NCAA Elite 8 Sunday Picks

Memphis -3.5 over Texas
Kansas -9.5 over Davidson

My results against the spread:

All my picks 35-24
My overdog picks 31-16
My underdog picks 4-8

An all-overdog strategy still leads the pack at 39-20.

NCAA March Madness Elite 8 Saturday Picks

North Carolina -5.5 over Louisville
UCLA -6 over Xavier

My results against the spread:

All my picks 30-23 57%
My overdog picks 26-15 63%
My underdog picks 4-8 33%

An all-overdog strategy still leads the pack at 34-19 or 64%.

NCAA Sweet 16 Picks and Results So Far

My results against the spread:

All my picks 27-22 55%
My overdog picks 24-14 63%
My underdog picks 3-8 27%

I can't claim to be an NCAA Nostradamus when an all-overdog strategy is way ahead at 32-17 or 65%.

My picks for the sweet 16 match-ups:

North Carolina -7.5 over Washington State
Louisville -2.5 over Tennessee
West Virginia -1 over Xavier
Western Kentucky +12.5 over UCLA
Kansas -11.5 over Villanova
Wisconsin -4.5 over Davidson
Memphis -4.5 over Michigan State
Texas -1.5 over Stanford

NCAA 2nd Round Picks

My results against the spread through the first round including the play-in game:

All my picks 19-14 58%
My overdog picks 16-7 70%
My underdog picks 3-7 30%
Spread a factor 3-0 100%
All overdog strategy 23-10 70%

My 2nd round picks:

Saturday

Xavier -3 over Purdue
Duke -3 over West Virginia
UNLV +13.5 over Kansas
Wisconsin -4.5 over Kansas State
Pittsburgh -2.5 over Michigan State
Washington State -2.5 over Notre Dame
Stanford -3 over Marquette
UCLA -10 over Texas A&M

Sunday

Louisville -7.5 over Oklahoma
Tennessee -4.5 over Butler
Memphis -9 over Mississippi State
Texas -6 over Miami
North Carolina -11 over Arkansas
Georgetown -5 over Davidson
Villanova -5 over Siena
Western Kentucky -5.5 over San Diego

NCAA 1st round selections

I was chuffed to start 1-0 by backing Mount St. Mary’s -7 in the play-in game.

In the first round, Lady Godiva fancies:

Thursday games

Belmont +20 over Duke
West Virginia -2 over Arizona
Xavier -8.5 over Georgia
Purdue -3 over Baylor
Portland State +22.5 over Kansas
Kent State -2 over UNLV
Wisconsin -11.5 over California-Fullerton
USC -2 over Kansas State
Notre Dame -6.5 over George Mason
Washington State -8.5 over Winthrop
Michigan State -6.5 over Temple
Pittsburgh -9 over Oral Roberts
Cornell +14.5 over Stanford
Marquette -6 over Kentucky
MVSU +32 over UCLA
Texas A&M -2 over Brigham Young

Friday games

Clemson -6 over Villanova
Vanderbilt -6.5 over Siena
Drake -4 over Western Kentucky
Connecticut -11 over San Diego
Mount St. Mary’s +31.5 over North Carolina
Indiana -1 over Arkansas
UMBC +16.5 over Georgetown
Davidson -2 over Gonzaga
Texas Arlington +24.5 over Memphis
Mississippi State -2.5 over Oregon
Austin Peay +15.5 over Texas
Miami -1 over Saint Mary’s
American +19.5 over Tennessee
Butler -4 over South Alabama
Boise State +13 over Louisville
Oklahoma -1.5 over Saint Joseph’s

22 overdogs and 10 underdogs

Be warned. I do not follow basketball. My picks are based on academic analysis of statistics.

It is interesting to note how different college basketball is from professional football. Overdogs favoured by more than 12 points win 95% of college basketball games outright. I don't know the comparable NFL stats, but I doubt that the outright winning percentage of heavy NFL favourites is anything close to 95%. However, in college basketaball the point spread turns nearly half of overdogs' straight-up victories into losses or ties against the spread.

NCAA Stats and Qualification Game Pick

Toy boy Archibald's pal Jarhead refuses to release his NCAA men's basketball pool picks. I shouldn't be surprised that an ex-military man would classify everything top-secret.

