OK, let’s start with the updated rankings here.  

            The way this works is, I personally figure the rankings; Dave Studenmund cuts the rankings out of the article and puts them into a program, created by Tony Pellagrino, which causes the updated NFL power rankings to sometimes appear on the front page of BJOL when you log in.   Not that you need to know about that or care; I just thought I’d share.

            The power ratings aren’t changing much any more.   The teams that were at the top of the rankings three or four weeks into the season—Indianapolis, New England and New Orleans—are still there, and the teams that seemed to be the worst in football very early in the season—Kansas City, Oakland, Detroit and St. Louis—have in fact proven to be the worst teams in football.   We missed a lot of predictions on games by mis-rating a few teams based on early-season results:  San Diego, Denver, the Giants, Chicago, Tennessee.   A few teams have moved up or down during the season and screwed up our predictions; most teams are about where they started out.   Pittsburgh has gone in the tank; don’t know if anybody saw that coming.   Washington is playing much better now than they were a few weeks ago.

            Predictions. . .we were 13-3 last week, sneaking our predictions into the Hey, Bill, section about five minutes before the games started on Sunday.    I do that because I don’t know how to post articles myself; I send them to Dave and he posts them, only sometimes I run out of time, so I put them into “Hey, Bill” because I know how to do that so that whatever I have written shows up on the site immediately.  On Sunday we were “right” on every game except San Diego at Dallas, but we missed on the Thursday game and the Monday Night game.   We’re now 90-54 on the season, 62.5%.    I think we’re still below our prediction record last year, but. . .. who cares?   Not what it’s about.  More on that in a moment.

            This is how the 32 teams rank, offensively:

            And this is how they rank on defense:

            Let us suppose that Indianapolis were to play at Jacksonville, which they will on Thursday.    Indianapolis’ offense is at 16.8; Jacksonville’s defense is at 11.6, add them together that’s 28.4, minus 1.25 because Indianapolis is on the road, that’s 27.15; we will predict that Indianapolis will score 27 points in that game.    Jacksonville’s offense is at 6.9; Indianapolis’ defense is at 5.5.   Add it together, that’s 12.4, plus 1.25 because Jacksonville is at home, that’s 13.65, Jacksonville should score 14 points.   Indianapolis should win the game 27-14, which is a really good football score, as opposed to 25 to 11; when it comes out 25 to 11 I have to change it to something that looks like a football score.

            Let us suppose that Dallas were to play at New Orleans, which they will on Saturday.    Dallas’ offense is at 10.9, the Saints defense is at 11.0, that’s 21.9, subtract 1.25 because Dallas is on the road. . .Dallas should score about 21 points in the game.    New Orleans’ offense is at 24.6, the Cowboys’ defense is at 7.6, that’s 32.2, add 1.25 because New Orleans is at home; the Saints should win the game about 33-21.    These are our predictions for all the games this weekend:

            Houston/St. Louis shows at 29 to 12, which is a bad football score, so let’s make that 30-13.   Minnesota/Carolina is 27-19, which is not a great football score, so let’s make that 27-20. 

            That chart shows the Miami/Tennessee game ending in a 26-26 tie, which, since I actually don’t have a clue who will win that game, I’d like to go with.   But unfortunately the rules of the exercise prohibit that, so I have to pick a winner. 

            Both Miami and Tennessee started out this season digging themselves into a deep hole, which they are still attempting to crawl out of.   In fact, that’s the dominant story of the season, isn’t it—that, and the two teams still undefeated?    It is the most common cliché of sports fiction:   a team finds itself in an impossible, can’t win situation, and then finds a way to win.   We root instinctively for the underdog, for the team that can’t win but won’t give up.   But has there ever been a season that had so many of these real-life stories?    Miami started out 0-3; now they’re one game behind in their division.   Tennessee started out 0-6; since then they’re 6-1.   San Diego started out 2-3, and has now so completely taken charge of their division that they seem likely to start the playoffs with a bye week.   Green Bay starts out 4-4, but now the Wild Card is theirs to lose.    It’s been that kind of a year.

            But the difference between Miami and Tennessee is this.   Tennessee early in the season was really bad.   Since then they’ve been really good.   These are Tennessee’s game output scores for the season:

            They were playing bad football early in the season, with game output scores down into the 80s and 70s.   They’re playing very well now, having been over 100 for seven straight weeks.  Miami’s not like that; Miami was playing pretty well early in the year, they’re playing pretty well now.    The difference there is that their early schedule was by far the most difficult in the NFL, whereas their schedule since has been more reasonable.  

            Playing Indianapolis, San Diego, New Orleans, the Jets twice, New England. . .you’ve kind of got to expect to lose some of those, even if you play pretty well.   The Dolphins are not really playing any better now than they were early; it’s just the schedule.

            What I’m saying is, if we based our predictions on how the teams have played in the last six weeks, we would pick Tennessee to win by about four points—which is in fact what I would predict, were I free to make predictions based on my own intuition.   But I am operating out of a strict analytical structure that bases predictions on the entire season, and, by that system, Miami comes out ahead, 26.4 to 26.0.   So we’re picking Miami, 27-26.    We’re probably wrong, but. . .that’s the system.

            Miami’s schedule has still been the toughest in the NFL this year, but they’re now only 1.6 points per game worse than Tennessee—about 21 points worse, on the season.  “S O S” stands for “Strength of Schedule”:

            Green Bay and Minnesota have had the softest schedules in the league, in part, because they have played the Lions twice each.    Dallas has been the most consistent team in the league, and Tennessee the least consistent:

            The formula for “consistency” is simply twenty minus the standard deviation of the Game Output Scores.   Tennessee’s Game Output Scores, given above, have a standard deviation of 8.9, therefore the team has a “consistency score” of 11.1. 

