Draws for PROFIT...Bugger it!

[Drawsandmore]

Right then, this will probably be my final attempt at posting a profitable draws system on Punters Lounge…If this one doesn’t work, I’ve really got to get a life away from Excel spreadsheets! Background information: This latest incarnation is basically a modified version of the system I used in my “Draws for Profit” Parts 1 and 11 threads, (which ultimately was not actually profitable, but it did come pretty close). This version was built using results from the 2005/06 and 2006/07 seasons. The results were then used to create home and away ratings for every team in the leagues I’ve chosen to cover. (More about which leagues and why I chose them in another post if anyone is interested). Then I used those ratings to rate every league match played in the 2007/08, 2008/09 and 2009/10 seasons...This was the “fun” part! Adjusting pretty much every variable I’d put into the spreadsheet, while always trying to be logical throughout the tinkering. Finally I finished messing about with the variables, and had something that looked fairly promising, (doesn’t it always?!) I was then left with three variations on a theme, with all leagues/divisions rated in exactly the same way. I then back-tested each of them, (with no further changes) on the 2010/2011 season. Somewhat surprisingly, all three variations performed pretty much as they had in the previous seasons. So, for 2011/2012, I went “live” with all of them…Sadly, no money exchanged hands, because all three of them were profitable…It was bloody difficult to keep betting real money on the old system when the performance of all the new ones was so much better! So, what’s different? Not a lot really…Instead of a goal being worth a pre-determined value, (as it was in the old system), it is now a percentage based on each teams current rating. I can’t believe I didn’t do this earlier. Logically, if I’d rated a team at say 20 points, with a goal being worth 4 points, how can a goal also be worth 4 points to a team rated at only 6 points? It seems to me that this goes some way towards explaining the performance of the original system in the lower divisions. Other than that, and a couple of other minor tweaks, it is pretty much the same system as before…Still trying to find a way of creating ratings that weight the very different values of class and current form, and the effect they have on predicting the correct result. The profit and loss figures below are all based on Bet365 draw odds from football-data.co.uk, staking 4 points on every bet. I have assumed that I can improve the odds by a fairly conservative 5% to arrive at the quoted profit figures. During the live testing stage, (and in my “Draws for Profit” thread), I consistently managed to improve the odds by around 7%, so I think the figures below are realistic. I apologise for including the figures/tables below…I’m not after-eventing, and obviously the figures will all be reset to zero once I post “live”. I’m just trying to put some “meat on the bones”, and