Fifth Pick Week Selections

Oct 16-18 College Football Picks And Post-Mortem On Last Week

Last weekend the analytics were an adventure in pure mediocrity. The positive strainer picks were 3-1 and and the negative strainer picks went 3-3 for an overall performance of 6-4. For the season-to-date the analytics are an underwhelming 17-24. This system has badly underperformed my own expectations based on the back-test of the 2013 college football season. The upcoming weekend will mark the midpoint of the 2014 college football season and I'll probably make a decision after the weekend whether or not to continue the football selections. If I don't, I'll begin to turn my attention toward the college basketball season.

One reason I'm leaning toward keeping the football portion of the blog going is the performance of some of the sub-components of the football analytics. Let me try to shed some light on what I mean.

If you have talked to me about my football analytics, the first thing I do when screening for football games is to screen the football slate for a high likelihood of line misses, i.e. games where the actual outcome has a high likelihood of being significantly different than the line. The three strainer groupings have a range of values between A) -0.05 and 0.05, B) -0.08 to -0.05 and 0.05 to 0.08 and C) less than -0.08 or greater than 0.08. During the 2013 back-test the C group had the greatest likelihood of a big line miss. The C group during 2014 has also demonstrated a high likelihood of being a big line miss.

However, I've used the same analytics as I did during the 2013 back-test this year but performance in the C strainer group has failed while the performance of the A strainer group has been excellent. In fact, the strainer group A has a 19-2 record year-to-date and strainer group B has a 17-9 performance through four picks weeks. The surprising feature of the 2014 season is that strainer group C has a 16-25 record year-to-date. During the 2013 back-test the A group was successful only 56% of the time, the B group was successful about 63% of the time while the C group etched out an impressive 73% success rate. I don't have an explanation why the performance of the ABC strainer groups are inverted this year.

The obvious answer, however, is that one season was not a sufficiently large enough sample set to make these determinations on.

Now that you are thoroughly bored and confused, here are this week's picks. The analytics flagged eight games total. There are five favorites and three underdogs.

Positive strainer picks

South Florida -2.5 @Tulsa

@Colorado State -5 Utah State

Clemson -6.5 @Boston College

Virginia +2.5 @Duke

Kansas State +8 @Oklahoma

Negative strainer picks

Georgia Tech -2.5 @North Carolina

@Arkansas +3.5 Georgia

Army -6 @Kent State

As always I wish everyone the best.

More to come soon - Jack

Year-To-Date Record

17-24