Friday, August 28, 2009

AFL 2009 Season - Round 22 Predictions

Finally a good week for the AFL tips since I started publishing them - tipping 7 out of 8 correct games, the game I missed being 1st place St Kilda losing to 13th placed North Melbourne. Still, some of the unusual predictions the model threw up were justified, the Western Bulldogs over Geelong in particular.

So here are the final tips for the home-and-away matches for the year, quite a few 50/50 propositions, and again it has tipped the Western Bulldogs as huge favourites in a game the bookies have Collingwood as slight favourites.

Team 1Team 2VenuePr 1Pr 2
West CoastRichmondSubiaco50.2% 47.7%
GeelongFremantleSkilled Stadium64.3% 33.8%
HawthornEssendonMCG52.4% 45.5%
CarltonAdelaideTelstra Dome45.8% 52.1%
SydneyBrisbaneSCG70.1% 28.2%
Port AdelaideNorth MelbourneAAMI Stadium71.9% 26.5%
MelbourneSt KildaMCG2.3% 97.5%
CollingwoodWestern BulldogsTelstra Dome14.4% 84.5%

Total tips for the season: 118 from 168 (70.2%)

Thursday, August 27, 2009

EPL 2009/10 Season - Week 3 Predictions

A decent week last weekend for the EPL model, correctly tipping 6 out of 10 results. The season tally is sitting on 54%, just a smidge below where we'd like to be to challenge for the top spot at
http://www.footytips.com.au/

Tipping just the one away team this week, Liverpool at Bolton. The only concern this round looks to be tipping Pompey over Man City - it's still too early in for the model to pick up on any inherent gain in form Man City have.

DateHome TeamAway TeamPr HomePr DrawPr Away
29/08/2009Blackburn RoversWest Ham United43% 30% 28%
30/08/2009Bolton WanderersLiverpool13% 21% 66%
30/08/2009ChelseaBurnley88% 8% 4%
30/08/2009Manchester UnitedArsenal54% 26% 19%
30/08/2009Stoke CitySunderland51% 28% 22%
30/08/2009Tottenham HotspurBirmingham City65% 22% 13%
30/08/2009Wolverhampton WanderersHull City53% 27% 20%
30/08/2009Aston VillaFulham56% 26% 18%
31/08/2009EvertonWigan Athletic68% 20% 12%
31/08/2009PortsmouthManchester City36% 30% 33%

Total tips for the season: 14 from 26 (54%)

Tuesday, August 25, 2009

A-League 2009/10 Season - Round 4 Predictions

The A-League model threw up 2 winners for the week, so we've dipped below 50% again for the season so far. Only one small bet for the week on Central Coast against Sydney was denied when they missed a last minute penalty.

Tipping a couple of away teams this week which is always a good sign - moreso that one of those teams is the Gold Coast, signs that the model has picked up on their form after only 3 games.

DateTeam 1Team 2VenuePr T1Pr DrawPr T2
28/08/2009Adelaide UnitedNorth QueenslandHindmarsh Stadium47% 26% 27%
28/08/2009Perth GloryMelbourne VictoryMembers Equity Stadium31% 26% 42%
29/08/2009Queensland RoarCentral Coast MarinersSuncorp Stadium47% 26% 28%
30/08/2009Newcastle JetsGold CoastEnergyAustralia Stadium36% 27% 37%
30/08/2009Sydney FCNew ZealandSydney Football Stadium49% 25% 26%

Total tips for the season: 7 from 15 (46.7%)

Friday, August 21, 2009

EPL 2009/10 Season - Week 2 Predictions - Part 2

The model performed quite well in the midweek games, tipping 4 out of 6 results correctly - taking the season tally to 8 correct predictions from 16 games - meaning as with the A-League we have reached the 50% mark, from which we should only improve.

A bit of analysis has come up with what seems to be a profitable betting strategy for the EPL model. Basically for teams playing at home, if they are rated at least 25% chance to win, and have a 15% overlay, then this becomes a bet. For away teams, provided they are at least a 30% chance to win, then any advantage over the bookmaker becomes a bet. It is rare to bet on a drawn outcome given, so the best advice would be to just leave these alone. For more information on how to calculate if an advantage over the bookie exists, and to work out an optimal amount to bet, refer to my earlier posts on the Kelly betting criteria.

