, for Saturday's home match against Kilmarnock ... and for simplicity I'll just outline the 'HomeLoss' element. 1. Calculate the benchmark odds Using the Scottish Premier League odds/results from the last five years (31/10/04 - 30/10/09 in this example) from football-data.co.uk, calculate the profits from backing Celic at home after they had failed to win their previous home game and group these according to odds ranges. The results of this are: Odds ....... Profit =1.50, >=1.60, etc.) which produce the best profit when laying the team; the rest is the same,just that we are laying rather than backing, and in part (3) we are checking that the current match odds are equal to or greater than the benchmark odds. And the 'RoadWin' is the same as the 'HomeWin' angle, but only looking at teams on the road. It can get a bit messy to explain, especially with the filters, but hopefully I've explained the four stages clearly enough and they may raise some ideas. Obviously, this could take a very long time to complete manually, so Excel macros to automate the tasks and run them for more than one team at a time is vital! Quote Link to comment Share on other sites More sharing options...
Hooloovoo Posted November 3, 2009 Share Posted November 3, 2009 Re: The 'boucebackability' system Very interesting read :clap Quote Link to comment Share on other sites More sharing options...
madnesstiger Posted November 3, 2009 Share Posted November 3, 2009 Re: The 'boucebackability' system Very interesting read :clap I'll second that - an interesting thread. I'm always interetsed when somebody creates their own unique method and you've explained clearly how you go about analysing each match. Credit to you, Sir, and will follow this with interest. :ok MT Quote Link to comment Share on other sites More sharing options...
madnesstiger Posted November 3, 2009 Share Posted November 3, 2009 Re: The 'boucebackability' system ...just seen the price of the book mentioned earlier in the thread, by William Mallios... William Mallios book @ Amazon :loon Quote Link to comment Share on other sites More sharing options...
Hooloovoo Posted November 3, 2009 Share Posted November 3, 2009 Re: The 'boucebackability' system ...just seen the price of the book mentioned earlier in the thread, by William Mallios... William Mallios book @ Amazon :loon LOL at the price :loon Quote Link to comment Share on other sites More sharing options...
lawalahmed Posted November 3, 2009 Share Posted November 3, 2009 Re: The 'boucebackability' system Hi friend, it take me 30mins 2 digest ur explanatn, it quit interesting with a unique ideal. what do u mean by benchmark odd and how do i check that? Quote Link to comment Share on other sites More sharing options...
Orpheus Posted November 3, 2009 Author Share Posted November 3, 2009 Re: The 'boucebackability' system lawalahmed The purpose of finding the 'benchmark odds' is to find that range of odds at which backing a team to win after losing at home is (most) profitable. The calculation for this is explained in part (1) of my above posts: take the last five years of results/odds and calculate the cumulative profits for each odds range. In the example above, the profits had been largest when backing Celtic to win at home after losing their last home match and the odds had been 2.40 or lower. Hence 2.40 are now the 'benchmark odds' for Celtic and the 'HomeLoss' angle. This is the purpose of part (1) of my four-part approach outlined above. Let me know if I need to explain it differently. Quote Link to comment Share on other sites More sharing options...
Hooloovoo Posted November 3, 2009 Share Posted November 3, 2009 Re: The 'boucebackability' system 4. Check the profitability of backing the selected team at the current match odds This filter was introduced to ensure that we didn't end up backing teams who were profitable at only very small odds ranges and that was because of one or two 'freak' results which produced large profits at large odds (e.g. Hull in the first half of last season). In this example, we calculate the profitability of backing Celtic whenever they are 1.26 (really it is at the odds group of I don't get this filter - First you find a match that should be profitable because of all the criterias mentioned, and then you add this filter that says that it must have shown profit for all matches within this odds range, otherwise it's a no-bet? Doesn't this defeat the entire purpose of the process? :unsure Quote Link to comment Share on other sites More sharing options...
.thumb.JPG.8c5a15b710f7a4188ce438df64354c1e.JPG)
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.