By Kurt Bullard
Now that it’s late July, football fans everywhere are looking forward to the opening of training camp, preseason, and the not-so-distant start of the regular season. Even though the so-called football experts struggle to forecast the outcomes of the 16-game season, I’ll try to put together a prediction model for the NFL season using a more quantitative method than the likes of Trent Dilfer.
The biggest challenge obviously is to come up with a sound way to estimate team strength, an endeavor that’s demanding considering the amount of personnel turnover each offseason and the lack of advanced statistics to evaluate player interactions. The method that I came up with uses Pro Football Reference’s Approximate Value statistic, the site’s best measure of trying to tease out individual talent. Then, using ESPN’s NFL depth charts, I aggregated each team’s per game approximate value of what I considered to be the “core” makeup of an NFL team: QB, RB, 2 WR, TE, Top 2 OL, the Top-4 “Front Seven” defensive players, and the Top-2 players from the secondary.
There were some exceptions to simply using last year’s AV. If a team had an absent starter that was injured or suspended for the majority of last year (e.g. Adrian Peterson), I used the player’s 2013 AV value. And, if ESPN listed a rookie as a starter, I took the AV of the backup with the reasoning that, if the rookie ends up starting, he should perform at least as good as the person that is backing him up. So, I used the per-game AV of Josh McCown as a substitute for Jameis Winston in my model since predicting rookie performance is another battle of its own. This will inflate the odds for teams who plan to stick with a struggling rookie through thick and thin, and hurt teams who find a phenom rookie.
To make sure this was a sound method, I tested it out on last year’s data and ran a regression to see if AV was predictive of the end-of-regular season Elo ratings as reported by FiveThirtyEight. Aggregated AV was indeed significant with a T-stat of 8.57. It was also a strong predictor of Elo, as the regression returned a .72 R-Squared value.
This model does not account for aging, but I make the assumption that in aggregating these AV totals, the positive and negative effects of aging on an individual will, for a team, net out to around zero. So this model favors aging teams and may hinder up-and-coming teams.
I then converted the aggregated AV for each team into an Elo rating so that I could later use that value to calculate the win probability of each team in each game this season.
With the mean Elo rating set at 1500, I set the possible range of Elo values between 1320-1900, since the standard deviation of Elo ratings has traditionally been 90 points. So, the Raiders, who had the lowest AV aggregate (76.34) were set to 1320, while the Seahawks (166.19) were set to 1680. The rest of the teams were set on the scale based on the following formula: 1320 + (360/(166.19-76.34))*AV. Some familiar teams fall to the bottom, while the Super Bowl favorites Packers and Seahawks floated their way to the top.
But Elo ratings don’t paint the whole picture, as teams who finish with worse records the previous year tend to benefit from easier schedules. I therefore ran a Monte Carlo simulation of each team’s season, calculating win probabilities based on the Elo ratings using the following formula: 1/(10^(Opponent Elo – Elo)/400)+1). Using Benjamin Morris’ conversion table from wins to playoff odds, I then calculated the odds that a team would make the playoffs for the upcoming year. I then normalized it so an average of 6 teams would make the playoffs from each conference every year.
To very little surprise, the Seahawks have the best odds of making the playoffs after having made the Super Bowl the past two years. The Dolphins may finally dethrone the Patriots in the AFC East, boasting the third-highest probability of playing into January, fifteen percentage points above defending-champion New England.
The much-touted Bills revamping, on the other hand, may not have the desired impact. The team only has a 40% chance of making the playoffs – the lowest in the AFC East.
In a similar vein, a potential 49ers collapse this year may not be the sole product of Harbaugh leaving, as the out-flux of talent brings the 49ers to a one-in-seven shot of making the playoffs. The Raiders, Titans, and Jaguars seemed destined for mediocrity again.
We’ll see who beats the odds this autumn. I’d just prefer to know sooner rather than later.
Could you explain where you get such math formulas? I don’t quite follow the math. I find it surprising that you put Falcons at a 51% chance to go to the playoffs and the Ravens a 9% chance. The ravens went to playoffs 5 of the past 6 years. The falcons on the other hand had one of the worst records last year. Also isn’t it strange how the AFC East teams are ranked really high but the NFC East teams are ranked pretty low. In fact every team from AFC East has a higher chance of making the playoffs than every team from NFC East. If 1 team from each division qualifies for playoffs then clearly something is a bit off. Sure all the teams from AFC East could indeed be better than all the teams from NFC East but that doesn’t mean that those teams have a better chance of making the playoffs. (Remember when Seahawks made playoffs w/ 7-9 record?) You can still make the playoffs as long as you win the division no matter how good or bad your team is. Now in the end, this is all just data and none of this matters when it comes to game time. I’m just curious. If you could further explain or perhaps clarify any of my misunderstanding I would appreciate so.
You’re not accounting for wildcards. The best division ever delivers 3 teams to the playoffs. A terrible division will only deliver 1.
Yes, he is absolutely accounting for wildcards. If his dependent variable was “whether a team won their division,” then he wouldn’t have accounted for it. His dependent variable is “whether a team made the playoffs,” which absolutely encompasses the wild card. An r2 of .72 is absurdly high. I’m anxious to see how this pans out, but based on him retroactively fitting the model to the previous year, I feel pretty confident in his method.
Take a few statistics courses and you’ll get it. Nobody can create a perfect mathematical predictive model from scratch. All you can do is make a model and apply the data and see what the results are. If the model was trying to predict the actual teams that go to the playoffs, the author would have accounted for that. He’s not predicting that though, he’s attempting to calculate each team’s odds of making the playoffs using his model that had an r2 of .72 when fitted to previous data. Put very simply, he’s calculating odds, not predicting specific playoff teams.
Your armchair statistics make me bubble with rage. Everything you just said was garbage and all of the holes you tried to poke in this approach were terrible. This was a purely statistical argument with no adjustments made. The numbers are what they are. The method is what it is. Just leave it at that. If you want a prediction that is based on ‘gut’ go read any other prediction written by some so-called ‘expert’. THIS IS MATH
You cannot reduce a sport to “just math”! It just does not lend itself to that. Many have tried, and think that they can do it – see the “Moneyball” methods pioneered by Billy Beane. However, in the end, it still takes athletic ability and the drive to succeed to win in sports. If it could be reduced to “just math”, math gurus and geeks would dominate – I know, I was one growing up. I can tell you that it also takes some athletic ability to win – a skill set I was lacking in compared to others when I played. Brains and analysis and overcome many things, but the raw physical ability to throw a baseball 95-100 mph or to run 40 yards in 4.1 to 4.3 seconds or to throw a football 60-90 yards accurately is hard to overcome with mathematical analysis.
