Jump to content
Download the PL Forum App!

a1ehouse

Regular Members
 View Profile  See their activity

  • Posts

    504

  • Joined

  • Last visited

 Content Type 

Posts posted by a1ehouse

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops Another "all selection" today. Lewes not exactly rocking in form...

    TopAndThePlops
    Date Division HomeTeam AwayTeam WinExpectancy GoalSuperiority B365H B365D B365A HomeOverall HomeForm AwayOverall AwayForm
    17/03/2009 Conference Kettering Town Lewes 92.56% 15 1.22 5.25 11 LDWWDW 11 LDWWWW 13 LLLLLL 0 LLLLLL 0

    Elo Access Function

    in Betting Systems & Strategy

    Posted

    Re: Elo Access Function Wow, you will need lots of :beer after all that!! So you're OK with all the ELO stuff how its worked out and all that?! Thing is with most stats and ratings, if it shows things consistently you could be on to a good thing, just won't be able to back test by the sounds of it. You are looking at a lot of variables there which is both good and bad but if there's method in your madness go with the flow and see how it runs!! A 40% hit rate would be amazing for draws given the good odds you can get. Good luck!

    My Draw System including loss recovery staking plan

    in Betting Systems & Strategy

    Posted

    Re: My Draw System including loss recovery staking plan

    Of course there is a chance I will go bust until the bank is doubled but I am confident in my picks, there isn't a system out there that doesn't have risk, and I have seen thousands and they all have one thing in common, the more risk the greater the reward. Every system out there depends on good picks and I do believe I can find enough draws to make this system work, as I keep saying you need to get into the heads of the teams that are playing and access whether a draw is an acceptable result for them, the stats only tell you so much.
    Good post mate, and well done so far. As long as you are comfortable with losing a bank I say go for it! :notworthy good luck!

