So, tweak one is this: I modified my algorithm to account for scoring differences. There's a ranking system called the Simple Ranking System. You can read about it here: SRS. I am downloading the SRS numbers and dividing them by 100 to bring the decimal to the right place and adding the points to my current scores.
Why the change? I believe that because there are so few out of conference head to head games that matter, we need to measure whether or not teams are winning the games they're supposed to by the amounts they're supposed to. The 1st and 2nd level points handle the first part, and the new SRS component answers the second part.
Tweak two: New calculation for the actual rankings. Instead of just adding up all the numbers and ranking the teams based on that, I decided to use rankings within individual components, averaged. If Team A is ranked #1 in 1st level pts, #3 in 2nd level pts, and #5 in SRS, its average is 3. If Team B is ranked #2, #1, and #8, its average is 3.667. Team A would be ranked above Team B. So far I like it, and there's very little difference between this and just adding the points.
Tweak three: I added tiebreakers. Once everyone had finished their last games (except Army v Navy), I discovered more than a few ties. I won't bore you with the tale of how long it took me to automate the spreadsheet to break ties (most of Sunday). Suffice to say that I got it working. First tiebreaker is 1st level points, second tiebreaker is 2nd level points. So far, zero ties using that. I can go to wins next if needed.
Here's where I'm at:
- I finally figured out how to embed my Google sheet into the blog, and I made that the Featured Post. I think that should pin it to the top, and it should reflect any changes made to the spreadsheet on the fly.
- I feel like the latest iteration of the rankings is fairly comprehensive, and it doesn't require me to know multivariate calculus to figure the SRS since the folks over at Sports Reference have already done the heavy lifting. With my system being retrodictive and SRS being predictive, I think I'm covering all aspects. The next question is weighting. That is, how much weight to give SRS as the 1st and 2nd level points are auto-weighted by design.
- I will be doing a weekly "No Season Left Behind" look back with my system once I get a few things ironed out in the spreadsheet. I'll be taking historical data from SR and dropping it into my system to declare new (maybe) national champions.
Anyway, there's more on my mind, but here's my take on Rivalry Week outcomes and impacts. Really, only 3 games mattered because these 2 were the only ones without foregone conclusions:
#7 Utah vs #55 Colorado - I include this for the implications for the CFP and not necessarily because of the impact on Ray's Rankings. I'd seen Utah play a little this season, but this was the first time I was able to stay up late enough to watch an entire game. In the first quarter I was more impressed with Colorado than the Utes. Eventually (right around the start of the 2nd qtr) Utah stopped making mistakes and just crushed the life out of the Buffs. At times they looked sharp enough to beat anyone, at others they looked like a mid-tier Pac12 team. Can't make a judgement after watching one game, but they jumped up 4 spots in this weeks rankings.
#16 Alabama vs #17 Auburn - I include this one also for the implications for the CFP. First. this was a thoroughly entertaining game, probably the best game of the weekend from a pure fan point of view. Back and forth, neither team caving under the pressure (to be expected from the Tide, never quite sure which Tiger team we'll see) until the very end. When Bama scored to make it 45-40 with around 13 min left, I didn't know how Auburn would react, but Bo Nix just took them down the field with a 5 minute drive capped with a TD and a 2pt conversion. What happened next was the 2nd most inexplicable thing I saw all game: a Nick Saban team driving the length of the field to the Auburn 9 with 3:37 left and then
- Run for 1yd loss
- False start
- Run for 5yds
- Time-out, Bama
- Pass complete for 1yd
- Field goal doink (not too unusual - Saban apparently doesn't like to recruit kickers?)
Auburn gets the ball back, Bama with 2 time-outs and we get the #1 most inexplicable thing I saw all weekend: Nick Saban being just plain out coached by Gus Malzahn. The 4th down call with Bo Nix behind center and the punter out wide fooled Saban utterly and caught the Tide in an illegal substitution. 1st down Auburn, victory formation, and far more questions than answers about Bama.
Yes, I saw the weird and probably wrong field goal at the end of the half. If the game and your season really hinge on one blown ref call, you did something wrong.
#1 OSU v #11 UM - Shea Patterson lit up the OSU D in the first half, throwing for 250yds and basically creating a sinking feeling in the guts of Buckeyes everywhere. The Bucks certainly felt the absence of Shawn Wade, a potential 1st round corner, as the Wolverines picked on his backup Riep. Riep had the last word though, picking Patterson off in the 2nd half as the #1 defense in the land clamped down and and held UM to 55yds passing in the half and 91yds rushing overall. Just another blowout win for the Bucks. Oh yeah, UM was held to 2/13 on 3rd down.
Other games I enjoyed watching:
The Apple Cup - Both Washington teams have had mediocre seasons, but this game is always fun. Except this time. WASU was never in it; Peterson's Huskies calmly and methodically put the beat down on the Cougs.
Wisconsin vs Minnesota - for the first 2.5 qtrs, anyway. Who knew Wisconsin could pass? In the snow? Not PJ Fleck, that's for certain.
Sorry I missed it:
Virginia vs VaTech - 15 years since the Cavs beat the Hokies. I expect getting that monkey off their back was a relief. Let's see if they have anything for Clemson.
My predictions were 79.66% this weekend.
AP vs RR: Right now, the AP rankings average about 1 place higher than Ray's Rankings. That doesn't tell the whole story, though. The teams the AP ranks higher are ranked an average of 5 places higher, while the teams ranked lower are ranked an average of 3 spots lower. Even better are the AP conference rankings as a whole:
- SEC ranked an average of 3 places higher than RR
- B1G ranked an average of 3 places lower than RR
- Pac12 ranked an average of 2 places lower than RR
I'm not saying there's bias. OK, yes I am. That's the whole point behind Ray's Rankings, right? To expose and replace biased rankings.
Ray's Playoff Picture - I think my projected 16 team playoff looks pretty tasty. I moved a couple of teams (swapped 10&11 and 12&13) to keep from playing within your conference in the first round.
Anyway, let me know what you think of the new system with adjusted point differential added.
No comments:
Post a Comment