This Math is putting the people who do have the athletic ability to do everything in the right places this doesn’t mean they are putting nerds on the field
The difference here is that the author is not using statistics to build a sports team from scratch. Instead, he is calculating odds to determine what NFL teams have the best odds of winning their respective conference. In the author’s defense the Miami Dolphins can and may very well win the first 6 games of their regular season. While 6 wins may not guarantee you a spot in the playoffs it’s certainly a big step in the right direction. I see the dolphins ending with a 9-7 or a 10-6 record with a ticket punched for the playoffs and from there we’ll see what happens.
I like that you use quantifiable metrics on speed to argue its not about math.
However you posting program says that my answer is too short.
It may be math but it’s WRONG. The Jets still have Geno Smith. Cowboys still have the best O-Line. The Eagles still have a completely new team. The Panthers still have to beat out 3 of the worst teams in football from last season. Math means nothing when you’re making up a bunch of nonsense and applying it to the real world. Well 2+8=10 then we multiply that by 10 and I have a 100% chance of going and taking a shower right now. Well I better go do what the numbers are telling me to do.
One last thing 2x5x10 =100 & 2x5x0=0. This shows that 100% of my time was wasted and I got 0% of useful information from this article.
LMFAO!!!! How is your Geno Smith comment working now? hahahah
Also, Geno Smith is terrible! Hes a decent back up at best lol
All I know is my Panthers will be in the mix.
you sound a lot like an eagles fan
I was going to reply to this but its just to stupid. Carolina at 19 percent! Yeah right top 7 perhaps but not that low. CHCHCHCHCH Not buying your formula peeps
Too stupid my bad. CHCHCHCH
Yeah I know. I guess they won the division so their schedule is slightly harder but they easily play in the worst division in the entire NFL. Mediocre teams is god awful divisions usually win the division. Life’s short enough, I really don’t need to be wasting it reading this crap.
Carolina will be in the mix they can say what they want Attrition was our biggest opponent last yr. Plain and simple. Go Panthers
You stupid idiot, Math is way smarter than any of us.
He says in his formula that he just has it for the top 12 make it and doesn’t count for divisions/ wildcards
This is just predicting who makes the playoffs. So there is a good chance atleast 2 afc east teams make the playoffs. And theres a good chance a 3rd or maybe all 4 teams are in the playoff picture week 17. All this means is Buffalo has a better chance of making the playoffs than any one Nfc East team. Although 1 will make it there is no clear favorite.
I do also agree that most of those numbers fall straight off the sky. But, as usual, the field will prove maths wrong. This study sees the Dolphins battling with the Chiefs for the AFC superbowl spot while it has been historically proven that in football, the player who recently go the biggest paycheck before the beginning of the season tend to tank their stats (the Redskins have seasons upon seasons of living proofs, and that’d take out Miami’s Suh and the Chiefs’ Houston out, which their respective teams’ best players) and when it comes to the Jets, the nucleus of the team is yet another a big paycheck seeker position that just got paid (corners). As of the Giants, well, the best defender is missing a finger while trying to reach big money and the rest team has been in disarray for quite a few years now. The Lions have their #1 offensive weapon recovering from injury and their #1 defensive weapon counting his stacks of Benjamins in South Beach. All these are just a few examples. Basing the ELO from the previous year definitely does no justice for a few teams (My Chargers dealt with the Pats, Seahawks and the Broncos twice this past year for example) while I would think that basing in through even five years would be more accurate, but unfortunately for Jets, Texans and Chiefs fans, that would bring their teams down by a bit. And realistically, who would say that Miami has 15% more chance than the Patriots to make it to the big game, even without Brady at the helm? Anyway, I like the attempt, and it sure gave me something interesting to read!
This logic is a joke, lol. The Ravens replaced everything they lost, and improved some positions, yet they are behind the 49ers who have had a coaching change, a handful of stars retire, and have an over rated quarterback, lol.
I don’t know if ATL truly has a greater chance, but I’m sure it figures that ATL is in a really weak division while BAL is in an ultra-competitive division.
I totally agree with this guy!!
falcons 2-0 now and ravens 0-2
I am a Dolphins fan so I hope you’re correct but really have my doubts. How often do we see big free agent spenders win big in the fall? Really, never. And while our QB has shown progress, Ryan Tannehill has a lot to prove-the deep passing game is not his strength nor are our offensive guards. Also remember, this team has NO playoff pedigree to fall back on if they do get in.
My bigger question though is how do the Baltimore Ravens grade out so low? What do you dislike about them? Their best OL in football, super bowl winning QB, their ability to beat teams like the Steelers and Patriots? Their defense may not be 2000 caliber but it should still be decent. They’ll be around in January. I have a hard time with that one.
He doesn’t dislike anyone. It’s a formula. And I can see why the ravens are so low. They have a mediocre core on offense as he explained and a week defense too. He took the values of Joe Flacco, Steve Smith, Breshad perriman, Justin Forsett, marshall yanda, eugene monroe. on offense. That is a pretty terrible offense is it not? And the defense stars include dumervil and suggs. the rest of the defense was pretty below average. I can totally see why the Ravens suck.
Lol you are as clueless as the writer. The Ravens are in the playoffs every year and making runs in the playoffs too. Yeah, they’ve never really had superstar WR in Baltimore, and the defense isn’t the big bad bully it used to be….truthfully, besides the Seahawks, no other team in the NFL is even playing anything close to shut out defense anymore.
This is what the Ravens are; they are a very well coached, confident, and resilient team. They are always in the playoffs and are always a difficult out. They have the wins to back it up.
Whatever credibility this garbage might have had went up in flames when the Ravens, a perennial playoff team was raked at the bottom with the likes of the Jags and Bucs. Even the Browns, Redskins etc are ranked higher than the Ravens. Lmaooo what a freaking joke! Obviously composed by an idle Ravens hater.
Ty Mike,couldn’t have said it better myself
Always in the playoffs? 2002, 2004, 2005, 2007, 2013 ring a bell? Stupid ratbird fans…
Ravens fans are idiots… Just like most of their players and coaches
Be nice, denial is a fantastic weapon.
Yup. 6 of the last 7 years: 2008, 2009, 2010, 2011, 2012, 2014. Sounds like a typical knuckle-dragging, iron city swill drinking Yinzer post.
Thank you mike for saying what i was thinking lol some people really hate the ravens probably cause they destroy dreams in the playoffs
Flacco is alright during the regular season but elite during the post season. Yanda is graded as the BEST guard in the NFL isn’t he? Lets not forget Forsett…probowl season last year. Lets get one thing straight….the only reason we didn’t play in the superbowl is because we had 6 CB’s in the IR and gave up two 14 point leads. I think perhaps the ravens aren’t getting any love because we do have some rookies or unproven players stepping into key positions (WR2 + TE for example) and he did mention the formula penalizes up and coming teams.
