NFL Elite 8 Picks and Wild Card Results

GREEN BAY -8 over Seattle
NEW ENGLAND -13 over Jacksonville
INDIANAPOLIS -8.5 over San Diego
DALLAS -7.5 over New York Giants

Results from last weekend:

All my picks 2 wins 2 losses
My home fave picks 2-1
My home dog picks 0-1
Lady G’s best bets 1-1
Readers’ choice (SD over Tenn) 1-0

I’m sticking with my overdogs-first approach for the playoffs despite the large point spreads. History shows that the spreads come into play less often in the playoffs than in the regular season.

I didn’t know what to do with the home underdog last week and lost by taking the Steelers as a home dog. Toy boy Archibald tied the game in his pool because his posted spread was Pittsburgh +2, not +1.5.

I have done a bit more analysis of how my underdogs-first approach went wrong during the 2007 regular season.

Recall that overdogs won 52% of all games against the point spread in 2007 – up from the 48% historical average. Overdogs’ better-than-expected performance in 2007 can be attributed to two sources.

First of all, overdogs won 69% of all games outright – up about two percentage points from the 67% historical average, but still well within the 64% to 70% range in two out of three years that we would expect to see based on random variation around a 67% long-run average. A 69% outright winning rate for overdogs would have generated a 50/50 year if the normal 19 percentage point gap had prevailed between overdogs’ outright and against-the-points winning percentages.

Unfortunately for yours truly and all other underdogs-first pickers, this gap shrank to 17 points in 2007 – at the bottom end of the expected 17-21 point range.

In other words, two powerful forces converged to favour overdogs in 2007 – a higher than normal outright winning percentage without the point spread and a lower than normal rate of reversal once the spread was factored in.

Time permitting between now and September 2008, I would now like to do some more analysis to see whether overdogs’ outright winning percentage without the spread has been creeping up by enough in recent years to suspect that a statistically significant change has occurred.