Playoff Probabilities: Week 10

Welcome to this week's edition of playoff probabilities, now complete with bonus graphics! As always, these numbers come courtesy of Chris Cox at NFL-forecast.com and are generated with the help of the NFL-Forecast software app, which uses the win probabilities generated by the team efficiency model to simulate the NFL season 5,000 times. And if you don't buy the game probabilities from Advanced NFL Stats, you can tweak them as much as you like to generate your own playoff projections. I encourage everyone to download the app and test out your own scenarios.

How 'bout them Giants? In the most exciting game of the year thus far, Eli Manning drove the Giants to victory over the Patriots, giving New York a two game lead over their nearest opponent in the NFC East. At #9, the Giants sit relatively high in the rankings, yet they're projected to win the East in only 55% of simulations—less often than both the Chargers and the Ravens, neither of whom are even in sole possession of first place in their divisions.

A lot of this comes down to the degree of difficulty of the Giants' upcoming games. Over the weekend, I heard a lot of sports chatter about how tough their remaining schedule is, and the team efficiency rankings bear this out. The chart below lists the average GWP of each team's remaining opponents:

Remaining OppGWP

As you can see, the conventional wisdom is right on this one: the Giants have the toughest schedule ahead of them of any team by far, with their remaining opponents averaging .601 GWP. As a result, the model expects them to win an average of just four more games over the rest of the season.

High-Leverage Games of the Week

Pittsburgh at Cincinnati | Sunday, November 13 | 1:00 pm

Playoff Prob. CIN WinPIT Win
PIT 7896
CIN 6028

If the playoffs started today, Cincinnati would be the #1 seed in the AFC, but the model is not yet convinced of their viability as a playoff contender. The Bengals have yet to play a team ranked in the top ten in GWP, but beginning with this Sunday's game against the Steelers, they embark on a very challenging stretch of games which should give us a better sense of their true potential.

Regardless of whether you consider the Bengals to be a legit contender, competition for the AFC North title is projected to be intense. As expressed in this nifty chart, presented here in a form too small to be read by human eyes (click to embiggen), we can use Brian's game probability model to estimate the joint probability of Baltimore and Pittsburgh each finishing with a certain number of wins:

BAL-PIT Joint Win Probs Week 10

Baltimore's number of wins are along the horizontal axis, Pittsburgh's along the vertical, and the middle contains the joint probability of each combination. The marginal probabilities of either team finishing with a certain number of wins are listed in bold in the bottom row and rightmost column, and at the very bottom in large numbers is the total probability of one team finishing ahead of the other. (Blank cells have probability zero; cells that contain "0.0" have a non-zero probability that was rounded down.)

So what do we learn? For one, that Cincinnati is going to have to perform very well over the rest of the season if they want to win the North—the model projects that either the Ravens or the Steelers will finish the season with 11+ wins 92% of the time.

That said, while a win on Sunday won't make Cincinnati the favorite to win the division (it would actually help the Ravens' chances most of all), it does have significant upside potential for their Bengals, raising their probability of a wild card berth to 45% and their overall playoff probability to 60%.

New England at New York Jets | Sunday, November 13 | 8:20 pm

Playoff Prob. NYJ WinNE Win
NE 6691
NYJ 6632

After back-to-back losses, the Patriots find themselves in a three-team tie atop the AFC East. And like last week's game between the Jets and the Bills, the outcome of this week's matchup will have large implications for the division race. There is a crowded wild card field in the AFC, with each of the three leaders in both the East and North having chances of greater than 20% at a wild card berth. Given that, neither team will want to lose ground here.

A New York victory would move the Jets into first place and turn the division race into a true toss-up between themselves, the Patriots, and the Bills, with each team having about a one-in-three chance of winning the division. A Patriot win on the other hand would put the Patriots back on top of the East, giving them a 66% chance to win the division. The Jets meanwhile would fall behind, with their chances of a division title dropping to only 6%.

News & Notes
  • The model projects either the Texans or the Packers as the Super Bowl champion in over 48% of simulations. Place your bets now.
  • The Colts' chances of winning the Andrew Luck Sweepstakes are up to 82%. My chances of making it through the season without using the phrase "the Andrew Luck Sweepstakes" have dropped to 0%.
As always, the probabilities below may not add up to 100 (in percent form) due to rounding. Enjoy.


AFC EAST
Team1st2nd3rd4th
NE5629150
NYJ2338391
BUF2133451
MIA00298
AFC NORTH
Team1st2nd3rd4th
BAL612991
PIT3253141
CIN717697
CLE01792
AFC SOUTH
Team1st2nd3rd4th
HOU99100
TEN172270
JAC027703
IND00397
AFC WEST
Team1st2nd3rd4th
SD6522102
OAK21313117
KC10303228
DEN4162752
NFC EAST
Team1st2nd3rd4th
NYG5528143
DAL3135277
PHI13314313
WAS151776
NFC NORTH
Team1st2nd3rd4th
GB90910
DET965270
CHI127711
MIN00199
NFC SOUTH
Team1st2nd3rd4th
NO801820
ATL1862174
TB1144936
CAR073261
NFC WEST
Team1st2nd3rd4th
SF100000
SEA0383230
ARI0293635
STL0323235