I totally disagree with everything that you just said. You clearly have no idea what you are talking about. Joe Flacco was just shy of 4,000 yards passing last year and lead the ravens to yet another playoff victory to add to his career post season success. Steve smith was a top 15 reciever last year, and perriman has the looks of a top reciever with great size and 4.25 speed. Justin Forsett led all running backs in carries of over 20 yards and rushed for well over 1000 yards which earned him a pro bowl visit last year. As far as offensive line if you take the two best lineman, you would see that the ravens have the first ranked offensive lineman in football in Yanda, and the third ranked offensive guard in osemele. Doesn’t sound too terrible does it. As for the defense, how could you fail to mention all pro cj Mosley and his sidekick Darryl smith who were both top five in the league in tackles last year. The corner situation has been fixed as gm ozzie newsome has added mccourty from the patriots and Ladarius Webb and jimmy smith are now healthy again. The front line should be stacked this year with Brandon Williams and Timmy Jernigan leading the way. I don’t think the ravens suck in the slightest Jameson.
Lol… Pretty terrible D? How about CJ Mosely, Jimmy Smith, Daryl Smith etc…
The Ravens were extremely close to beating the SB champion Pats in the playoffs last season, and the team has only gotten younger (more athletic) and better. The pass defense should be MUCH improved because of the return of Jimmy Smith and addition of Kendrick Lewis at safety.
The problem with using math at this point in the (off)season, is that we have a whole draft full of rookies that can’t be adjusted for in the statistics. They are statistical wild cards. This means that teams that typically draft well and scout well by adding valuable veterans who may have been in bad situations are discounted far more than others.
Math is very powerful but the unpredictable human variable in sports (bad calls, injuries, poor years, bad plays, choking at the wrong time, etc) makes it almost impossible for these types of analyses to hold much weight.
I love statistics. but i would like to see more explanation on how the Ravens are so low. In many years, the Ravens have been undervalued and have gotten to the playoffs. I am very excited for this season, and I think the Ravens are a better team than last year. But don’t worry Raven’s fans, we can get in with a 95% chance or a 9% chance.
I think the big factor that’s being neglected is schedule. Miami has an easy schedule where the tougher teams are backloaded. It’s realistic to think that Miami might end up going 8-2 going into Week 12, then back their way into the playoffs. Remember, this is based on odds of making the playoffs, not doing well in them.
Dude. Do you even football? Lol seriously though this isn’t gonna prove even close to accurate.
How does nobody understand statistics. This study only ever claimed to calculate odds, it never claimed to predict who makes the playoffs. If it did, the author would have some teams with 100% probabilities. According to this model, it’s possible for the Raiders to make the playoffs and the Seahawks to miss the playoffs, it’s just highly unlikely.
You may, or may not, understand statistics but you certainly don’t understand English. I take it that English is a second or third language for you. Let me help. Calculating odds, as you put it, IS making a prediction about an event along with the probability of the prediction coming true. Any beginner’s book on Stats will tell you that.
I’m not sure you understand his colloquialism Mr. Stats dude. When someone says “do you even football?”, sure they could be an idiot, but they also could be JOKING?
Yeah, I think we’re gonna beat this guy up over just the Ravens prediction alone here. 🙂
Who’s overrated on this list?: Dolphins, Chiefs, Jets, Falcons
Who’s underrated on this list?: RAVENS, Cowboys, Eagles, Cardinals, Vikings (I think)
But it’s all for fun, we shall see won’t we? 🙂
I’m a Steelers fan and even I think the Ravens are being underrated by this. The Ravens are one of the toughest teams in the league! Sure they’ve lost some talent but they’re well coached and whoever steps up tends to perform well. As long as they’ve got Harbaugh and Flacco they’ll always be challenging for the division and there’s not a chance in hell that they finish below the Browns. Sure I want to see my Steelers beat them but I still respect who they are as a team. The ravens are a top 10 caliber team, have been for a long time.
Wow Sam i’m impressed never thought i would see such praise from a steelers fan,, but i agree with you on the steelers,, stomach is in knots the hole day when they play ea other.I know a lot of our playoff success depends on if we see you guys or not.
I’m high on both the Chiefs and the Falcons. As a real long-shot I probably would bet the Jags at an implicit 33-1. Really that division is wide-open if one player has a sub-par year (Luck).
The Giants being ahead of the Eagles and Cowboys is just bizarre.
With Tony Romo at the helm the Cowboys are NEVER underrated
Interesting stuff, and I agree with the thread so far that the Ravens are way too low. I would also bet that the RAMS have better odds with their defense and now Nick Foles and Todd Gurley in the mix. Also the Vikings – I think you are not calculating a key factor in that they are now in year 2 of an entirely different offensive and defensive scheme, they also sustained a lot of injuries last year and have improved their roster on both offense and defense. Too many factors not included – mainly coaches, coaching staff, schemes and systems that the players will have to adapt to or be more familiar with in 2015.
The Eagles and Cowboys are under 40% chance to make playoffs? Eagles last year had one of the highest scoring offense’s and best special teams and one of the beat take away defenses. The Eagles now have a better defense and a backfield filled with RBs and a quick offense in general. The past two seasons the Eagles have a total of 20 wins and 12 losses. 10 wins for each season.
The Cowboys have one of the best offensive lines and a high caliber offense. The Cowboys went 12-4 last year and could have been in the SuperBowl easily.
The Chiefs are put at a 66% chance to make the playoffs but yet their offense is very mediocre and is pretty much a one-trick pony because they are a running team.
The Ravens are at 5% chance. They have been to the playoffs 5 out of their last 6 seasons. Out of their last six seasons they have an 83% playoff rate. Now they are at 9%???
The football season will prove these percentages wrong.
The ravens have made the Playoffs 6/7 not 5/6. Anyways you’re still on point.
Ya thanks for backing me up haha. I can’t believe these statistics.
“teams who finish with worse records the previous year tend to benefit from easier schedules”
Not true. Only two games each each year have anything to do with a team’s previous season record – and those two games are against teams that finished in the same position in another division. Under the current system, opponents for 14 of the 16 games in a season can be predicted years in advanc e.
What you said was once true, but it hasn’t been true since 2002, when the current scheduling system was adopted.
That’s not entirely true either. Previous seasons record still plays affects which teams in each division you play at home/on the road.
I acknowledge the fact that this is pure statistics, but the beauty of football is that statistics are not quite as accurate as they seem. As a ravens fan I am obliged to have a say about them. You say you used PFF’s method, well as a matter of fact ESPN released an article ACCORDING to PFF how many players each team needed to reach the Super Bowl. The ravens were the second team that needed the LEAST players, with only two, and Dallas being the first with NO players needed, this article also has them very low. I am aware of the fact that the Ravens lost Haloti Ngata, Pernell Mcphee and Torrey Smith. And of course Dallas lost DemarcoMurray. Specifically the Ravens drafted their replacements mostly. You have no way of proving how rookies will play. So of course you did something that seems very mediocre in the likes of an article like this one (about stats), which is that you put veterans ahead of the rookies. In the case of my Ravens, do you expect Marlon Brown to perform as well as our first round pick Perriman? The Ravens have been–besides the Patriots–the most consistent team in the past 7 seasons (since the Joe Flacco era began). Out of seven seasons they made the playoffs 6 times, that is roughly 86% of the time. As a loyal fan of my team, you saying that we have a 9% chance of making the playoffs–with inaccurate stats–is very disrespectful. So If you know nothing about the rookies–which virtually you can’t–then you can’t write an article like this. That’s the beauty of football, the MYSTERY. I find this article to be rubbish and I feel sad about all the time you wasted making it.
