NFL Point Spread Picks Week 7 - 2013

Pick 1: Baltimore +1.5 incorrect NFL point spread pick
Pick 2: Carolina -6.5 correct NFL point spread pick
Pick 3: Philadelphia -2.5 incorrect NFL point spread pick
Pick 4: Seattle -5.5 correct NFL point spread pick

Very little time to talk this week. t-10 minutes til the kickoff of Thursday's game. Great week last week!

We continue by betting 15% of bankroll. For Week 7 we have a total of 4 picks so we bet evenly across all games, $375 per game. Good luck everyone and the only best ROI for the season.

GameVegas LineEstimatePrediction-VegasConfidence
BALTIMORE @ PITTSBURGH -1.5 7.6 9.1 60.7%
ST LOUIS @ CAROLINA -6.5 -13.5 -7 60.3%
DALLAS @ PHILADELPHIA -2.5 -4.8 -2.3 59.6%
SEATTLE @ ARIZONA 5.5 8 2.5 59.0%
CHICAGO @ WASHINGTON -1 4.2 5.2 54.0%
TAMPA BAY @ ATLANTA -7 -7.2 -0.2 53.1%
BUFFALO @ MIAMI -8 -8.4 -0.4 53.1%
CINCINNATI @ DETROIT -3 0.1 3.1 56.9%
SAN FRANCISCO @ TENNESSEE 4.5 16.2 11.7 53.2%
NEW ENGLAND @ NY JETS 4 9.8 5.8 52.1%
CLEVELAND @ GREEN BAY -10 -11.6 -1.6 52.1%
SAN DIEGO @ JACKSONVILLE 8 12.2 4.2 52.0%
MINNESOTA @ NY GIANTS -3.5 7.2 10.7 51.0%
HOUSTON @ KANSAS CITY -6 -7.8 -1.8 50.0%
DENVER @ INDIANAPOLIS 7 8.0 1.0 49.1%

How to read the table:
  • Vegas Line: A NEGATIVE number implies the point spread favors the HOME team
  • Estimate: NFL Pickles' point spread prediction
  • Pred-Vegas: Subtraction of the previous two. POSITIVE implies VISITING team will cover point spread.
  • Confidence: The probability that the point spread pick is on the correct side.


Just said...
This comment has been removed by the author.
Just said...

If I am reading the chart right Dallas should be your pick. Unless something else is wrong.

Jaime Brugueras said...

You're right, but my spread was entered incorrectly. A minus sign mistake, yet the probability is the same. Philly is the pick. Thanks for pointing it out.

Jonathon Grant said...

Jaime - Tried doing some reading up on how you calculate confidences. It's hard searching through old posts, but from what I gather, you have 80 different buckets that each game can fall under, and the confidence is simply how often that bucket has been correct. Am I right on this?

My hunch is that you have 8 or 10 Vegas Line columns, and 10 or 8 Estimate rows (from the model). It's irrelevant exactly how it's set up, but have you looked at smoothing the performance of the buckets? E.g. if a bucket is only hitting 75%, but all its adjacent buckets are only hitting 40%, is the 75% accurate, or could it just be noise.

I think some form of smoothing could be useful. Looking at week 2 (unofficial, I know), STL@ATL and SD@PHI both had lines of -7. Your estimates were -.2 and +3 respectively. It doesn't make sense to me that you could be more confident in the -.2 game than the +3 game.

Not trying to be critical. I've been using this site for years and love it - just want to throw some ideas out and see if you've looked at them already or want to consider them.


Jonathon Grant said...
This comment has been removed by the author.
Jaime Brugueras said...

@jonathon you got it buddy. I do that back tested for 15 years. Your smoothing and preserving order ideas are very good ones. We should talk one day.

Jaime Brugueras said...

Also, sample sizes are considered although not published