But, Jarhead did point me to two academic analyses of NCAA historical data:

http://bpp.wharton.upenn.edu/jwolfers/Papers/PointShaving.pdf

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1002691#PaperDownload

The professors disagree about the particular issue they are debating -- whether large numbers of college basketball games are fixed. In my personal opinion, the two professors are more persuasive who discount the game-fixing story in favour of their explanation of how basketball games unfold as the clock winds down.

But, all the professors agree about the basic data. And, for purposes of playing friendly, non-profit office pools, all that counts is whether there appear to be any small opportunities to exploit. One interesting fact to come out of these papers is that a simulation reducing lines by half a point raises the overdog winning percentage against the point spread by 2 full percentage points.

Based on what I learned in these articles, here is my pick for the qualification game:

Mount Saint Mary's -7 over Coppin State

NFL Post-Mortem

Apologies to loyal Godivans for my late post-Super Bowl post. The week after Super Sunday I was racing to finish a project with a 31 January deadline that I missed by a kilometre. Then, toy boy Archibald and I spent the past month shuttling back and forth caring for Archibald’s dad, who has now recovered I am chuffed to report.

Super Sunday provided a fitting ending to a dismal year for Lady Godiva. My only consolation from my 3-8 playoff record is that 27% in the playoffs makes my 47.5% regular season record look good.

I hope to file a new post Thursday with NCAA March Madness tournament picks if toy boy Archibald can persuade his pal Jarhead to give up his computer-generated picks. As a former schoolgirl netball player back in dear olde Coventry, I have only the vaguest idea of what basketball is all about. So, I will be relying on Jarhead who claims to have a system for playing friendly, non-profit NCAA pools that will work more often than not. We shall see if Archibald can convince Jarhead to share his secrets.

To cap off the gridiron season just past, here are some random observations:

Overdogs still rule in the Super Bowl with a 21-19-2 record or 52%. But, a sample size of 42 games is too small to draw any firm conclusions. I have been treating the Super Bowl like any other overdogs-rule playoff game, but Professor Levitt may well be right that playing the Super Bowl underdog is the best approach. Unfortunately, even a long lifetime of 90 years is probably too short to avoid the small-sample problem. Playing the Super Bowl underdog may be a sensible bet over the long run, but as the great economist Keynes quipped: ``Ìn the long run, we are all dead.`` In other words, it will take far longer than 100 years to find out whether Lady Godiva or Professor Levitt has a better handle on Super Sunday.

I did not realize this until just recently when I checked the record, but my playoff picks have been dismal – 43% from 2002 through 2008 vs. my 52.4% record over the past 7 regular seasons. My playoff record has been even worse than the 45% winning record for overdogs from 2002 to 2008. In other words, my few deviations from my overdogs-first playoff approach have served only to make matters worse. By contrast, my 52.4% regular season average is a touch above the 51.7% winning record for a plain-vanilla, all-underdogs strategy.

My faith that overdogs dominate the playoffs is based on overdog dominance from 1992 to 1998/9 when favourites won 75% of all playoff games outright and 60% against the spread. But, from 1999/2000 on overdogs have won just over 65% of all playoff games outright and only 46% against the spread. The overall playoff record of overdogs looks good over the past 16 years – 70% winners outright and 52% against the spread. But, this record can be divided into two distinct periods and the most recent 9 years have been dreadful for overdogs.

In any case, 176 playoff games over the 16 years from 1992 on is too small a sample to draw definite conclusions. Overdogs won 52% of 256 games played during the 2007 regular season, but one regular season is not enough to support a NFL pool strategy for all time to come. Based on overdogs’ 48% record over the past 16 regular seasons, I will continue recommending an underdogs-first approach when the 2008 regular season kicks off in September.

But, the same logic leads me to question my overdogs-first approach to the playoffs. Contrary to my picking approach, playoff and regular season games may not differ at all. Overdog playoff dominance from 1992 through 1998/9 may simply be a random deviation from the underdog norm in the same way that overdogs’ 52% record over the 2007 regular season deviated from the 48% historical norm. If time permits, I would like to dig up regular season and playoff records prior to 1992 to see whether underdog dominance in the regular season and the overdog playoff advantage persists in the pre-1992 data.

Check back next week to see whether I have obtained Jarhead’s recommendations on how to play March madness NCAA pools.