So without further ado here are the games for the upcoming weekend of English Premier League matches.
DateHome TeamAway TeamPr HomePr DrawPr Away
23/08/2009ArsenalPortsmouth81% 13% 6%
23/08/2009Birmingham CityStoke City44% 30% 27%
23/08/2009Hull CityBolton Wanderers35% 31% 34%
23/08/2009Manchester CityWolverhampton Wanderers58% 25% 17%
23/08/2009SunderlandBlackburn Rovers36% 31% 34%
23/08/2009Wigan AthleticManchester United10% 18% 73%
23/08/2009BurnleyEverton21% 27% 51%
24/08/2009FulhamChelsea13% 22% 65%
24/08/2009West Ham UnitedTottenham Hotspur38% 30% 32%
24/08/2009LiverpoolAston Villa73% 17% 9%

Total tips for the season: 8 from 16 (50%)

AFL 2009 Season - Round 21 Predictions

Murphy's Law still kicking around - only managing to tip 3 winners last week, combined with the 2.5 winners from round 19, when the lowest round I had prior to making these tips public was 4.

Anyway we soldier on, though looking at the first game has cause to make me nervous, having the Western Bulldogs as an 85% chance against Geelong - mainly due to the profile of players selected. The player-based aspect of the model will be getting a serious overall in the off-season, but as it stands it is sitting on 111 tips so far, so hopefully this game is an exception rather than the rule. This game aside, the other predictions look quite reasonable.

Team 1Team 2VenuePr 1Pr 2
Western BulldogsGeelongTelstra Dome84.9% 14.0%
CarltonMelbourneTelstra Dome87.2% 11.9%
AdelaideWest CoastAAMI Stadium89.1% 10.2%
BrisbanePort AdelaideGabba65.8% 32.4%
RichmondHawthornMCG20.3% 78.3%
St KildaNorth MelbourneTelstra Dome92.3% 7.1%
CollingwoodSydneyMCG72.2% 26.2%
FremantleEssendonSubiaco53.5% 44.5%

Total tips for the season: 111 from 160 (69.4%)

Thursday, August 20, 2009

A-League 2009/10 Season - Round 3 Predictions

A decent week of tipping in week 2, tipping all home teams and having 3 of them win. That brings the season average up to 50% so far - if we can stay above that figure for the rest of the season I'll be happy.

A bit of analysis has shown that there is some consistent profit to be made by betting on home teams where the probability of winning is less than 55%, yet still have an advantage over the bookie (see earlier posts regarding the Kelly betting method for further discussions on this). Having said that, I'd avoid betting on games involving Gold Coast or North Queensland for a while yet until their ratings stabilise somewhat.

For the first time this season the model has thrown up a tip on an away team. The home team dominance a testament to the fact that given then fickle nature of form in the A-League, even a slim case of home-team advantage will win out.

DateTeam 1Team 2VenuePr T1Pr DrawPr T2
21/08/2009Adelaide UnitedGold CoastHindmarsh Stadium43% 26% 31%
22/08/2009Central Coast MarinersSydney FCBluetongue Stadium40% 27% 33%
22/08/2009North QueenslandMelbourne VictoryDairy Farmers Stadium36% 27% 38%
23/08/2009Queensland RoarNew ZealandSuncorp Stadium51% 25% 24%
23/08/2009Perth GloryNewcastle JetsMembers Equity Stadium37% 27% 36%

Total tips for the season: 5 from 10 (50%)

Tuesday, August 18, 2009

EPL 2009/10 Season - Week 2 Predictions

An average start to the 2009-10 Premier League season, only managing to correctly tip 4 out of 10 winners - even more average when you consider that there were no draws involved. The two teams that have supposedly improved in the off-season, Tottenham and Manchester City, both recorded impressive wins so as mentioend previously if these teams keep doing well then the model will react to this before too long.

Most teams are involved in mid-week game so here they are - no real surprises here.