also to give anyone with an understanding of probability, an opportunity to point out why I might be disappointed at the end of the season…I know past performance is no guarantee of future success, and I am prepared to lose my bank, but if somebody can see a reason that I’m likely to fail, I’d rather know before I lose everything! The tables: I think most of the headings are fairly self-explanatory, with the possible exception of the last column…This is the percentage improvement the system achieved when compared to the actual results. So if in 100 matches, 25 were draws and the system found 30 draws, the % improvement is 20%. Finally, The results! Option 1 [TABLE=width: 823]

[TR] [TD][/TD] [TD]All matches[/TD] [TD]Actual draws[/TD] [TD]Win %[/TD] [TD]Losing run[/TD] [TD]Start Bank[/TD] [TD]Max. Bank[/TD] [TD]Min. Bank[/TD] [TD]Final Bank[/TD] [TD]Profit or loss[/TD] [TD]Bets[/TD] [TD]Wins[/TD] [TD]Win %[/TD] [TD]Yield[/TD] [TD]% Imp.[/TD] [/TR] [TR] [TD]07/08[/TD] [TD]4116[/TD] [TD]1090[/TD] [TD]26.48%[/TD] [TD]15[/TD] [TD]100.00[/TD] [TD]322.65[/TD] [TD]72.00[/TD] [TD]245.69[/TD] [TD]145.69[/TD] [TD]338[/TD] [TD]110[/TD] [TD=align: right]32.54%[/TD] [TD=align: right]10.78%[/TD] [TD=align: right]22.89%[/TD] [/TR] [TR] [TD]08/09[/TD] [TD]4116[/TD] [TD]1112[/TD] [TD]27.02%[/TD] [TD]12[/TD] [TD]100.00[/TD] [TD]171.98[/TD] [TD]-1.77[/TD] [TD]131.73[/TD] [TD]31.73[/TD] [TD]387[/TD] [TD]116[/TD] [TD=align: right]29.97%[/TD] [TD=align: right]2.05%[/TD] [TD=align: right]10.95%[/TD] [/TR] [TR] [TD]09/10[/TD] [TD]4098[/TD] [TD]1103[/TD] [TD]26.92%[/TD] [TD]10[/TD] [TD]100.00[/TD] [TD]265.63[/TD] [TD]109.66[/TD] [TD]247.29[/TD] [TD]147.29[/TD] [TD]359[/TD] [TD]115[/TD] [TD=align: right]32.03%[/TD] [TD=align: right]10.26%[/TD] [TD=align: right]19.01%[/TD] [/TR] [TR] [TD]10/11[/TD] [TD]4116[/TD] [TD]1157[/TD] [TD]28.11%[/TD] [TD]13[/TD] [TD]100.00[/TD] [TD]223.00[/TD] [TD]86.10[/TD] [TD]215.00[/TD] [TD]115.00[/TD] [TD]379[/TD] [TD]119[/TD] [TD=align: right]31.40%[/TD] [TD=align: right]7.59%[/TD] [TD=align: right]11.70%[/TD] [/TR] [TR] [TD]11/12[/TD] [TD]4116[/TD] [TD]1111[/TD] [TD]26.99%[/TD] [TD]13[/TD] [TD]100.00[/TD] [TD]294.29[/TD] [TD]86.88[/TD] [TD]290.29[/TD] [TD]190.29[/TD] [TD]401[/TD] [TD]131[/TD] [TD=align: right]32.67%[/TD] [TD=align: right]11.86%[/TD] [TD=align: right]21.03%[/TD] [/TR] [TR] [TD]Total[/TD] [TD]20562[/TD] [TD]5573[/TD] [TD]27.10%[/TD] [TD][/TD] [TD]500.00[/TD] [TD][/TD] [TD][/TD] [TD]1130.00[/TD] [TD]630.00[/TD] [TD]1864[/TD] [TD]591[/TD] [TD=align: right]31.71%[/TD] [TD=align: right]8.45%[/TD] [TD=align: right]16.98%[/TD] [/TR] [/TABLE] This really is “warts “n” all and shows every bet in every single division, in all of the leagues that I’ve covered. There isn’t really any data-mining going on here. It certainly seems to show promise, but I’ll either need to reduce the stake, or increase the start bank to allow for 2008/09 happening again. Option 2 [TABLE=width: 823] [TR] [TD][/TD] [TD]All matches[/TD] [TD]Actual draws[/TD] [TD]Win %[/TD] [TD]Losing run[/TD] [TD]Start Bank[/TD] [TD]Max. Bank[/TD] [TD]Min. Bank[/TD] [TD]Final Bank[/TD] [TD]Profit or loss[/TD] [TD]Bets[/TD] [TD]Wins[/TD] [TD]Win %[/TD] [TD]Yield[/TD] [TD]% Imp.