    Elo Access Function

    in Betting Systems & Strategy

    Posted

    Re: Elo Access Function

    glad i found this ive been trying for month to get something like this to work myself but have no acces or excel experince but i came up with this elo rating/(hw+al*100/totalgames) seems to work but im trying to adjust the draw as it can be quit high i would be interested are they your own record's on ratings cause ive found no sites which keep such type of stats an have no way of back testing what i have so for as it seemed to do well at predicting draws
    Hello servecold Not too sure what you mean by "elo rating/(hw+al*100/totalgames)", but I will put some explanation behind my calculations. Asc's is quite similar except the percentage of the kitty differs and goal weighting is not included. For me, ELO ratings are actually quite straight forward and only get complicated when you introduce variable weightings such as goals scored and home advantage. The difficulty lies with updating the records which is why I went down the Access rather than excel route. There is some information scattered around punterslounge and the internet, but to summarise how my ratings are calculated: '***************************************1997 Bob Runyan adapted Elo*************************************** ' The ratings are based on the following formula: ' ' Rn = Ro + KG × (W - We) ' ' or ' ' P = KG(W - We) ' ' Where; ' Rn = the new rating ' Ro = the old (pre-match) rating ' K = the Weight Constant for the tournament played ' * 60 for World Cup finals; ' * 50 for continental championship finals and major intercontinental tournaments; ' * 40 for World Cup and continental qualifiers and major tournaments; ' * 30 for all other tournaments; ' * 20 for friendly matches. ' G = the Goal difference adjustment for the Weight Constant ' It is increased by half if a game is won by two goals, ' by 3/4 if a game is won by three goals, ' and by 3/4 + (N-3)/8 if the game is won by four or more goals, ' where N is the goal difference. ' W = the result of the game (1 for H, 0.5 for D, and 0 for A) ' We = the Win Expectancy - the expected result ' We = 1 / (10(-dr/400) + 1) ' dr equals the difference in ratings plus 100 points for a team playing at home. ' P = Points Change '*************************************************************************************************** Lets look at an example of adjusting ratings: Two results, score in brackets: (4) Team1 vs Team2 (2) (1) Team3 vs Team1 (5) The 3 teams start with 1000 points each and this could be at the start of each season or from the earliest record you have. Now, in order for the ratings to work, we need to update both teams rating after every game. To get our numbers for the equations, refering to the info above, lets get the easiest one first!: Difference in ratings including home team weighted with 100 points (dr): dr = ((RatingHomeTeam + 100) - RatingAwayTeam) dr = ((1000 + 100) - (1000) dr = 100 Win Expectancy (We): We = 1/(10^(-dr/400) + 1) We = 1/(10^(-100/400) + 1) We = 1/(10^-0.25 + 1) We = 1/(0.56 + 1) We = 1/1.56 We = 0.64 so Team1 had a 64% chance of winning, Team2 36% WeightConstant for league games (K) K = 30 Goal Difference is 2 (G) G = 1.5 Result of the game was H (W): W = 1 Win Expectancy (We): We = 0.64 Points change (P): P = KG(W - We) P = (30 * 1.5) * (1 - 0.64) P = 45 * 0.36 P = 16.20 Team1 gets the points: 1000 + (16.20) = 1016.20 Team2 loses the points: 1000 - (16.20) = 983.80 For the next game: Team3 v Team1 Difference in ratings including home team weighted with 100 points (dr): dr = ((RatingHomeTeam + 100) - RatingAwayTeam) dr = ((1000 + 100) - (1016.20) dr = 83.80 Win Expectancy (We): We = 1/(10^(-dr/400) + 1) We = 1/(10^(-83.80/400) + 1) We = 1/(10^-0.2095 + 1) We = 1/(0.62 + 1) We = 1/1.62 We = 0.62 so Team3 had a 62% chance of winning, Team1 38% WeightConstant for league games (K) K = 30 Goal Difference is 4 (G) G = 1.875 Result of the game was A (W): W = 0 Win Expectancy (We): We = 0.62 Points change is: P = KG(W - We) P = (30 * 1.875) * (0-0.62) P = 56.25 * -0.62 P = -34.78 Team3 gets the points: 1000 + (-34.78) = 965.22 Team1 loses the points: 1016.20 - (-34.78) = 1050.98 So we can clearly see how Team1's rating improves with each win, while Teams 2 and 3 decrease. The important thing to take from all of this is that you should have both the old and new ratings for both teams for your match records. In my database I have (among other things compiled and updated regularly from the football data site): HomeTeam AwayTeam FTHG FTAG FTR RoH RoA RnH RnA where Ro = Rating Old, Rn = Rating New This allows me to do all the back testing and analysis I need when devising and developing selection systems. Any questions, post away! :dude

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops Thanks davidc, I agree that there will need to be some adjustments following the paper trial. Basic back testing led me to the 5%/95% cut off, but as there are a lot more winning selections that are coming through I'll need to do some further analysis. Having read a number of posts (and as you are alluding to) it could all be about finding "fair" booky price. Always happy to share opinion as the system develops! I have recently been looking at regression testing all the ratings (over 10000 games) with some very interesting and positive results. Top graphs I plotted rating against percentage of times the result happened Bottom graphs I plotted rating against total number of occurrences Blue = Home Green = Away Red = Draw regression.jpg

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops Downloaded the fixtures and looks like I missed one on 10th. Again missed the 95% cut off and posting for completeness. Need to make a mental note to check the downloads on Tuesdays and Fridays! Good job I keep all pre and post match ratings!!!

    TopAndThePlops
    Date Division HomeTeam AwayTeam WinExpectancy B365H Result
    10/03/2009 Division 2 Brechin Stranraer 92.64% 1.25 :dude

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops

    How would you have been getting on if you had not been paying attention to the 95% to qualify?
    Hey aldric, looking at all selections I've posted so far we'd get:
    Starting Bank100
    Current Bank£94.50
    Total Staked£160.00
    Number of Bets16
    Winning Bets12
    Losing Bets4
    Strike Rate75.00%
    Yield-3.44%
    P/L-£5.50
    A decent strike rate, just that the odds will mean I have to be more selective on what makes the cut. Back testing reduces the number of selections, but increases the likelihood of profit. Suggestions welcome!!