They basically stole this article at nfl.com http://www.nfl.com/news/story/0ap3000000503110/article/harvard-students-model-picks-dolphins-seahawks
I hope this was done with the authors approval.
you do realize that link links back to this page right? this is the source
Looks like this method will always over value those teams with a crop of elite players but with a poor supporting cast, and undervalue teams that may not have so many superstars but have a quality cast of backups and defensive players across the field outside of those accounted for.
thats why miami have gone in ahead of the pats, their top 12 players are great, but the rest of the roster (not really accounted for here) is a little way behind whereas i’d put the pats as having a better 53 man roster all things considered. Maybe miami’s top 12 players are better than NE’s top 12 players but that doesnt win you all the games.
Really interesting work, and it doesn’t look too far off so I think you are really onto something. But I think this needs some refinement.
My two cents: You need to account for every position on the field. You can’t just ignore things like over half the defense and 60% of the offensive line. On many teams, the #3 WR is much more important than the TE. And an outlier at K will drastically change the fortunes for a team.
The key is to give the proper weighting to each player on the roster, and that probably should include “coach”. For example, a couple of the places where your odds diverge from the betting markets is that you have Miami and KC with higher odds in the AFC, and Denver and NE with slightly lower odds. I suspect this is because your formula isn’t valuing the QB position enough. In the modern NFL, the QB position is probably as important as all the other offensive positions combined. Also, NE consistently outperforms its “talent on paper” because Belichik is such a good coach.
While I enjoy this as a quantitative exercise, it’s a problematic analysis on both football and statistics levels.
First, the study equates (not associates) team success with a pool of player talent. We can intuit that “dream team” organizations would rate very highly in this rubric, but that’s questionable in the real world. For example, the deeply flawed Dallas teams of the last 10 years featured some top-end talent, but no depth. While this obviously has a very profound influence between the lines towards the end of the season, it doesn’t appear to be accounted for in the math. As mentioned by other posters, variables like cohesion, coaching, and organizational culture are ignored. These are highly relevant–just ask Dan Snyder. To be bluntly logical, the structure of the study represents the fallacy of composition–the quality of the parts determines the quality of the whole. Composition is always involved in a generalized analysis, but we have to keep it in the foreground.
As a researcher, I think in terms of questions, and there seems to be some ambiguity here. The question the model is *intended* to address is “what is each team’s likelihood of reaching the playoffs?”, but as I read it, the question *actually* being answered is different: “which team has the best collection of core talent according to PFR’s AV metric?” This is important, because the independent variable here is not talent–it is someone else’s assessment of talent. Without validation, we’re really testing the AV metric, not the teams.
I was pleased to see that Kurt validated the model against last season, but I submit that a good-looking regression can be deceptive. What does that high R^2 actually mean? It means that teams with a better talent pool were more likely to reach the playoffs. That’s true, but trivially so–it circles back to the independence problem mentioned above. Shouldn’t the question be “to what degree does the AV pool associate with success?” Don’t we have to answer that question before we can validly progress to the predictive question?
I don’t want to knock the project too much, because I think it is an interesting start. The AV variable looks like a good way to develop a talent index, which could then be used in a multiple regression or PCA/factor analysis that would provide a more robust prediction.
Anyway, interesting read–keep statsing!
This is an enjoyable statistical exercise, but has some serious flaws. The ranking of the Ravens, a perennial contender, highlight the limitations. One, no weight is given for coaching prowess — it assumes that Joe Philbin and John Harbaugh are coaching equals. That’s nuts. Second, no value is assigned for special teams and the kicking game, a traditional Ravens’ strength. [The loss of Jacoby Jones, however, could hurt the Ravens more than most Ravens’ fans want to admit.] Third, no value is given to rookie or “cheap” veteran replacements and Ozzie Newsome has had quite a bit of success with both. Steve Smith and Justin Forsett are examples of “cheap” veteran pick-ups from last year.
Bottom line – don’t worry Ravens’ fans, they don’t really have a worse chance of making the playoffs than the Browns.
I’ve been a loyal and faithful Dolphins fan since 1970. If I get cut, I bleed aquamarine. Statistical analysis of a given team’s player’s abilities is a nice staring point. However, I want to address the comment(s) made by Gus Abbott. COACHING is a huge part of a team’s win/loss record. It plays an even BIGGER part when a team makes the playoffs. Joe Philbin is a LOSER. I gave him PLENTY of chances to impress me with his coaching abilities, and MORE IMPORTANT, his Coaching DECISIONS. Here are Ryan Tannehill’s SITUATIONAL Stats: http://www.nfl.com/player/ryantannehill/2532956/situationalstats There is a SUBSTANTIAL difference between Tannehill’s QB rating between his opponent’s 49 yard line to their 20 yard line AND elsewhere on the field: 77.8 Rating versus 81.4, 94.8, & 101.8. He’s had 13 sacks & 4 interceptions between the opponent 49 and 20 yard lines. This is all due to Philbin’s insistence Tannehill remains in the pocket. ALL THE TIME. Well….this will be Tannehill’s 4th season at QB. He came out of college with enough mobility, one could call him a semi-scrambler. Now I DON’T advocate Tannehill become a running QB. But a GOOD/GREAT Coach will utilize a given player’s natural abilities. When the Dolphins have to overcome LONG YARDAGE for ball control are the times when Tannehill gets sacked or gets picked off. I’ve SEEN Tannehill on set designed runs/scrambles. That’s a great way to slow up the pass rush, give confidence to a young QB, and keep the defense honest AND guessing. He’s EXCELLED in this role. The very few times I’ve seen this, I’ve also seen Tannehill SHRED the pass defense LATER on other drives. Such is the EFFECT of his mobility. I don’t understand why Philbin REFUSES to incorporate such plays on a regular basis. His STUBBORNNESS makes him an INEPT HACK. I’m excited about the Dolphin’s acquisition of Ndamakong Suh and the other DTs they drafted. I’m a firm believer that the Dolphins’ Defense should and will be better what with their line and secondary. But in important games, the defense will get tired out way too quickly. Most of that will be because Tannehill will be forced into high risk situations, get repeatedly sacked or intercepted. That will bring the ‘Fins defense back out on the field. Meanwhile, Philbin will be scratching his head, CLUELESS as how to rectify Tannehill’s errors.
