Monday, 13 September 2010

Mathletics


A great weekend for anyone using the lay system written about last week with 9 winning lays from 13 selections and a first weekend post-commission ROI of 78.28%.

Germany 2 bets, 2 winners, +3.90 points

France 6 bets, 4 winners, +3.17 points

England 5 bets, 3 winners, +3.10 points

Football Elite had a great weekend too with the one Recommended Bet of Lecce winning 1-0 at 3.5. The Short-List selections fared well too, with Nice winning at 3.2, Cesena beating AC Milan at a nifty 7.4, and Brescia beating Palermo at 3.35. They also had Real Zaragoza to beat Malaga, but after spotting Malaga a 5-0 lead, Zaragoza could only pull three goals back. The last team to score 5 (or more) away goals before HT in La Liga was Barcelona who were 6-0 up at Real Sociedad in October 2000. Football Elite's final short-list selection was Hannover who drew with Bayer Leverkusen.

Elo Summary - England: Birmingham's draw with Liverpool was predicted by the Elo ratings, and was a short 3.3. No doubt the previous six matches between these two teams all finishing as draws was a factor! The other predicted draws in the Premier League were Wigan Athletic v Sunderland (3.35) and tonight's Stoke City v Aston Villa match (3.3). Teams expected to win by two were Arsenal and Chelsea. Arsenal added on late to miss the Correct Score, but Chelsea obliged winning 3-1.

Spain: One draw forecast here, but no winner as Athletic Bilbao lost 1-2 to Atletico Madrid. The 2+ goal expected winners were Barcelona and Real Madrid. No one who is into football betting, or who reads this blog, will be unaware that Barcelona started at a very short 1.07 at home to Hercules, and had not lost at home to a promoted club since 2000. It's rare enough that they are down at home at HT, with it being 16 months since they trailed Osasuna by 0-1 and they couldn't save that game either. Hercules won 2-0. Real Madrid won 1-0 v Osasuna.

Italy: No less than five draws predicted here, but just the 3-3 draw at Juventus obliging. A two goal win was expected by Internazionale who won 2-1. AS Roma lost away by 4 goals for the first time in 20 years (since going down 0-5 at Juventus in 1990) and for the first time in 12 years, AC Milan have lost four consecutive away games.

Germany: The league that defies the odds had two predicted draws, but resulted in 2-0 and 2-1 home wins.

France: Three draws predicted, with one winner (3.15) at Brest.

This weeks sees five inter-league matches with the European Champions League Group matches getting underway. There is also one Europa League match where I have ratings for both teams. Last season, the ratings in these inter-league matches performed better than I expected, a trend I hope will continue.

Not such a great weekend for the self-titled Sports Betting Professor, who continues to send me his picks. I may start using them - as lays. He had no less than 10 College Football picks this weekend, and nine lost. That's not easy to do!

How he comes up with so many picks this early in the season is beyond me. In professional sports, off-season turnover is significant enough to warrant caution on the fisrt few games of a new season, but the nature of college sports means that quite often the majority of a previous season's starting line up has graduated or is no longer eligible to play. There is often simply no continuity whatsoever, meaning that early season predictions are based almost entirely on the subjective opinion of a few 'experts'. The Sports Betting Professor appears to be no expert.

I had some time to start reading a new book called "Mathletics" by Wayne L Winston yesterday. Sub-titled "How Gamblers, Managers, and Sports Enthusiasts Use Mathematics in Baseball, Basketball and [American] Football".

The opening chapter covers Baseball's Pythagorean Theorem, something I touched back in April 2009, though not in too much detail. The more runs a baseball team scores, the more games the team should win. Conversely, the fewer runs a team gives up, the more games the team should win. Anyone who read Michael Lewis's excellent "Moneyball" will be familiar with Bill James and sabermetrics who discovered that the percentage of games won by a baseball team can be well approximated by the formula:

Runs scored [squared] / (Runs scored [squared] + runs allowed [squared])
In the final stages of the regular 2010 season, it's a good time to look at the records of teams and see who is currently under-performing as the play-offs near. The St Louis Cardinals lead the way, a full 6 wins behind 'where they should be', followed by the Atlanta Braves, Washington Nationals and San Diego Padres who are 4 wins behind schedule.

Over-achievers who might be expected to slip a little in the final few weeks are the Houston Astros (+8), Pittsburgh Pirates (+6) and the Kansas City Royals (+5).

Of these teams in the race for the play-offs, the Atlanta Braves are just one game behind the Philadelphia Phillies, and the San Diego Padres are tied with the San Franciso Giants (and just 1.5 games ahead of the streaking Colorado Rockies). The Cardinals are 6 games back from the Cincinnatti Reds, and probably out of it. Interestingly, the Padres play three games at Colorado starting today, followed by four at the St Louis Cardinals. They have 20 games left to play, but this week will have a big say on their play-off chances.

I haven't yet reached Parts II and III on Football and Basketball, but I hope they are as interesting. Unlike baseball, where the play-off schedule means that teams typically use just three starting pitchers as opposed to five in the regular season, and thus regular season performances are a relatively disappointing predictor of play-off prospects, (53.8%), figures from basketball and football should project play-off results more accurately.

Has any similar formula for football ever been proposed? It seems reasonable to think that there should be a correlation between corners, shots, shots on target, and goals, but I've not seen anything serious on the subject.

The NFL is back, although I unfortunately had to miss all the games on Sunday's schedule. A few punts (pun intended) resulted in a small profit, but Week One of the season is not a time to be too bold. Incidentally, the Sports Betting Professor is no better in the NFL it seems, with his three picks for Sunday's game all failing to win, although the Cleveland Browns +3 @ Tampa Bay Buccaneers was a wash.

The New York Jets take on the Baltimore Ravens tonight followed by AFC Champions in waiting San Diego Padres at the Kansas City Chiefs.

My Elo ratings have the Ravens favoured by 1, and getting 2.5 they are available at 1.93.

In the second game, my ratings have the Chargers favoured by 6, and giving 4.5 are 2.02.

2 comments:

Kobayashi said...

Cassini, great blog, I've not kept up that much recently but having more time now so I can check in every few days.

Would you do (or maybe you but I missed it) a post on how the ELO setup you have was created? I don't expect you to give it all away, but I'd like to do something similar myself, concentrating on just one league.

It's really the analysis side of it that interests me, with any small winnings just showing if I was right or not :)

On the subject of the SBP, what bothers me most (apart from his cack record so far) is he seems to leave out "valid" selections (as per his manual) from the picks he sends. I assume he applies some analysis of his own, but doesn't say what. Do you feel there's actually anything to suggest the spreads he uses have any edge?

Cheers,
Matt

Anonymous said...

The Mathletics book has some useful stuff in it. Nothing specifically on "soccer" as it's US based but plenty which "crosses over".

You will likely find the section on ratings useful to further develop your "soccer" ratings.