DateHome TeamAway TeamPr HomePr DrawPr Away
18/08/2009SunderlandChelsea8% 16% 76%
18/08/2009Wigan AthleticWolverhampton Wanderers49% 28% 23%
19/08/2009Birmingham CityPortsmouth41% 30% 29%
19/08/2009BurnleyManchester United8% 15% 77%
19/08/2009Hull CityTottenham Hotspur24% 29% 47%
19/08/2009LiverpoolStoke City84% 11% 5%

Total tips for the season: 4 from 10 (40%)

Friday, August 14, 2009

AFL 2009 Season - Round 20 Predictions

Well Murphy's Law of tipping came to the fore in the AFL last week - all season the model had not tipped less than 4 tips in a week, but publish them online and sure enough I only get 2.5 correct for the round - the biggest disappointment of course tipping Western Bulldogs as a 97% chance to win and having them lose - to paraphrase The Simpsons, when you're right 97% of the time, you're wrong 3% of the time.

Still, despite having a relatively big bet on the Western Bulldogs (at $1.08!) I was able to turn a nice profit thanks to good wins by supposed underdogs in Carlton, St Kilda and Fremantle. Had Essondon not lost a big lead to end up drawing after the siren then the profit would have been even higher.

So here are the model's predictions for Round 20 of the AFL. A couple of surprising probs throwing up big overlays, mainly due to the team's selected for this week. Best of luck to those following.


Team 1Team 2VenuePr 1Pr 2
HawthornAdelaideMCG78.7% 20.0%
RichmondCollingwoodMCG22.2% 76.3%
West CoastNorth MelbourneSubiaco57.2% 40.7%
BrisbaneWestern BulldogsGabba29.2% 69.0%
SydneyGeelongANZ Stadium55.4% 42.5%
MelbourneFremantleMCG16.7% 82.1%
Port AdelaideCarltonAAMI Stadium65.6% 32.6%
EssendonSt KildaTelstra Dome22.5% 76.0%

Total tips for the season: 108 from 152 (71.1%)

Wednesday, August 12, 2009

EPL 2009/10 Season - Week 1 Predictions

The time has come to put up up the first lot of match predictions for the 2009/10 season of the English Premier League.

This EPL model has performed extremely well in the past - for the 2008/09 season a model was built using data from the 5 seasons prior and applied to 08/09 season - resulting in 214 correct tips, out of 380 games for a success rate of 56.3%. - this figure might not sound like much for those more used to AFL / NRL-type tipping, but consider in soccer there are 3 possible outcomes to tip from.

To put this in perspective, last year on the tipping website http://www.footytips.com.au/ there were 33,400 people in the EPL tipping comp for last season. As yet I cannot find the number of tips the winner ended up on, but 2nd place finished on 212 tips - so there's every chance this model on 214 tips would have finished on top.

Of course the other measure of this model's success is profitability. Betting on all teams with an overlay (an advantage over the bookies price) saw a Return on Investment of 13%, a pretty good figure in anyone's books. However if a couple of simple filters are applied then this figure jumps to 36%, which is simply outstanding.

Of course talk is cheap, it's easy to boast about how the model has gone in the past - if this season's figures come close to last year's I'll be more than happy. So without further delay here are my picks for the first week of the Premier League.

DateHome TeamAway TeamPr HomePr DrawPr Away
15/08/2009Aston VillaWigan Athletic 65% 22% 13%
15/08/2009Blackburn RoversManchester City 43% 30% 28%
15/08/2009Bolton WanderersSunderland 54% 27% 19%
15/08/2009ChelseaHull City 91% 7% 3%
15/08/2009EvertonArsenal 32% 30% 38%
15/08/2009PortsmouthFulham 41% 30% 29%
15/08/2009Stoke CityBurnley 50% 28% 22%
15/08/2009Wolverhampton WanderersWest Ham United 37% 30% 33%
16/08/2009Manchester UnitedBirmingham City 88% 8% 4%
16/08/2009Tottenham HotspurLiverpool 18% 26% 56%

Monday, August 10, 2009

A-League 2009/10 Season - Round 2 Predictions

Putting the predictions up early for Round 2 of the A-League. Round 1 fared OK, 2 correct tips out of the 5 - which sounds poor on paper, but given there are effectively 3 outcomes to each game then at least this is better than random selection! Also those who did bet hopefully saw the value on Central Coast in the first game - close to $7 was available on them where I rated them a 24% chance to win.