In this game, QBR means more than SOS. If I factor in 2014 QBR (* mostly, – some fudging like using Sam Bradford ’13 season numbers; AND assuming NO regression (eg Big Ben) NOR progression (eg Bortles), in a really simple model (ELO*.10 + 1.5*QBR)): I get: GB, DAL, DEN, NE, PIT,SEA, NYG, NO, ATL, IND, SD, MIA, KC, BAL, DET, AZ, HOU, CIN, CAR, CHI, SF, PHI, MIN, WAS, CLE, TEN, NYJ, BUF, OAK, TB, JAX. (That said, I don’t think NYG is really better than PHI, or DEN/NE better than IND)
The Cardinals must have been doing it with smoke and mirrors, because I think they have the best roster they’ve had since moving to Arizona this year, yet only show a 29% chance. I guess we’ll see how predictive this formula is. Expect Mathieu, Peterson, Cooper, Weathers poon to be must improved after recovering from injuries. Oline and running backs will be much improved, which will help the passing game and play action, especially with a healthy Palmer. I expect Brown will build on his rookie season and Floyd and Arrington to bounce back. Defense lost Cromartie, but should be improved overall.
The Vikings are way too low on this list. And I don’t understand how the Browns got to 20%. A little silly to me.
Whoever made this knew there’s no point and it wouldn’t make sense, whether you’re critical about certain teams or not if you really look at how teams performed last year and their off season moves you can make a more sensible list of probability for each team, although anything can happen but before it’s even the preseason you know it’s pointless
Really appreciate this article.
First of all, when an article’s title is “A Way Too Early Prediction of the NFL Season” you need to take it all with a grain of salt. I think it is assumed that someone of the author’s intelligence understands that this game isn’t simply a mathematical model. Don’t get pissed if his model doesn’t support what you personally believe.
Let’s just touch on a few teams/divisons that have been mentioned in comments:
Ravens: They lost Ngata, McPhee, Smith and they gained who exactly? Rookies – and I do believe some very talented rookies – but he clearly states how they evaluate rookies and states this could be incorrect. In my opinion the Ravens are in a rebuilding year. Queue the angry Raven fan comments now (or re-queue them…). Totally understand the lowered % chance and totally expect them to miss the playoffs, probably just for this year. 2016 I expect them back in the 50% range.
Falcons: Had some key injuries (OL was a mess) yet were top 10 in yards and passing yards last year. Plus their divison is weak and his formulas account for their opponent’s strength.
AFC East: This divison is, in fact, very good. The Dolphins were good BEFORE Suh, and the Jets are a QB away from being a force. Have you seen their defensive depth chart? The Bills, lest you forget, were dominant at times last year (beat the Packers with Kyle Orton as their QB). Oh, and the Patriots won the Super Bowl.
NFC East: Cowboys lost Murray. Cowboys downplay losing Murray. Cowboys will indeed struggle to replace Murray. That team will be SO different without him and if you disagree, you’re lying to yourself. Redskins are jekyll and hyde, but mostly hyde. The Giants have been the most injured team the last couple years and this has resulted in 2 poor seasons. The Eagles have potential, but their QB situation is surely dragging their % down.
Thanks for the article and the formulas, Kurt.
The fact of the matter is that you CAN’T replace incoming rookie starters with veterans. We don’t know if the rookies are going to be good or bad. It makes no sense to write this kind of article when you don’t know all of the stats. It was a huge waste of time.
For what it’s worth, I handicap football every year. I bet last year at a 60% success rate, which on an individual basis is incredibly efficient over such a large sample size (three best on Sunday, One on Thursday, One on Monday – every week).
I bet this successfully not by statistical analysis, but by film study. To the guy who says the Ravens are replacing these NAMES with ROOKIES, it is obvious that he doesn’t realize that the rookies have names as well, and they also are performers. Carl Davis was a borderline first rounder as an athletic nose tackle, and he will be second on the depth chart to BRANDON WILLIAMS who graded out very favorably on PFF if you want to look at a statistical bend, but he was very powerful to the EYE on tape as well.
Timmy Jernigan will also help at DE as well as CJ Mosley at LB with Daryl Smith (check out HIS PFF score!)
Perriman will be an immense upgrade from Smith, and that’s at one of the easiest positions to eval coming out of college, WR.
In the AFC East, I have no qualms with your eval, the division is literally a toss up depending on how the Bradyless Pats operate, now THERE is a “wild card”, the QB position is very hard to evaluate, as bad poise can render good talent useless, but poise isn’t helping someone who can’t make the reads.
The Cowboys writeup you have is the biggest reach of all, they just added Randy Gregory, Byron Jones, and La’el Collins in the draft. And that’s to a team with very few weaknesses on their roster already. Tony Romo was the highest rated passer in the league last year, and many people are trying to attribute it to the running game, but that’s not the fact the tape reveals. DeMarco Murray, as of week 9 last year ran for MORE YARDS BEFORE CONTACT THAN ANY OTHER RUNNING BACK HAD TOTAL YARDS. This means that DeMarco Murray ran for more yards before being touched by a defensive player in week 9 than any other running back had actually run in the entire league.
This is a credit to Dallas’ FANTASTIC run blocking O Line, and the film backs it up.
Now, some backwards thinkers then instantly look at YAC for Murray and see that those numbers were also very high, but isn’t it obvious when a back is being tackled by a LB or even a little DB that he is going to break more tackles? Look for a huge regression for DeMarco in Philly, as he was best BETWEEN the tackles in Dallas, and Philly just let Evan Mathis go.
Good luck to all this year, but if you are planning on betting using this system please think again and just buy a subscription to NFL game rewind and watch EACH player, that means multiple viewings, and you will be successful.
Although I want to agree with this formula, it would be more supportive of almost all the top ten teams weren’t in the same conference, how can a team have good odds of there are 8 teams ahead of them with greater odds, not saying the chart should have exactly six teams from each division but it should be closer to prove a more accurate formula
Meant to say “if” a bunch of times and I also meant conference not division
The chart should have exactly 6 teams expected to make the playoffs from each conference. Anything else is by definition impossible under the current rules.
This is fantastic. The people replying negatively to this thread I think are just looking at the charts and not looking at the formula behind it. Please continue to post stuff like this. I’ve grown tired of the NFL reporters blind speculation.
The guy generated flawed odds by normalizing them. He was lazy and not only didn’t account for conferences, but didn’t account for divisions, and wildcards. The numbers are completely worthless. Probabilities that ignore structural restrictions aren’t even really probabilities at all. At least not probabilities with any substance to them.
Absurd. The Ravens are clearly 1 of the 6 most likely teams in the NFL to make the Super Bowl. That is a true statement nearly any season, particularly this upcoming season. These predictions have them 5th from the bottom, beneath the Browns, Redskins…..absurd, beyond absurd.
This model fails to account for the fact that only 6 teams make the playoffs from each conference. If each 100% adds up to 1 playoff team, you have 6.3 playoff teams from the AFC, and only 5.7 from the NFC.
The final table does nothing to take Divisions or Conferences into account. The table suggests that based on probability the AFC will somehow have 8 of the 12 playoff teams
Not quite, but they didn’t factor conference into account. If you factor in all the probabilities, you’d have 6.4 AFC teams and 5.6 NFC teams.
