I tried to edit my post, but was too late:

Edit: wait, the example used a 4 teamer, so that is the only reason right?

So technically if 4 teams were wongable we would have (4) 3 team teasers in addition to the (6) 2 teamers, and (1) 4-teamer? That would tease every possible combination.

What bet amount is ideal at each teaser level?

5% for 2 teamers
3%? for 3 teamers
2% for 4 teamers
1%? for 5 teamers

Read the linked post by TomCowley.

Anyone know if there is a correlation between WIN_PCT and the o/u for Wong-able games?

One of the things I'm going to look at once I put my database of score archives together. I would logically expect there to be one, but I have no idea how big it is.

i know there is a big difference in college...lower totals have higher win_pct...don't know what you are asking though

Hmmm, maybe I am asking the wrong question, because one would think that lower o/u matters more if the line has been teased up to 7.5-8.5 for a dog than if teased down to 1.5-2.5 for a favorite. Or maybe it doesn't matter at all.

Pretty negligible difference, but the line of thought likely has other applications...

Will it work on college games?
NO

math in
this thread


oh yeah, I have some definitive news in general

basic teaser strat (-7.5 to -8.5 and +1.5 to +2.5) has not been +ev for NCAAF from 1993 through 2006 for -110 2-team pricing

here's the breakdown for NCAAF:

Home Faves of 7.5 to 8.5
232-101-2 for 69.67%

Road Faves of 7.5 to 8.5
166-66 for 71.55%

Faves of 7.5 to 8.5 (home, road, nuetral)
422-180-2 for 70.10%

Home Dogs of 1.5 to 2.5
203-85-2 for 70.49%

Road Dogs of 1.5 to 2.5
245-94-1 for 72.27%

Dogs of 1.5 to 2.5 (home, road, nuetral)
494-193-3 for 71.91%


note1: I also looked at 6.5, 7, and 10 pt teasers. The additional points added very little and I'd call them a total ripoff for NCAAF.
note2: pushes excluded for percentages

I've been crunching numbers on NCAA football data tonight to determine if there are any subsets for which Wong teasers could still be profitable. I think I have found some significant data but I'd like some others to look it over and see whether you agree.

I looked at all NCAA division 1 football games from 2001-2006 and the current 2007 season, for which the lines were between 1.5 and 2.5 or 7.5 and 8.5.

The overall numbers agree with what has previously been posted, that the winning percentages are a bit below the breakeven percentage for 6-pt 2-team -110 teasers (0.7237):

ALL
Season Wins Losses Win %
2000 64 19 0.7711
2001 72 24 0.7500
2002 76 40 0.6552
2003 70 33 0.6796
2004 74 22 0.7708
2005 77 26 0.7476
2006 70 29 0.7071
2007 34 15 0.6939
Total 537 208 0.7208

However, the numbers for Road teams, specifically Road Underdogs are better:

ROAD
Season Wins Losses Win %
2000 29 5 0.8529
2001 37 10 0.7872
2002 35 16 0.6863
2003 30 17 0.6383
2004 35 14 0.7143
2005 37 7 0.8409
2006 31 12 0.7209
2007 14 6 0.7000
Total 248 87 0.7403

ROAD UNDERDOGS
Season Wins Losses Win %
2000 14 2 0.8750
2001 19 7 0.7308
2002 22 9 0.7097
2003 18 4 0.8182
2004 26 5 0.8387
2005 26 4 0.8667
2006 22 11 0.6667
2007 9 5 0.6429
Total 156 47 0.7685

Next, I looked at a subset of these games where the O/U total was less than 50 points. The overall numbers were uninteresting (222-86, 0.7208 - same percentage as the overall totals), however the edge for road teams and particularly road underdogs became even more noticeable:

ROAD, total < 50
Season Wins Losses Win %
2000 7 1 0.8750
2001 11 3 0.7857
2002 10 4 0.7143
2003 14 6 0.7000
2004 16 6 0.7273
2005 16 1 0.9412
2006 26 8 0.7647
2007 6 2 0.7500
Total 106 31 0.7737

ROAD UNDERDOGS, total < 50
Season Wins Losses Win %
2000 4 0 1.0000
2001 8 2 0.8000
2002 7 1 0.8750
2003 7 1 0.8750
2004 12 4 0.7500
2005 13 1 0.9286
2006 17 7 0.7083
2007 5 2 0.7143
Total 73 18 0.8022

Next I looked at games with the O/U total less than 47 points. This data showed even more of an edge for Road teams and particularly Road underdogs.

