Search This Blog

Monday, September 24, 2007

Gooners NFL Spreadsheet review.

OK ... we've been running the NFL prediction spreadsheet for 3 weeks of the 2007 season and it's time to look at some results.

WE run 4 prediction systems over the spreadsheet, running from sports priced winners, to betting ATS, to value betting, and underdog plays. We've provided nicknames for each system to make it easier to follow.

Interestingly - after 3 weeks the systems are ranking almost exactly as I expected them to be - unfortunately none of them are making profits at all.

Although it's close ... so let's take a look.

#1 VINCENT (Value) 34 punts 12-21 (1 to play) -0.21% yield

Vincent is a value betting system. It will back a team if they're considered to be at odds that are better than they should be - given their chances of winning.

IF the Titans win tomorrow then Vincent will make a profit for the second straight week after a rubbish 2-8 start in week #1.

This system has virtually broken even in the first three weeks - and the ratings should get better as the effects of new players are now factoring in. I have hope for Vincent paying off.

#2 ABE (Value) 27 punts 12-12 +2 pushes (1 to play) -1.70% yield

ABE is our system for betting Against The Spread - and it's started OK - holding it's own with 12 wins, 12 losses and 2 pushes. It's also on the Titans +4 tomorrow.

Again - the system has virtually broken even - down just under 2% and should get better as we progress through the season to the playoffs.

#3 MARVIN (Most likely) 21 punts 14-7 -8.90% yield

Marvin is your "banker bet" system - simply picking the most likely winners on moneyline (ML) without caring what the price is. He went 5-1 this weekend and made a small profit of +11%.

Overall though this system is about 9% down - which almost exactly matches the over-round at the bookies - and is where I expected it to be.

It's hard to make money regularly by betting on the obvious favourites - and I'll be amazed if this makes a profit as the bookmakers would go broke if they were that easy to beat.

#4 EDDIE (Extreme Value) 19 punts 4-15 -49.42% yield

Ah ... Eddie Eddie Eddie.

This system only wants extreme value - which usually means big underdogs - and unfortunately this systems method seems only to be proving that extreme value is not there in the betting market to be found.

Basically if you're getting a great price - it's because there are other things going on that the sheet does not know about - and you're far better off ignoring these picks altogether.

It did get the Giants to beat the Redskins this week (@ 2.76) but went 1-3.

I will say that the amount of loss suprises me - but perhaps Eddie has also been a bit unlucky too - as he had Cleveland @ 2.55 but they missed a 40yd FG on the last play.


OK - so there we have it - four statistical systems for betting on the NFL - none of them beating the bookies yet (it's tough to do) - but the top two are going close enough for some hope at just 1% and 2% losses - especially as we get into the middle of the season with more data.

And the interesting difference between Vincent (value) and Eddie (Extreme Value) which highlights the same sort of "efficient betting market" arguements that power my discovered value systems for European football.

Interesting? Well it is to me ...
I'm already hanging out for the odds lines and next weeks games ...

Check out the Forum on Friday for the new sheet.