I really don’t think that they considered divisions. Because how can you have 3 teams in the same division ex. (Dolphins, Patriots, and jets) with over a 50% to make the playoffs?
Unfortunately, this is nothing more than meaningless entertainment.
The algorithm cannot measure matchups, heart, coaching blunders and injuries.
In the NFL, poor matchups get exploited, great players play above the talent level, and in the end, the healthiest teams have the best chance to survive 16 weeks of regular season and four weeks of playoffs.
So the ravens have 1-91 odds. Where do I place my bet?
So Carolin is below the Falcons and Saints after beating both with a depleted roster…hmmm..if you say so
You know 6 teams make the playoffs from each conference, right? There are an expected 5.6 teams from the NFC and 6.4 teams from the AFC projected to make the playoffs. You know that can’t happen right? Did they stop teaching statistics at Harvard?
So is it safe to say that the Raiders Jags Titans and Tampa wont win to many games this season and during the season will you do weekly predictions?
Mike, you don’t “really think that Den/NE are better than IND”?
I guess 45-7 meant nothing?
Oh, and for good measure, the Colts are the team who turned in the lighter footballs to the league. You think the Pats aren’t coming out breathing fire vs that defenseless Indy team @ Indy when they meet? To me, the Colts have done nothing to convince me they can stay on the field with New England. Why am I wrong?
This is flawed as it fails to take into account coaching/staff changes and scheme changes. You need to factor in scheme…sometimes even at the pro level effective schemes and adaptability of coaching staff to adjust gameplans plays a huge factor in outcomes
How is this called a “model” and “simulation”…? All it’s doing is crunching team schedules with relative strength variables (that the author had no part in developing). It doesn’t simulate actual gameplay, it’s just a fancy ranking system. The sophistication is the equivalent of adding up all the bullets one army has and all the bullets another army has and says the one with the most wins. It assumes that the bullets are the best way to determine the chances of winning, and all that is required to win is to have more bullets, regardless of how they’re used. So a top defensive line against an otherwise stronger offense but with the OL as its weakest point can easily turn the tide of a game, which this in no way will account for. Even if its statistically significant (of course it is, the people that developed AV and Elo wouldn’t have published it if it weren’t) is still only giving you a correct ranking 72% of the time… so 9 of 32 places are likely to be incorrectly predicted. I’m not saying that the work is wrong, I’m sure the author is aware of its shortcomings; the game is infinitely more complex. My issue is with making this out to be more than it is… just some numerical computation that takes others’ original work (AV and Elo) and applies it to an existing scenario. Yawn.
I think you’re being unnecessarily harsh. This is an undergraduate project, and a fairly sophisticated one for anyone outside of a rigorous mathematics department. His two-step methodology is very clear; both ‘model’ and ‘simulation’ are appropriately used. I agree that there are limitations (as I pointed out in my earlier post), but any model is an abstraction. I’m confident that you don’t invent your own inferential tests, but use tools developed by others (T-tests, ANOVA, chi-square, etc.). So it’s senseless to call him on the carpet for using existing analytical tools like Elo or AV.
In the end, analysts analyze. That’s what the author is doing, and he deserves credit for creating an analysis and putting it out for public consumption/comment.
Well Said, Eric.
It’s really sad that people can’t just appreciate this for what it is. Just a tool at analyzing distinct data sets.
Psh Harvard kids know football as well as they know their respective opposite sex… ravens only 9% chance… right
Dolphins in Super Bowl 50…have you guys ever been right before? And isn’t Harvard in New England? I figure Tom Brady is motivated enough without you guys telling him Miami will win “his” division…last time Fins did that was 2008 when Brady got hurt…hmmmmmm…and like everyone else you have Ravens (a team I do not like at all) way too low. So this means you have Miami winning AFC East w/ Pats wild card, KC winning AFC West w/Broncos wild card, Pittsburgh winning AFC North and Indy winning AFC South. I think Dolphins get a wildcard but not division title. I see KC or San Diego winning division with other getting WC (or more likely, Ravens). In the NFC Giants win NFC East, Packers win NFC North w/ Detroit WC, Atlanta winning NFCSouth w/ Saints getting WC, and Seattle winning NFCWest. (I think Vikings and either Saints or Cardinals getting WC). Thing is most of the teams you guys have making playoffs so do I. But I’d love to know what your record for predictions for playoffs is.
Oh, glad you mentioned that. You have the Ravens probability wayyyy too low. Look at their team. Look at how they’ve performed in their division. They’ll either win the division or get in on a wild card.
Interesting article. I think you have some weighing issues as the Cardinals and Cowboys are rated very poorly, with teams that have large question marks currently marked much higher than them. An example is the rankings of Chicago, the Saints, and the Giants. How can you rank two playoff teams with loaded rosters as having less of a probability of making the playoffs than 3 of the worst teams in the NFL the previous year? The roster turnover is not enough to warrant that kind of single season swing for two playoff teams and then also three non-playoff teams. I think you’ll need to work on your predictions a little bit.
It would be very interesting to apply the models to data available at the end of 2013 to yield its prediction of 2014 odds. Can you please do this?
I was going to request the same thing to validate how close the model is…. use 2012 data and project 2013 and so on…
Three rounds should do the trick…
I looked through your numbers, and I get that they are the odds for reaching the playoffs, not for making it (or winning) the Super Bowl as some idiot journalists are saying.
However, when I add up your numbers, I get 1200% – which makes sense, given that 6 teams from each conference make it to the playoffs (4 divisional winners plus 2 wildcards).
Unfortunately, you normalized the numbers across all 32 teams, not within each conference. The NFC has a total of 560, while the AFC has a total of 640.
Please recalculate and repost your new numbers. (It’s not a simple “increase the NFC odds by 600/560 and decrease the AFC by a factor of 600/640”, because then the Seahawks have a 101.7% chance of making the playoffs.)
This proves something that I’ve known for a long, long time: Harvard graduates and professors are not nearly as smart and intelligent as they are reputed to be. To pick Miami and Kansas City as having better chances to make the Super Bowl over Baltimore is absurd. These teams do not have the talent level, experience, or coaching staffs to make it that far. I can understand New England and Denver being rated highly to make it, but, the Ravens should be in the mix as well. No wonder our economy is in trouble – with stupid Harvard grads and advisors like the person who came up with this study running the economy, why are we surprised?
Stacey, I think you are not as intelligent as you think you are. If you actually knew how to read, you would see that this article is not at all attempting to calculate the chances of making the Super Bowl. Rather it is calculating the playoff odds for each team based on a model that the author clearly acknowledges has its faults and is just an approximation, hence the title as well. You and I and Kurt all know that the Ravens are generally superior to Miami and Kansas City and are probably more likely to make the playoffs than those teams, but to say that this author and other Harvard-affiliated people are all stupid is completely ignorant and perhaps you should actually read the article, not just skim before making such comments.