DateTeam 1Team 2VenuePr T1Pr DrawPr T2
14/08/09Central Coast MarinersNewcastle JetsBluetongue Stadium44.2% 26.2% 29.6%
15/08/09Gold CoastNorth QueenslandSkilled Park45.3% 26.0% 28.7%
15/08/09Melbourne VictoryQueensland RoarTelstra Dome44.5% 26.1% 29.4%
16/08/09New ZealandPerth GloryWestpac Stadium42.6% 26.4% 31.0%
16/08/09Sydney FCAdelaide UnitedSydney Football Stadium40.3% 26.6% 33.0%

Total tips for the season: 2 from 5 (40%)

Friday, August 7, 2009

AFL 2009 Season - Round 19 Predictions

Predictions for Round 19 of the AFL season follow here. This model is built on a combination of team based data - that is, the ratings for each teams based on previous results - and player based data - so the actual set of player named each week are used as inputs to the model. In effect this means if a star player was replaced at the last minute and replaced by a rookie, then their team's chance of winning is reduced.

As it stands the model is sitting on 105.5 tips for the season (counting the draw as half a tip). At the moment the leading Herald Sun "expert" tipster is on 104 tips, whereas celebrity tipster Shane Warne is on 106, no doubt buoyed in some part to tipping St Kilda every week - so the signs of this model's success look quite promising.

Of course another application for this model is to bet on the outcome of games. An easy way to see if it is worth betting on a team is to multiply the probability of them winning by the odds available, with a result over 100% indicating an advantage over the bookmaker. For example, Essendon are available at odds of $2.60 - so 63% x 2.60 = 163%, so a 63% advantage over the bookmaker.

Once an advantage is found the next question is how much to bet on the team? There are numerous methods, however the method I have used for years is a (not surprisingly) mathematically proven technique called Kelly Criteria. Some useful information can be found about it by reading these pages:

http://www.puntingace.com/bettingguide/kelly.htm http://en.wikipedia.org/wiki/Kelly_criterion

To use the example above, with our 63% advantage over the bookie for Essendon, the amount to bet would be 63% / (2.60 - 1) = 39% of your bank. Of course, this is a large amount of your bank to stake on one game, and most professional punters betting on numerous games at once will scale this figure back quite a bit.

For the record, this model is sitting on a +21% return on investment for the year so far, which is an extremely good outcome in any scenario.

I'll discuss more about the gambling side of these predictions shortly, but for now here are the probabilities for the upcoming round. Interesting that the Western Bulldogs are rated near-certainties, even at odds of $1.05 they are still apparent value.

NOTE - the predictions for the 3 Sunday games have been altered from the original picks to account for the finalised squads

Team 1Team 2VenuePr T1Pr T2
CarltonGeelongMCG37.1% 60.9%
Western BulldogsWest CoastTelstra Dome97.1% 2.7%
HawthornSt KildaAurora Stadium37.3% 60.7%
EssendonBrisbaneMCG63.4% 34.6%
AdelaideCollingwoodAAMI Stadium58.2% 39.7%
North MelbourneMelbourneTelstra Dome82.5% 16.3%
RichmondSydneyMCG56.6% 41.4%
FremantlePort AdelaideSubiaco43.5% 54.4%

Thursday, August 6, 2009

EPL 2009/10 Season Predictions

Following are my predictions on how I expect the English Premier League season to pan out.

Without going into too much detail, the entire season is simulated thousands of times. The ratings assigned to each team are updated based on the results of each game - so if a lower ranked team wins more games than expected, it's ranking throughout the simulation will increase to reflect this - and the converse for a higher ranked team.

Not too many surprises here, things look reasonably similar to last year. One point worth discussing is Manchester City - they've made quite a few off-season signings so the bookies rate them a lot better than I do. As my model is team based (as opposed to player based), all we can do is sit back and see if their performance follows the hype. If they start doing well their ratings will soon reflect this.

When comparing these probabilities against the bookies odds, it looks as though Manchester United and Chelsea are actually the only teams showing value - it doesn't often happen that the favourites in these types of markets end up being the value bets, but there you go.

Anyway I'll update this table as the season goes on and compare back to this original table to see how close or far these predictions were.


