2014 College football rankings: Updated S&P+ projections

Right after Signing Day, as is my custom, I posted some preliminary S&P+ projections based on initial returning starting data and new two-year recruiting rankings. As is also my custom, I have updated those projections after spring football to account for attrition, better estimates of returning talent, and whatever tweaks I have made to my projection system.

This time around, my major tweak was in looking more closely at offensive and defensive trends instead of full-team trends as a whole. Instead of using a weighted five-year history figure for the team, I tinkered with the proper weights for offense and defense. It's the same with returning starters -- attrition affects the units differently.

Using eight years of returning starter data, bouncing it off of S&P+, and using some general regression, I was able to derive the following:

Inexperience matters more on offense than on defense, while experience makes more of a difference on defense, if that makes sense. Let's put that another way: returning almost no starters will hurt an offense more than a defense, while returning 10 starters helps a defense more than an offense. And returning 5-8 starters affects each unit pretty much equally.

(And yes, using returning starter data is and will always be a hindrance. The eventual goal is to come up with something better -- % of yards/TFLs/whatever returning, for instance -- but I never leave myself enough time in the offseason.)

So with these and other general tweaks in mind, here are the updated S&P+ projections as we enter the summer months. Brian Fremeau has his projected FEI figures out as well, and we will make combined use of these numbers for the F/+ projections in the Football Outsiders Almanac 2014 in July. But for now, here's S&P+.

Thoughts:

• There's still a huge lump of teams between fourth and about 18th. Because of my formula tweaks, it looks like some teams moved up or down quite a bit -- Ohio State up from 10th to fourth, Clemson up from 14th to fifth, LSU down from fifth to 12th -- but that's really not the case. A slight improvement or regression can affect your ranking quite a bit when you're clustered together with so many other teams.
• Yes, Clemson fifth. I've removed draft points from the equation for now, simply because I don't like the way I do it and don't think it has much of a positive effect, so losing Sammy Watkins will look like simply losing a starter. But the major positive impact for Clemson comes from experience on defense; the Tigers have improved by quite a bit over the last couple of years, and they return a relatively experienced unit in 2014. They're one of only three teams projected in the top 15 on both offense and defense. (The other two are pretty obvious.) You don't have to actually believe the Tigers are a top-5 team if you don't want, but you might want to set the bar a little higher than you were thinking.
• Yes, Oklahoma 17th. And UCLA 20th. I expect both teams to be better than that ... but not necessarily a LOT better than that. If nothing else, consider this a warning sign for the top-5 hype. Both teams will be good, but we need some leaps of logic to call them elite when they very much were not last year.
• When I looked into proper weighting for my weighted five-year history, the major tweak was to give last year quite a bit more heft. Five-year history is still far more effective than looking ONLY at last year, but things change a little too quickly for an evenly weighted five-year history to be effective. It stands to reason that last year carries quite a bit more weight than 2009. Anyway, one of the major beneficiaries with this tweak was Baylor, for obvious reasons. (The other obvious beneficiary: Florida State, which jumped Alabama into the top spot. They're far and away the top two.)
• LSU: 12th in the country, sixth in the SEC.
• The Kansas State projection scares me a bit. I'm really excited about the Wildcats' offense this year, but the projections are reminding me that the defense has a lot of talent to replace, even if losing starters on defense doesn't hurt as much as we might think.
• Gotta say, a No. 26 Mississippi State team is going to rack up quite a few wins with that schedule.
• Your major-conference bottom 5: No. 108 Purdue, No. 104 Kansas, No. 99 Colorado, No. 83 Wake Forest, No. 82 California.

I'll try to hammer out some win projection data with these numbers at some point. But it might take me pretty far into the summer to get to it.

Anyway, I've always viewed these projections as a way to start the conversation in the right place. There's plenty of "yeah, but..." context to add, but this is a safer place to conversation than "Team A looked AWESOME in the __ Bowl!"