Week 14 BTBS Picks

by Bill C. on Dec 3, 2009 9:00 AM CST in National Football 2009

Sigh. Let's just get this season over with already. Vegas has found that screwing me over is a lot of fun, and I'm not appreciating it.

Last Week: As if the week before weren't bad enough, I missed 13 of last week's 49 games by 3.5 points or less. THIRTEEN. Just keeping up with the scores on the ticker, I thought I was doing wonderfully on Friday and Saturday. I was not. Needless to say, it is hampering my confidence in the abilities of the S&P+ system to be a good money maker. Yes, my first stab at season-long projections went strikingly well, but week-to-week ATS picks? Not so much...yet.

Category	Last Week	Season
All Games	21-28 (42.9%)	346-302-8 (53.4%)
Big 12	1-4 (20.0%)	46-39-2 (54.0%)
"LOCKS"	1-1 (50.0%)	30-30-2 (50.0%)

The offseason plan is two-fold: 1) I'll do some analysis on which type of games I was most likely to miss and by how much (the goal is to create a spreadsheet I can send to people via request to see if they can spot trends I can't), and 2) once I have 2004-06 play-by-play data entered, go back to the beginning of 2004, set up the system to pick games week-to-week from 2004-09, and tinker until I find the "system" that works the best.

Oh yeah, and I plan to do a ton of work on my preseason projections system (having 2004-09 data to look at instead of 2007-08 is going to make a world of difference, and I'm really excited about the prospects) and different ways to measure defensive success. It is indeed possible that I'll be busier in the offseason than I have been during the season.

Anyway, here are this week's picks. Conference title games (and de facto title games) are in bold.

Date	Time	Game	Projection	Spread	ATS Verdict
Thurs., 12/3	6:00pm	Arkansas State at Western Kentucky	WKU by 6.5	WKU +6	WKU (WIN)
	8:00pm	Oregon State at Oregon	UO by 21.7	UO -10	UO
Fri., 12/4	7:00pm	Ohio vs Central Michigan	CMU by 4.4	CMU -13	Ohio (WIN)
Sat., 12/5	11:00am	Cincinnati at Pittsburgh	Pitt by 3.6	Pitt +2	Pitt (WIN)
		Houston vs East Carolina	UH by 3.5	UH -3	UH
		West Virginia at Rutgers	WVU by 10.6	WVU +1.5	WVU (WIN)
	11:30am	Fresno State at Illinois	Illini by 3.3	Illini -3	~~Illini~~
	1:00pm	San Jose State at Louisiana Tech	La Tech by 18.6	La Tech -23.5	SJSU
	2:00pm	New Mexico State at Boise State	Boise by 47.9	Boise -47.5	Boise
	2:30pm	Arizona at USC	USC by 3.2	USC -7	Arizona
	3:00pm	Florida vs Alabama	UF by 6.6	UF -6	UF
	5:30pm	California at Washington	Cal by 2.4	Cal -7	U-Dub
	6:00pm	Florida Atlantic at Florida International	FAU by 0.1	FAU +2	FAU
	7:00pm	Georgia Tech vs Clemson	CU by 5.2	CU +1	CU
		South Florida at Connecticut	UConn by 11.2	UConn -7.5	UConn
		Texas vs Nebraska	UT by 11.8	UT -14	NU
	10:30pm	Wisconsin at Hawaii	Wiscy by 2.3	Wiscy -12	Hawaii

Really, Vegas? Really? I have Boise by 47.9, and the line is Boise -47.5? Really? GET OUT OF MY HEAD, VEGAS. Ugh.
In all, six of 17 games are projected within three points of the line (and another handful are within four or so), so depending on my luck, this could be really good or really bad. And considering my luck's been atrocious lately, assume really bad.

Tweet Comment 5 comments | 0 recs |

Do you like this story?

Comments

Tracking +/- 7 Points

Just a thought before you dive too deep into your offseason analysis…
Have you tracked your entire numbers when looking at a differential of 7 points in your projection and the spread?

I think there might be some value at that difference just from eye balling it through the picks week to week, although I have not actually tracked the numbers myself.

by Escoot on Dec 3, 2009 2:08 PM CST reply actions

Looking for trends, etc.

I would definitely be interested in looking for trends. I actually began doing some “meta” analysis on your picks this year (I posted about it in a comment thread several weeks back) and ended up stopping only because putting all the data into an excel spreadsheet proved more tedious than my first-year-associate-attorney day job would permit. So, yeah, I would love to take a look at a pre-compiled spreadsheet and test out some of my theories as to why a lot of the projections are spot on, but then there are the occassional few that make no sense to me at all. For example:
-it boggles my mind how a computer would favor WKU over anybody outside of a Junior Varsity High School team.).
-there were certain teams that remained overvalued all season
-Why did “locks” do worse than the entire set of picks? Along these lines, I have a hunch that there is a certain range of where the projections likely did the best. I would venture to guess it is the range just south of the “locks” range…something liike a 9-12 point discrepancy or something. Those would be the first numbers I would run. Break all the picks into bins depending on how much the projection differered from the spread:
Bin 1: 0-3 points
Bin 2: 3-7 points
Bin 3: 7-10 points
Bin 4: 10-14 points

See which bin had the highest rate of success, and then speculate on why the additional “Lock” discrepancy range didn’t go as planned.

I will shoot you an email because, believe it or not, my legal name is actually not “kidbourbon” (although it is a very solid moniker, if you ask me).

________________________________
Eric Berry is better at football than you.

by kidbourbon on Dec 3, 2009 2:50 PM CST reply actions

I have been tracking that

0-3 points: 110-112
3-6 points: 88-66 (57%)
6-10 points: 81-59 (58%)
10+ points: 65-65

What’s interesting is that the 10+ difference from spread category was by far the most reliable category through week 6 and since then it has been by far the worst category. As if to say there was an advantage early on and now the 10+ differences are just outliers than Vegas has fixed and this system has not.

by danman31 on Dec 3, 2009 4:12 PM CST up reply actions

Is this around the time that last season/pre-season stopped working for this seasons' outcomes?

This is really interesting.

Will Ebners Hit Parade, Pain TV; Channel 32; All the time! (PDT)

by MarioVanPeebles Republic of China on Dec 3, 2009 7:15 PM CST up reply actions

Projections

When you run projections for your numbers, do you come up with actual score predictions, i.e., USC 41, Missouri 21, or just that your projections say USC by 20?

I am wondering how well the S&P+ Ratings would do in predicting totals for games, in addition to spreads, if in fact you can run numbers for that.
It seems logical having OFF S&P+ and DEF S&P+ go “against” each other to determine totals.

by Escoot on Dec 4, 2009 11:49 PM CST reply actions