[/TD] [/TR] [TR] [TD]07/08[/TD] [TD]3619[/TD] [TD]979[/TD] [TD]27.05%[/TD] [TD]14[/TD] [TD]100.00[/TD] [TD]361.84[/TD] [TD]76.00[/TD] [TD]315.83[/TD] [TD]215.83[/TD] [TD]329[/TD] [TD]114[/TD] [TD=align: right]34.65%[/TD] [TD=align: right]16.40%[/TD] [TD=align: right]28.09%[/TD] [/TR] [TR] [TD]08/09[/TD] [TD]3537[/TD] [TD]978[/TD] [TD]27.65%[/TD] [TD]13[/TD] [TD]100.00[/TD] [TD]165.47[/TD] [TD]-27.66[/TD] [TD]159.97[/TD] [TD]59.97[/TD] [TD]376[/TD] [TD]115[/TD] [TD=align: right]30.59%[/TD] [TD=align: right]3.99%[/TD] [TD=align: right]10.61%[/TD] [/TR] [TR] [TD]09/10[/TD] [TD]3564[/TD] [TD]975[/TD] [TD]27.36%[/TD] [TD]11[/TD] [TD]100.00[/TD] [TD]255.08[/TD] [TD]105.91[/TD] [TD]253.98[/TD] [TD]153.98[/TD] [TD]336[/TD] [TD]110[/TD] [TD=align: right]32.74%[/TD] [TD=align: right]11.46%[/TD] [TD=align: right]19.67%[/TD] [/TR] [TR] [TD]10/11[/TD] [TD]3550[/TD] [TD]1026[/TD] [TD]28.90%[/TD] [TD]11[/TD] [TD]100.00[/TD] [TD]224.10[/TD] [TD]88.40[/TD] [TD]204.92[/TD] [TD]104.92[/TD] [TD]349[/TD] [TD]110[/TD] [TD=align: right]31.52%[/TD] [TD=align: right]7.52%[/TD] [TD=align: right]9.06%[/TD] [/TR] [TR] [TD]11/12[/TD] [TD]3605[/TD] [TD]995[/TD] [TD]27.60%[/TD] [TD]13[/TD] [TD]100.00[/TD] [TD]358.41[/TD] [TD]102.06[/TD] [TD]346.41[/TD] [TD]246.41[/TD] [TD]370[/TD] [TD]126[/TD] [TD=align: right]34.05%[/TD] [TD=align: right]16.65%[/TD] [TD=align: right]23.38%[/TD] [/TR] [TR] [TD]Total[/TD] [TD]17875[/TD] [TD]4953[/TD] [TD]27.71%[/TD] [TD][/TD] [TD]500.00[/TD] [TD][/TD] [TD][/TD] [TD]1281.11[/TD] [TD]781.11[/TD] [TD]1760[/TD] [TD]575[/TD] [TD=align: right]32.67%[/TD] [TD=align: right]11.10%[/TD] [TD=align: right]17.91%[/TD] [/TR] [/TABLE] A little bit of “mining” going on here! In every variation I’ve tried, the month of May is fairly hopeless…Too many teams with either everything or nothing to play for, and although May pretty much breaks even over the five years of testing, I think it probably isn’t worth the hassle, (especially as May tipped my previous thread on here into the red!). Additionally, I’ve dropped a couple of the lower divisions where the actual draw percentage over the last five years is consistently below 25%…The odds may be better in these matches, and there have been profitable seasons, but both of the excluded divisions are a bit too unpredictable, and frankly not worth the effort for the minimal profit gained. The profit is up compared to option 1, and the yield also shows a considerable improvement. The downside is that the 2008/09 season was an even bigger “Bank Buster”. Option 3 [TABLE=width: 823] [TR] [TD][/TD] [TD]All matches[/TD] [TD]Actual draws[/TD] [TD]Win %[/TD] [TD]Losing run[/TD] [TD]Start Bank[/TD] [TD]Max. Bank[/TD] [TD]Min. Bank[/TD] [TD]Final Bank[/TD] [TD]Profit or loss[/TD] [TD]Bets[/TD] [TD]Wins[/TD] [TD]Win %[/TD] [TD]Yield[/TD] [TD]% Imp.