AFC Percent Probability Playoff Seeding
Team1st2nd3rd4th5th6thTotal
HOU32362640099
BAL36168122992
PIT141350461290
NE101826261679
SD127560166
NYJ2614261948
BUF2611261542
CIN2320121939
OAK002180122
KC00090110
TEN0010157
DEN0004005
JAC0000001
CLE0000001
MIA0000000
IND0000000
NFC Percent Probability Playoff Seeding
Team1st2nd3rd4th5th6thTotal
GB78121091100
SF1447231500100
NO11632322486
DET4310531980
NYG31318203967
CHI0000233560
DAL06121231246
ATL028861236
PHI00391619
WAS0001012
TB0001012
CAR0000001
SEA0000000
ARI0000000
MIN0000000
STL0000000

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

12 Responses to “Playoff Probabilities: Week 10”

  1. James says:

    Considering how likely it is that Green Bay will get the #1 seed and San Fransisco the #2 seed in the NFC, that makes a significant impact on the other playoff teams' postseason success depending upon seeding pairings.

    Specifically, the #3 seed is vastly more desirable than the #4 seed, and the #5 more than the #6 (as the #6 is locked to play GB if they advance, whereas #5 has a chance to avoid GB). Obviously this is the way the seedings are intended to be, but I think this year it has the potential to be a much more significant difference than normal.

  2. Brian Burke says:

    Man, that heat graph really takes me back! Check out this one from the 2006 season.

    Fantastic post. I was just scribbling together a post about CIN and Josh just did all the hard work for me.

    An interesting thing about the PIT-BAL race is that they've both won a strenth-of-schedule game so far (BAL over NYJ, PIT over NE), but despite being in 1st place last year, PIT draws KC and BAL draws SD. I think SD is the much better team despite the fluky loss to KC.

  3. Anonymous says:

    Why average OppGWP instead of median? Average makes sense for normalizing stats that might scale more smoothly with opponent strength (YPA, SR%, etc.) but for # of games won, a game against a particularly strong team can lose you at most 1 game, regardless of how strong they are (and vice versa for particularly weak teams.)
    A quick spreadsheet to show the difference: https://docs.google.com/spreadsheet/ccc?key=0At2xGu_ne0lcdGZjZjBlN0pLNTkwakJLMHNGa3NFS0E

  4. Josh Katz says:

    James: That's a great point. I hadn't realized how often San Francisco makes it in as the #2 seed.

    Brian: Nice! I must have seen that graph at some point in the long, long ago and filed it away for inspiration to strike.

    Anon: As for mean vs. median (though to be honest, I didn't give it that much thought beforehand), while it's true that as you say, a game against a strong team can lose you at most 1 game, you are much more likely to lose that game, so it has a greater impact on a your expected wins.

    For example, if Team A plays teams with GWPs of .450, .500, and .900, their number of expected wins is going to be much lower than Team B playing teams with GWPs of .450, .500, and .550 (even though the medians are identical).

  5. Matt says:

    good point anonymous...

    I can see the merits of both sides. Median will basically tell you how many games are "tough opponents" vs "easy opponents".

    On the other hand the Mean gives you and average of how good or how bad each team is... obviously a GWP team of .9 is a much tougher matchup than one at 0.6, and your chances of winning that individual game are much less, but then looking at the average it skews the rest of the games.

    I think you probably need both pieces of information to get the real answer.

  6. Josh Katz says:

    Mistake! Sorry guys, I realized the OppGWP chart erroneously included the games from Week 9... Chart now updated to reflect the real values.

  7. Brian Burke says:

    OAK now jumps ahead of SD with their win last night. Total chance of making playoffs: OAK: 46%, SD: 39%.

  8. robbbbbb says:

    This is fun stuff; thanks.

    Can you please start putting some information in on draft position? Those of us who are fans of also-ran teams (I'm looking at you, Seahawks) would love to see some data on what the draft's likely to look like.

    Stuff on #1 pick probabilities is fun, but maybe a little deeper look (top 3? top 5?) would be informative and fun, too.

    Thanks much!

  9. Chris says:

    robbbbbb --

    You should run my software. It's easy and free.

    Here are odds fro the worst 5 teams and the first 5 picks:

    Indy: 82%, 11%, 4%, 2%, 1%
    Miami: 7%, 25%, 17%, 12%, 9%
    Arizona: 4%, 18%, 16%, 13%, 11%
    Seattle: 2%, 12%, 13%, 11% 10%
    St. Louis 1%, 9%, 10%, 10%. 9%

  10. bigmouth says:

    Tangentially related question that arose on Niners Nation. Some posters were quite insistent that it's bad practice in statistical analysis to round up probabilities to 100%. Someone even claimed it's in the first chapter of any Intro to Stats textbook.

    Yet I've never, ever heard this before. Anyone else?

  11. Josh Katz says:

    bigmouth: I've never heard that before either, and it's certainly not in the first chapter of my Intro to Stats book.

    That said, I suppose I could understand the argument if you're in a situation, as here, where a team could feasibly have a 100% probability and you want to differentiate between them and a team having a probability of less than 100% but greater than 99.5. Likewise for differentiating between a 0% probability and non-zero probabilities that have been rounded down.

    For the record, neither Green Bay nor San Francisco has clinched a berth yet, and the playoffs remain a mathematical possibility for every team (yes, even the Colts).

  12. Zed says:

    If you know, or if it's not too much trouble to run it, how does the Bears-Lions game look in terms of leverage?

Leave a Reply