I did read the article, and I still feel the methodology is flawed and the exercise in and of itself worthless. I thought Harvard grads and advisors were smarter than this. This exercise does not take into account many aspects regarding the past performance of players in relationship to how they may perform in the future, which, in reality, is one of the only true benchmarks of determining future success. Winners will win – period. It’s really still the one true measuring stick of performance. Ryan Tannehill (Miami) and Alex Smith (KC) have had some success in their careers (college and pro), but, then again. Tom Brady has won in the NFL (4 times) and at Michigan (people forget that), Joe Flacco won at Delaware (he took his teams to DI-AA title games) and in the NFL (SB47), and Peyton Manning has won in high school, college, and the NFL. Winning is winning, and there is no substitute for it. Far too many head coaches and GMs are now drawing unemployment checks or serving as scouts and assistant coaches because they fell in love with a player’s “metrics” or measurements and thought they could “motivate” or “teach” them to win. As a former player, I know that all too well. It does not take intelligence to see this, although I suspect it will take more than you possess. Winners win – plain and simple.
As someone who does Data Science for a living, this is laughable. This model should be a starting point, not the end product. Your assumptions of what constitutes a “core” of a team are puzzling. The fact that you haven’t mentioned weighting of data elements that are more predictive than others is concerning.
If people made bets based on this table you produced, they would be broke.
The Ravens were listed as Peter King’s #1 in Power Rankings, and you think there’s a 9% chance they reach the playoffs. PK isn’t exactly a seer, but this should be alarming to you because he’s clearly someone who watches more NFL than you and anyone who helped with these predictions…. Yikes…
Part of the reason I love to be a Raven’s fan is everyone underrates us, in every single playoff game we played in 2012 we were projected to lose. One example is how underrated Joe is even though he has broken so many records.
Most starts by a quarterback in first season: 19
Most starts by a quarterback in first two seasons: 37
Most starts by a quarterback in first three seasons: 55
Most starts by a quarterback in first four seasons: 73
Most starts by a quarterback in first five seasons: 93
Most wins by a quarterback in first 80 consecutive starts: 54
Most combined regular and postseason wins in first three years as a quarterback: 36 (tied with Dan Marino)
Only quarterback to start and win a playoff game in each of his first five seasons
Most road playoff wins by a quarterback: 6
Most touchdowns in a postseason: 11 (tied with Joe Montana and Kurt Warner)
Most touchdowns without an interception in a postseason: 11 (tied with Joe Montana)
First quarterback to have a passer rating over 100 in all four games of a single postseason.
Most consecutive playoff games with three passing touchdowns: 3 (tied with Bernie Kosar, Kurt Warner, and Aaron Rodgers)
Note that after only five years Flacco already has more road playoff wins than any other QB in the HISTORY of the NFL. And his best years are ahead of him.
That’s why we don’t rank high, we don’t put up big numbers on stat lines, we put them up in the win column.
If you believe in this so much — enough that you would publish it — then give me 11-1 odds that the Ravens make the playoffs and put your money where your mouth is. Otherwise, just a bunch of number crunching that has no real world application because it clearly is an oversimplification with bogus assumptions.
The math all looks pretty sound. I think the big problem with this model is the assumption about aging players and progressing rookies netting to zero for each team. Consistently good teams in the NFL, (the Patriots, Green Bay, Pittsburgh, Baltimore, Seattle, NY Giants just to name a few) are very good at assessing the degradation of aging players, who to sign and who not to, avoiding salary cap situation and proactively replacing starters through the draft and free agency a year or two before so they can progress. By following the assumptions made in this model, we assume that all teams are equal at these and that is simply not true and is really the basis for what teams have success in the NFL. The model also neglects to show the coaches influence on team success and anyone who knows football knows that the coach is more important than any single player or position. In addition by seeing some teams with no franchise Quarterback high on the list, it leads me to believe that the QB position was very undervalued when creating the formula. No teams win with a below average QB that’s just 21st century Football. I think this was a very interesting and I am glad to see more mathematically based models for predicting sports and statistics beginning to be developed, but I think in February the writer will be back to the drawing boards to come up with some tweaks to make it a lot more accurate, but I guess only time will tell
The current limitations to your model are that it underrated players who played through injuries last year, it fails to account for the quality of coaching staffs, difficulty of schedule, and completely ignores the impact of rookies.
The Browns have a better chance of making the playoffs than the Ravens?
This article is nothing but an example of bad science.
I would think you should factor in that 6 NFC and 6 AFC teams, including at least 1 from every division as there is statistically 0% chance of 8 AFC & 4 NFC teams making it.
NFC east for example has all 4 at 42% or less but it is 100% certain one will make it.
Not a great advert for Harvard
As it is and always will be “YOU HAVE GOT TO PLAY THE GAMES”.
Seriously RAVENS only a 9 per cent chance of playoff berth???
This is a fascinating exercise. Attempting to calculate the likelihood of making the playoffs is certainly complex. Each year, roughly half of the playoff teams turn over, with injuries, free agency changes and coaching changes.
Who will rise? Who will fall? Teams that have the most roster depth can overcome injury or free agency the easiest.
This would be fun to revisit at each quarter mark of the season, don’t you think?
Why did you choose Profootballref’s player evaluation over Pro Football Focus’s? Pro Football Focus watches the film as well as looks at the stats. It seems like a more accurate player evaluation and it includes offensive line and all defensive players. They also have player scores based on formations so if a player signs via free agency into a new defensive scheme, you can check their grade in that scheme.
I totally get why the Ravens are so low. They are relying on a 34 year old wide receiver who declined in the second half of last year, a 30 year old “one year wonder” as the lead running back with a rookie as the backup, a rookie wide receiver if he can win the second wide receiver job, and a tight end coming off of hip surgery or a rookie tight end (pick your poison). I don’t think Joe Flacco is elite and elevates his teammates tremendously to the point that this offense will be the same as it has been the last 5 years.
The defense is still above average but no longer elite by any standards.
Also, the division is a tough one. The Bengals have the best offense they have had in many, many years. The Steelers had the best running back and wide receiver in the NFL last year. Its going to be tough to win the division over both of those teams.
“The Raiders, Titans, and Jaguars seemed destined for mediocrity again.”
Mediocrity? Doesn’t that mean of moderate quality/value? Seems to me that those 3 teams are MUCH worse than mediocre. They’ll be terrible.
This is brilliant analysis that requires a minor correction to account for how NFL teams qualify for the post-season playoffs.
As a baseline, each of the eight NFL divisions has four teams. Every team in the NFL technically has a 25% probability of winning their division championship to advance into the playoffs.
Ultimately, the odds of making the playoffs comes down to the probability of having the best record in their respective division OR having the first or second best record of non-division champions in their respective conferences.
Overall, great work!