ROAD, total < 47
Season Wins Losses Win %
2000 3 0 1.0000
2001 6 2 0.7500
2002 7 1 0.8750
2003 10 5 0.6667
2004 11 5 0.6875
2005 10 0 1.0000
2006 20 4 0.8333
2007 5 2 0.7143
Total 72 19 0.7912

ROAD UNDERDOGS, total < 47
Season Wins Losses Win %
2000 2 0 1.0000
2001 5 2 0.7143
2002 4 0 1.0000
2003 4 1 0.8000
2004 9 4 0.6923
2005 8 0 1.0000
2006 15 3 0.8333
2007 5 2 0.7143
Total 52 12 0.8125

For the under 47 numbers, the sample sizes are getting pretty small as there just aren't that many games/year that meet the conditions. But I think the data clearly shows that 2-team 6-pt -110 Wong teasers have been profitable for road teams in expected low-scoring games over the past several years. Since there are so many more college games, the sample sizes for the Road teams in general are actually larger than for the NFL data that Wong originally used. I think betting these games with a total < 50 is a good bet, although that may not be the optimal cutoff number.

If people are interested in this, I can go back and add a few more seasons. The database I use goes back to the 1996 season but the data starts getting shittier pre-2001 (games listed with 0 as the total, etc.). I could also get the results for some different point totals besides 50 and 47.

Feedback is appreciated; if I'm way off here and/or overlooked anything, please let me know.

Interesting stuff Mogwai. Is the cover percentage ATS roughly 50%? The percentage that the teaser wins isn't nearly as significant as the percentage that hits on the teaser's "sweet spot."

For instance, if the subset went 55% ATS for some reason, of course we'd expect to see the subset's teased lines jump 5% as well. This leads to a false expectation of +EV for these teasers since betting straight up ATS might be better.

J

You are correct. The ATS is high in his sample.

this is exactly right and something I mentioned in my post

straight ATS are way above 50% for this subset...significant by any binomial test by a large margin
skip the teaser and just bet that subset ATS...you'll thank me later

they are actually >50% straight up to win the game over 300+ trials from 1993 to 2006, which means the ML is what you should really be hitting

Mogwai, please take a canary with you next time.

Ok, that makes sense. I'll run the numbers tonight and see what the exact results were for the unteased line, but I'm sure you're right. If straight betting this subset ATS is more +EV than teasing it, that's even better since you have more opportunities to do so (probably fairly often you only have one team per week in this situation). I can also look at expanding the line ranges further than {1.5-2.5,7.5-8.5}. Sorry if this is commonly known stuff, I'm new at this.. thanks for the replies.

um...definitely not commonly known or the lines wouldn't continue to show such a positive bias over time

I'd prefer not to make it commonly known

it's definitely known among frequent bettors that work at this stuff

a very good investigation all the way around mogwai

Mogwai,

do you mind sharing your raw data files?

Thanks man! Your posts have been a big help to me in learning this stuff.

I have been using atsdatabase.com. The site annoys me but it has been the best source of free data that I could find. Scroll all the way to the bottom of the main page and click on the "super database" for the sport you want. For most of the work I've been doing, I just grab all the games for one year at a time off of there, dump them into Excel, and then do filtering/sorting/math on the data there.

I'm definitely interested in finding a better database.

#ill give it a look. Thanks!

Sunshine Forecast is a good source for ATS info and it allows you to download into Excel (CSV file) or, if you prefer, as a XML file.

Here's the link...

http://www.repole.com/sun4cast/history/scoresmenu.html

http://www.sportsdatabase.com

It has queriable databases for NBA, NFL, NCAA football and MLB. Their sister site (http://www.killersports.com) has results of 6-point teasers as well as 10-point and 13-point monster teasers.