TeamPointsPr Win
Manchester United 87.7 40.5%
Chelsea 86.7 34.8%
Liverpool 83.9 22.0%
Arsenal 78.3 8.3%
Everton 64.9 0.5%
Aston Villa 58.1 0.1%
Tottenham Hotspur 53.4 0.0%
Manchester City 49.3 0.0%
Fulham 47.9 0.0%
West Ham United 46.4 0.0%
Blackburn Rovers 46.5 0.0%
Portsmouth 43.6 0.0%
Bolton Wanderers 42.5 0.0%
Stoke City 40.3 0.0%
Wigan Athletic 39.4 0.0%
Birmingham City 38.4 0.0%
Wolverhampton Wanderers 38.2 0.0%
Burnley 36.7 0.0%
Sunderland 35.2 0.0%
Hull City 32.4 0.0%

Wednesday, August 5, 2009

A-League 2009/10 Season - Round 1 Predictions

Here are my first lot of predictions, for Round 1 of the Australian A-League 2009/10 Season. Apologies for the formatting, it took me ages just to get it looking this good.

A few points worth considering regarding the A-League model in general:

  • The new teams of Gold Coast and North Queensland are rated as an average team as they enter the league - that is, there will be better and worse rated teams than these two at the start of this season. As they play more games their ratings will update accordingly and more accurately reflect where they sit with regards to other teams, so be aware of this for the first few rounds
  • In general the home ground advantage isn't as significant as it would be in say the English Premier League.
  • There is a lot more variability in the A-league than more well-established leagues. It's rare for teams to perform consisently well throughout the season - and moreso from season-to-season. Take Melbourne for example - in the four seasons so far they have finished 7th, winner, 5th and winner again. Newcastle won in 2008 yet finished last in 2009.
  • Given this, it's proven hard to make a profit betting on the A-League. Analysis has shown that only betting on home teams with a significant edge over the bookmaker's price was profitable last season

As the season goes on I'll update how many games have been tipped correctly and try and compare against any expert tipsters who publish there selections. I'm a bit wary of posting odds and betting results as like I've said A-League is just to volatile to bet on at the moment.

DateTeam 1Team 2VenueProb T1Prob DrawProb T2
Thu, Aug 06, 2009Melbourne VictoryCentral Coast MarinersTelstra Dome, Melbourne51.2% 24.7% 24.1%
Fri, Aug 07, 2009Adelaide UnitedPerth GloryHindmarsh Stadium49.9% 25.0% 25.0%
Sat, Aug 08, 2009North QueenslandSydney FCDairy Farmers Stadium43.0% 26.4% 30.7%
Sat, Aug 08, 2009Queensland RoarGold CoastSuncorp Stadium47.9% 25.5% 26.6%
Sun, Aug 09, 2009Newcastle JetsNew ZealandEnergyAustralia Stadium43.7% 26.3% 30.0%


Intro

Welcome,

If you found this site then you must like free money almost as much as I do.

Over the years I've developed models predicting outcomes various sports - with AFL, soccer and horse racing being the most prominent. I've finally decided to create this website and publish the predictions that have been profitable over the years, with the aim to track how are faring going forward.

I currently have models for the English Premier League and the Hyundai A-League, as well as the AFL which I'll aim to publish before each round of matches, along with odds for these matches and where possible suggested bets. As rounds are completed then I'll update the results for the week along with any profits / losses, as well as keep a running total throughout the season.

Also each year I come out with Brownlow predictions - from the usual overall winner through to who will post the most votes for each club. These have found varying degrees of success over the years - last year I rated Adam Cooney around 15% chance to win - paying $14 this equated to some good value - so I'll be posting those this year as well.

Horse racing has been something I've modelled for around 5 years now. It is difficult, though by no means impossible, to consistently make a profit from this given the myriad of factors and variables that come into each race. As the years go on I collect more and more data, and my modelling knowledge improves as well. When I feel comfortable with the long term success of the model then I'll post my predictions here as well, with the aim to post on a daily basis as the final fields are announced.

The modelling process I've recently developed for AFL and soccer can be reasonably easily translated through to other sports. My short term aim is to get models up and running for the NRL as well as the major US sports. Tennis modelling is also a long term aim – I see this as a huge market given there are matches almost every day of the year, and as it is a one-on-one match up – as opposed to team based with many influencing factors - I feel this has great potential for modelling and hopefully for profiting in the long term.

Anyway enough of the introductions. Given the A-League season kicks off tomorrow night I’ll put my predictions up shortly. EPL kicks off on the weekend so those predictions will appear before then as well. And once the AFL teams are announced for the weekend (yes, the players that are picked are factored into the AFL model) then they will be published. Stay tuned!