Week 4 S&P+ picks

So ... funny story. When I was redesigning S&P+ last winter, I set up my picks spreadsheet to point at S&P+ instead of F/+ (which is the combined S&P+ and FEI) so that I could test how the new number was doing in that regard. The week before the season began, as it was time to make the Week 1 picks, I dusted off those spreadsheets and gave them a quick perusal, and everything looked right. So I started in. Only ... the VLOOKUPs were still pointed at S&P+ and not F/+. And it took me three weeks to realize it. So what I've been calling F/+ picks for the first three weeks were actually picks generated from only my number, not mine and Fremeau's.

Good thing I don't advocate these for gambling purposes, huh?

So now that I've realized it, I'm going to try to do something about it. But I never leave myself enough time to actually do anything in timely fashion, so I haven't gotten it yet. So ... ladies and gentlemen, your S&P+ picks for Week 4. I really need a keeper.

So how has S&P+ done so far this year?

• Week 1: 20-17 (54%)
  
• Week 2: 25-26 (49%)
• Week 3: 28-23-1 (55%)
• OVERALL: 73-66-1 (53%)

I'll take it. As I've said before, I post these picks for validation. Since I don't really have the capability of showing everything under the S&P+ hood (and few would have interest in looking that far under the hood to begin with), I post these picks as proof that the numbers I use for analysis purposes are good enough to use. Over 50% against the spread? Win probabilities working out as they should? Then I'm happy. (This means that, when I do get the F/+ picks squared away, I'll run the picks for the first weeks of the season, too, so I can make sure they're performing as they should.)

Win probabilities

How are the win probabilities working out? Still a little too well, honestly, but that's probably because the FCS picks seem to have been quite a bit more in-line this year. Well ... with one exception. Congrats to UCF for becoming the first team with a 90%+ chance of winning ... to not win. Gotta watch out for those Furman Paladins.

As the FCS games cycle off of the calendar (until the SEC's SoCon Saturday, anyway), I would assume we'll see an evening out of those games in the 70-99% range. The evening out for 60-69% began last week, I think. If your percentages are going to be off, you want it to be in the "too good" range, but hopefully this still evens out. I use win probabilities for projecting conference races and whatnot, so I need those to fall in line to feel good about those posts.

Thursday

(No, the numbers haven't noticed that Gunner Kiel is doubtful to play tonight.)

Friday

Saturday