There is a calculation that is missing, That is the teams internal divisional play. You cannot have a probable calculation where the winner of the division has a 42% chance of making the playoffs as in the case of the NFC east. What you are implying here is more than likely a team or more in the conference will have a better record in other divisions but will not qualify for the playoffs. This may be true but the inevitable winner of a divisions weight is not factored into the calculations
Enjoyed reading the article, Mr. Bullard, but if I may make a suggestion:
The “core players” that comprise your aggregate should be weighted more than the other players on the field, but the other starting players should factor into the equation, too.
An example of my reasoning is that a defense with three outstanding players in the secondary is not going to give up big passing plays as often as a team with two outstanding secondary players and one mediocre player. Offenses look for holes in defenses and exploit them. Holes are not necessarily “core” players, but “core” players are all your aggregate considers.
What accounts for the vast devaluation of the Cowboys’ playoff chances taking into account last year’s success and this year’s acquisitions, despite losing Murray?
He has the Raiders at dead last so he hasn’t taken into account that it’s a completely different team, different coaching and massive upgrades at various positions.
Analysis is way off. In lots of places. I care too much about my own time to spend an hour picking this article apart, so I will just pick apart 1 aspect: a team’s defensive secondary is only as strong as its weakest link. The “top 2” are irrelevant. Liebig’s barrel theory. Good OCs/QBs will expose poor DBs.
It makes no sense how the Panthers are at 19% since theyve made the playoffs the last 2 years and got better players
Well this should be interesting. Either the Dolphins will, in fact, shock the football world, or this will be the last time anyone takes seriously a Harvard statistical analysis of an upcoming NFL season. My money is on the latter.
To all those complaining about the stats in this article —
Understand that the model produces the data, not the other way around. The results are what they are because of how the experiment was conducted, e.g., calculating a starter rookie’s AV by looking to their backup, which by the way is quite clever because you more or less establish a rookie’s floor. How would you predict rookie performance otherwise? As is stated in the article, that’s a beast on its own.
How can you take this experiment seriously when you have the Ravens at 9%?
LMAO. Dumbest thing I’ve ever read. “A Way Too Early Prediction of the NFL Season.” So 100% of useless calculations + it being “way too early” = purposely wasting everybody’s time with a stupid math problem that’s not only wrong but takes all the fun out of sports.
Math is powerful…. many who attempt to use it, not so much. I’d like to make some wagers with these analysts…. Miami and KC top two in AFC? We shall see.
Can you use the same analysis and do an assessment of the accuracy of this approach? Shouldn’t be too tough and would either back up or shut up any nay sayers.
If all teams were equal, each team would have a 25% chance of winning its division and a 12.5% chance of making the playoffs as a wild-card team (2 wild-card spots divided by 16 teams in the conference). That’s a total of 37.5% chance of making the playoffs for each team. 37.5% times 16 teams per conference is 600%.
Of course the teams aren’t equal to each other, so some teams will have a higher likelihood of making the playoffs than others. But as long as there are six playoff spots per conference, the combined percentages for all teams in the conference must be 600%.
The combined odds for the AFC teams in this list is 640%.
The combined odds for the NFC teams in this list is 562%.
Rounding up or down to the nearest integer can account for some of this discrepancy, but not all.
That’s what I said – the numbers don’t add up. He needs to rerun the numbers to account for the conferences (if not the divisions as well).
According to this, if I am reading this right, and I’m not sure I am, so bear with me, the most likely playoff picture will look like this:
AFC: Dolphins and Chiefs get the Bye.
The Steelers (division winner) play the Pats in Pittsburgh round one.
Colts (division winner) play the Broncos in Indy, round one.
NFC: Seahawks and Packers get the bye.
Falcons (division winners) play the Saints in Atlanta, round one.
Giants (division winner) play the Lions in New Jersey, round one.
Sounds cute, but I ain’t buying it. Time will tell though. I’ll come back at the end of the season and we’ll see how things shake out. These are humans playing this game, not numbers, and stranger things happen with people than with numbers.
Any set of numbers that has both the Falcons and Saints finishing ahead of the Panthers and the best defense in the NFC is math that doesn’t add up, no matter how well thought out.
THIS IS MATH….said some guy further up in the comments, and he’s right. Math don’t know football, and is nothing more than a cute exercise in number crunching.
Sabermetrics it ain’t. Not even close.
Nice try though. Kinda
You’re not reading it right.
These are (allegedly) odds of making the playoffs. They are by no means an indicator of overall record. Someone could have a high probability of making the playoffs with a horrible record simply because they are in a sucky division. Conversely, teams with amazing records don’t get as great a chance of making the playoffs (or they get in as a wildcard) because another awesome team is in the same division.
Odds != season win-loss record
BTW, when is he going to get around to recalculating the numbers to account for different conferences. His odds don’t add up evenly for each conference (should be 600% each, not 560% and 640%).
Your model most not have accounted for Adrian Peterson playing for the vikes.
I find it very funny you put the Tampa Bay Bucs so low. They are going to benefit from an easy schedule and and easy division. Their coach put together a great defense and everyone knows it takes time to learn the Tampa 2 because people have to trust their teammates. I understand we have a rookie QB but he should benefit from the tall receivers. I would say they have a great chance to at least go 7-9 if not more.
The Bills spanked the Jets twice last year by a combined 45 points. Both teams improved greatly in the offseason but the Bills greatest addition is not even being measured: coaching. The Bills had bar none the worst offensive playcalling in the league. The firing of Nate Hackett alone makes them legit contenders and still far better than the Jets
Essentially, this kid simply simulated an entire season of Madden without this year’s rookie class. That’s what the “math” is behind the madness.
And just like Vegas, this kid won’t be right either.
Everyone is crying after looking at the final graph and seeing their team in an orange/yellow color – boohoo kids, this is a prediction using math, its not a fool proof system, its not real life. The defensive replies from the low IQ individuals because someone did some math you cannot comprehend and you don’t like it. Read a book instead of posting “OMFG this guy has Ravens at 9% omfg they are so good, no wai! ”
Looking forward to seeing how the %s shake up! Thank you for the article.
I was slightly confused when you describe how you tested the regression, did you use the data from the 2013 season to predict the 2014 season, or did you test the data set from 2014 to predict results for 2014? Obvious issue being that you would have unpacked the AV statistic and simply predicted what would happen if present rosters played at the same competency but with a newly weighted schedule. Otherwise, impressive work!
I read this article and after observing that the odds your model have calculated are wildly different to the betting market, I was provoked into writing about it at:
How are the Vikings so low? There’s no way on Gawds green earth that Vikings are going to finish that low. They have Bridgewater, a top 10 defense, best RB in the league, a formidable WR corps and a decent Oline. I see them picking up at least a wild card spot.
How is that Dolphins Seahawks prediction working out for you?
You can never measure teamwork, drive, heart and luck no matter how much “data” you pull out.
Yeah how bout them phins????
lol seriously, who would ever pick the Dolphins to win the division. They’ll be lucky if they even finish third.