The most important thing at this point of the season is the playoff chase, so this week the model results will be down at the bottom and we’ll focus on how things stand with four games left to go. I’ll also improve a bit on my usual playoff projections by using nfl-forecast.com and ESPN’s playoff machine to be a little more certain about how things should shake out.
First off, let’s do the rankings and season predictions. Part of the reason to use the forecast and playoff machine is just to handle tie-breakers without me delving into pages of obscure rules, but also to pull teams apart a little by the most likely results. As you can see in the season predictions, pretty much everyone is expected to win two games from here on out. With only four games left, even a bad team would be ‘expected’ to win one and two isn’t outside the realm of possibility, and vice versa for a good team. Another way to put it is that with four games to play, teams are close enough in talent that ‘noise’ matters as much or more than talent. Thus it helps to put a hard number on things with a probability instead of saying ‘if team X wins twice and team Y wins one but that one isn’t this game….’ and so on.
Ok, let’s go by week and put the presumed winner for each game into the playoff machine to see what happens. Then we’ll wrap up with the probabilities and see if there are any interesting differences.
Week 14: After next week, the playoff picture has all the same teams as it does now, but the 3 and 4 seeds on each side have flipped: the Bengals take 3rd and push the Colts to 4th and the Eagles take 3rd and push the Lions to 4th. Lions fans get anxious because now they’re tied with the Bears and are only ahead on the tiebreaker.
Week 15: Things stay the same. The Lions are still in but sweating, there are no serious threats to the 49ers or Panthers to get the wild card spots in the NFC, and Baltimore is still hanging on to the 6 seed in the AFC at 7-7. Pittsburgh, Miami, San Diego, and Tennessee sit one game back and hope they can sneak in at 8-8.
Week 16: Lions fans sleep a little easier as Chicago falls to Philly but Detroit beats the Giants. The Bengals, who have run their record to 11-4, take over the 2 spot from the Patriots when they lose in Baltimore. Dallas is a game back of the Eagles and only Tennessee and San Diego can match Baltimore at 8-8.
Week 17: Dallas takes advantage of home field to beat the Eagles and take the 4th seed in the NFC; Detroit moves up to 3rd. Seattle and New Orleans get byes and Carolina takes the 5 seed over the 49ers. Despite losing, Baltimore holds on to the 6 spot over San Diego and Tennessee, all at 8-8. Denver and KC stay at 1 and 5, as they almost certainly have to, and Cincy gets the 2, New England the 3, and Indy the 4.
Again, that’s what happens if Luigi’s predicted favorite wins every time. But a lot of those are far less than done deals. So what are the odds of it playing out that way? I put all the actual probabilities into nfl-forecast (or as well as I could with the sliders) and had it predict the rest of the season a number of times. Then you count up how often something happens and bingo: probabilities for how often we expect a team to end up in a certain spot.
In the AFC, things are pretty much done in terms of who’s going. The Broncos, Colts, Patriots, and Chiefs are all over 99% to make the playoffs. The Bengals are at 96+%. Denver has a 70% chance of being the 1 seed and is otherwise most likely to be the 2 seed; they are unlikely to fall behind the Chiefs for the division (KC is the 5 seed about 90% of the time) or the other division winners. Despite the playoff machine prediction, the Pats end up the 2 seed more often than the Bengals, about 34% to 28%. They’ll be the 3 seed about equally often, and the Pats are less likely to be the 4 seed. The Colts end up in that 4 spot about half the time but have a 25% chance at the 3 seed, which means about a 20% chance of getting a bye. They are the only team in the AFC playoff hunt with no chance of being the 5 or 6 seed; if they lose the division, they’re out entirely.
So that leaves the 6 seed. While Baltimore might be the favorite in enough games according to Luigi, when you take the likelihood of winning into account it’s basically a toss-up between the Ravens and Miami at about 37%. The Steelers are next best at about 15%, and then you have a poo-poo platter of Tennessee, San Diego, the Jets, Oakland, and three teams with less than a 1% chance.
In the NFC, only three teams are fairly locked in: Seattle (apparently 100%), New Orleans, and Carolina (both over 97%). The Saints are a bit more likely to get the 2 seed and Carolina more likely to get the 5, but both have about a 10% chance to end up in the 6 spot. The Lions are next highest, at 82%, with roughly equal probability of being the 3 or 4 seed. Next up are the 49ers, with a 79% shot at the playoffs. If they get there, they will most likely be the 6 seed, about three times as likely as the 5 seed. The last slot is currently aimed at Dallas over Philly, 60% to 53%; Dallas has the edge thanks to currently having the tiebreaker and then having home field in their critical week 17 game. If Rodgers is healthy for week 15, it might be even closer than that. Chicago, Arizona, and Green Bay all have about a 10% chance at making it, although Arizona obviously only really has a chance at the wild card. As opposed to the AFC with its various hangers-on, there are four teams with 0 chance at the playoffs and the Rams and Giants apparently make it less than 1% of the time.
On to the model results! In all games Luigi was 7-8-1 (with the Eagles providing the push), but would have passed on four games (including the Eagles) and gone 6-6 on the rest. That gives Luigi a total of 88-83-5/71-64-3. Yoshi 1 was 8-7-1/7-7-1 for a total of 80-91-5/67-78-5, and Yoshi 2 was 6-9-1/5-7 for a total of 84-87-5/67-72-3. The consensus picks, where all three models agree, were 4-5 for a 49-46-1 total.
Luigi’s SuperContest picks were again 3-2, and would have been 4-1 if I could ignore the Packers pick and put Atlanta in instead. That puts my record at 37-26-2 for 38 points. The average score for people still picking is just under 31, and 38 points would have me in a tie for 41st. If I had actually been in the contest and done some Rodgers-related adjustments to my picks, my 40 points would have me in a tie for 12th and in the money. First place right now has 42 points. Yoshi 1 would have gone 4-1 or 5-0 with the Rodgers change for a record of 31-32-2. Yoshi 2 would have gone 2-3 (3-2) for a total of 35-29-1.
All three models had pretty bad weeks on the moneyline despite Jacksonville’s win; a few underdogs like Pittsburgh, KC, and Chicago were close but came up short. On the season overall, Luigi is pretty solidly in the black, Yoshi 1is a little bit in the red, and Yoshi 2 is solidly in the red.
Using his lines, Bill Simmons went 5-11. Luigi went 8-8, Yoshi 1 went 9-7, and Yoshi 2 went 7-9. That makes the counts Bill 70-87-5, Luigi 83-74-5, Yoshi 1 83-74-5, and Yoshi 2 77-80-5. Bill seems legitimately cursed this year.