[/TD] [/TR] [TR] [TD]07/08[/TD] [TD]1987[/TD] [TD]577[/TD] [TD]29.04%[/TD] [TD]8[/TD] [TD]100.00[/TD] [TD]334.17[/TD] [TD]77.21[/TD] [TD]324.16[/TD] [TD]224.16[/TD] [TD]191[/TD] [TD]74[/TD] [TD=align: right]38.74%[/TD] [TD=align: right]29.34%[/TD] [TD=align: right]33.42%[/TD] [/TR] [TR] [TD]08/09[/TD] [TD]1905[/TD] [TD]539[/TD] [TD]28.29%[/TD] [TD]11[/TD] [TD]100.00[/TD] [TD]242.37[/TD] [TD]45.55[/TD] [TD]227.35[/TD] [TD]127.35[/TD] [TD]210[/TD] [TD]72[/TD] [TD=align: right]34.29%[/TD] [TD=align: right]15.16%[/TD] [TD=align: right]21.18%[/TD] [/TR] [TR] [TD]09/10[/TD] [TD]1974[/TD] [TD]555[/TD] [TD]28.12%[/TD] [TD]10[/TD] [TD]100.00[/TD] [TD]265.20[/TD] [TD]109.66[/TD] [TD]262.02[/TD] [TD]162.02[/TD] [TD]181[/TD] [TD]66[/TD] [TD=align: right]36.46%[/TD] [TD=align: right]22.38%[/TD] [TD=align: right]29.69%[/TD] [/TR] [TR] [TD]10/11[/TD] [TD]1919[/TD] [TD]569[/TD] [TD]29.65%[/TD] [TD]11[/TD] [TD]100.00[/TD] [TD]237.33[/TD] [TD]96.40[/TD] [TD]230.15[/TD] [TD]130.15[/TD] [TD]200[/TD] [TD]69[/TD] [TD=align: right]34.50%[/TD] [TD=align: right]16.27%[/TD] [TD=align: right]16.35%[/TD] [/TR] [TR] [TD]11/12[/TD] [TD]1973[/TD] [TD]553[/TD] [TD]28.03%[/TD] [TD]9[/TD] [TD]100.00[/TD] [TD]290.70[/TD] [TD]105.66[/TD] [TD]275.10[/TD] [TD]175.10[/TD] [TD]233[/TD] [TD]82[/TD] [TD=align: right]35.19%[/TD] [TD=align: right]18.79%[/TD] [TD=align: right]25.56%[/TD] [/TR] [TR] [TD]Total[/TD] [TD]9758[/TD] [TD]2793[/TD] [TD]28.62%[/TD] [TD][/TD] [TD]500.00[/TD] [TD][/TD] [TD][/TD] [TD]1318.78[/TD] [TD]818.78[/TD] [TD]1015[/TD] [TD]363[/TD] [TD=align: right]35.76%[/TD] [TD=align: right]20.17%[/TD] [TD=align: right]24.95%[/TD] [/TR] [/TABLE] A bit more “mining”…I prefer to call it fine tuning! A further three of the lower divisions excluded, (two of which were actually profitable over the full five seasons, but they achieved the profit very erratically, winning a Hell of a lot in some seasons, and losing a Hell of a lot in others). The profit is not much better than in option two, but it has been achieved far more evenly, and the yield is also much better. My logic could be flawed and I might just be seeing what I want to see…Searching for patterns where none really exist, but there is a certain consistency about the results…Especially in option three. Obviously there are far fewer bets in this one, but I still feel comfortable that an average of 200 bets a year over a five year period should be a large enough sample size to start putting real money down in 2012/13 with a degree of confidence. What’s next? Well I really would like some feedback regarding the probability of any or all of these figures being repeated next season, or the season after etc. I’m intending to post (and back) all of the selections from Option three during the 2012/2013 season, but if someone with a better understanding of statistics and probability than me, (and that’ll be quite a lot of you), can show me why option one or two is better, I’m quite prepared to listen and learn.

[leony]

Re: Draws for profit...The last chance saloon! Looks interesting. I have 1 question: What is the average, winning, odds for the 3 systems?