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops Weekend 3:

    TopAndThePlops
    Date Division HomeTeam AwayTeam WinExpectancy B365H B365D B365A
    07/03/2009 League 1 Leicester Cheltenham 94.91% 1.28 5 12
    07/03/2009 Division 2 Ayr Stranraer 93.38% 1.22 5.75 13
    07/03/2009 Bundesliga 1 Wolfsburg Karlsruhe 91.41% 1.4 4.5 7.5
    Nothing again above 95% though.

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops As the full fixture list wasn't available from footy data site there would have been 2 additions that were above 90%, but as neither made the cut I am posting for completeness:

    TopAndThePlops
    Date Division HomeTeam AwayTeam WinExpectancy B365H Result
    03/03/2009 Premier League Liverpool Sunderland 90.98% 1.4 :dude
    03/03/2009 League 2 Wycombe Chester 90.91% 1.44 :dude
    04/03/2009 Premier League Rangers Inverness C 94.50% 1.16 :wall

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Re: The a1ehouse top and the plops

    TopAndThePlops
    Date Division HomeTeam AwayTeam WinExpectancy B365H B365D B365A
    03/03/2009 Premier League Liverpool Sunderland 90.98% 1.4 4.3 9.5
    Mid week English Prem, again nothing above 95% so no qualifiers, maybe being to cautious with the percentage, but will see how it runs for now.

    A1ehouse LtD (Lay the Draw) 2009/10

    in Betting Systems & Strategy

    Posted

    Some of you may have seen my post on ELO ratings. Well after some promising results with my prediction package, I am now going to paper trial the system for a month. Essentially I have all the downloadable data from the football-data site with a rating function applied to all games based on http://en.wikipedia.org/wiki/World_Football_Elo_Ratings and http://www.eloratings.net/system.html and included goals scored and home team advantage adjustments. Basically I will post all games where the ratings are over 90% and under 10%. Generally the backtested results throw consistent winners above 95% and below 5%, so will focus on them for the bets. Starting with 100 point bank I will stake 10 points per game and see how we go!! :hope Remember this is a PAPER TRIAL ONLY!!! Edit: 2009/10 Season starts at thread 61: http://www.punterslounge.com/forum/f21/a1ehouse-ltd-lay-draw-2009-10-a-80638/#post1441585 :hope 3000 point bank, Maria ratchet staking and live!! :nana

    Elo Access Function

    in Betting Systems & Strategy

    Posted

    Re: Elo Access Function Hi Asc, its based on all divisions from the footydata site. Win Expectancy is on the match itself rather than teams. I download the fixtures list for the week then link to my ratings table which contains the latest rating for each team (updated weekly). I link the table twice (alias Home and Away) and pass the rating through a function We(HT.Rating,AT.Rating) which includes home team earning an extra 100 on the rating difference (dr). The SQL pulls all records where WinExpectancy >=0.9 Or <=0.1 WinExpectancy: We(HT.Rating,AT.Rating) 

    Function We(sRating1 As Double, sRating2 As Double) As Double
    Dim dr As Double

    dr = ((sRating1 + 100) - sRating2)
    [B]We = 1 / (10 ^ (-dr / 400) + 1)[/B]
End Function

    I was ill with flu before Xmas so only managed some very basic back testing on about 1200 of the records resulting in a small profit. I'd need to do some more robust backtesting just haven't had the time over Xmas plus need to clean up the functions and database too. My ratings are based on http://en.wikipedia.org/wiki/World_Football_Elo_Ratings and http://www.eloratings.net/system.html which include adjusting the weight constant based on goals scored in the calculation.

    Elo Access Function

    in Betting Systems & Strategy

    Posted

    Re: Elo Access Function Cheers Smokin Joe, trouble is the odds columns were all over the place in every file so I left them out when compiling all leagues and seasons into a single Access table. Looks like I will have to do a bit more work, although I reckon most odds will be quite short anyway. Even though the strike rate is pretty good, the low odds may not see any decent profit.

×
×
  • Create New...