            That’s one way to look at it, but another way to figure “consistency” would be by looking at the week-to-week changes—in other words, subtracting each Game Output Score from the previous one, and summing up the changes.   By that method, Tennessee would score as much more consistent.  They were consistently bad the first six weeks; they have been consistently good since then.     Consistency, one could argue, is not playing at the same level all year; it is playing at the same level from week to week.

            Hmm. . .maybe I should figure that.   By that method, the most consistent team in the league is still Dallas:

            This figured by subtracting each game output score from the previous week’s game output score, and summing up the results.   We’re measuring inconsistency; consistency is the absence of inconsistency.   The most inconsistent team, by far, has been Oakland:

            I kind of think that’s a better system, isn’t it?  Maybe I’ll switch to that next season.   Oakland certainly seems more like the most inconsistent team in the league than Tennessee.   Tennessee will play well this weekend.   God only knows what Oakland will do.  

            OK, I’ve got everything in here except temperatures; these are the current temperatures of the 32 teams:

            That completes the required elements of the BJOL NFL report.  Now I wanted to talk frankly about what is wrong about the way I have been doing this.

            The common mistake that is made by virtually all amateurs who try to dive into the analysis of baseball is to attempt to leapfrog the discussion, and go immediately to the great questions that are the ultimate goal of the analysis.   People want to begin immediately by rating players, by deciding who should be in the Hall of Fame, and by predicting who will win the National League next year.   People want to begin an analysis of fielding by figuring out who the best fielders were.

            The essential contribution of the serious analysts who began in the 1970s—myself, Dick Cramer and Pete Palmer, and probably somebody else who I am offending by not including him—the essential contribution that we made was to break the discussion down into smaller and more manageable issues.   Small questions have actual answers.    Great questions do not.  Rather than beginning the discussion with an argument about how Mike Schmidt compared to George Brett, we began with an analysis of the relative value of a single versus a double, a double versus a home run, etc.  What is the relative importance of pitching versus fielding?  What is the relative value of a double play?   What role does the pitcher play in getting double plays, what is the role of the fielder, and to what extent do double plays simply result from having too many runners on base?    What role does the catcher play in preventing stolen bases, and what role does the pitcher play?   What is the role of clutch hitting in the creation of runs? By beginning with very small questions and worrying at length about how one of these questions related to the next, we were able to crawl gradually toward the great questions which had always been there waiting for us at the end of the discussion.   We still couldn’t answer those questions, and still can’t, but we could contribute toward the public’s understanding of those questions by resolving some of the internal issues.

            In the early 1980s, one of my editors wanted me to write a book which would have been called “Bill James Tackles Football.”  It seemed to him like a natural extension of the success I had enjoyed in writing about baseball.    The problem, I explained to him, was that I didn’t know anything about football. 

            But the analysis of football, honestly, has never taken off at the same level as the serious analysis of baseball, and eventually I decided I should try to tackle football—not pretending to understand it, not intending to offer judgment on the work done by anyone else, but simply trying to find a way to nudge the discussion forward.   These articles have been written as a part of that effort.

            But the biggest mistake I have made here has been to try to leapfrog the discussion, and to try to leap directly onto one of the great questions at the end of the discussion: How good, exactly, is each team?    I have tried to use that question as a foundation post to build outward, looking from that to offensive and defensive excellence, to consistency and to hot teams and cold, looking at home field advantage and the value of the bye week, based on this foundation of the team rankings.    It was a mistake, I think, to start with a question so large.   I should have started instead with much smaller questions.   What is the value of a down?   What is the difference between 1^st and 10 and 2^nd and 10?   What is the value of 5 yards on a punt return versus 5 yards on a run?   What is the value of a strong offense vs. a strong defense?  What is the value of a running game versus a passing game, a running defense vs. a passing defense?   What is the value of a linebacker versus a wideout?   What is the value of a sack?   What is the value of an interception?

            I should have started my analysis with questions of that nature.   I didn’t, really, for two reasons.   The smaller reason—the follow-on reason--is that I don’t have a data base to work with in trying to understand those issues.   The primary reason is that I just don’t know that much about football.  

            When I compare myself now to the person I was 35 years ago, when I was doing the baseball work which was the launch pad of my career, I think that my ability to identify relevant questions is no less than it was then, and my understanding of how to work toward the answers is certainly much better than it was then.    We have tools and techniques now that are far, far beyond anything we could have imagined in 1973, when we all lived in the Land Before Computers. 

            But what I do not have now is time.   There was no pressure on me to publish anything then; I could work five, six months on an idea without anybody asking me what was happening with it.   I copied data out of books, put it on paper and tried to piece it together; we didn’t have spreadsheets, but I had time--plus I had a neurotic anxiety about the answers.   I had to know whether baseball was 75% pitching; not knowing the answer to that was driving me crazy.  It still drives me crazy.   I can’t sleep at night if there is a question about baseball to which I don’t know the answer.  Football. . .I don’t care that much.

            But still, I think that’s what I need to do:  I need to focus on smaller questions.   I have enjoyed doing this football work; it makes following the NFL more fun for me, and I see some things and understand some things that I didn’t catch until the last couple of years.  Some of you have been kind enough to say nice things about this football work I’ve been trying to do, and I appreciate that, but I also know that my football work so far is not really meaningful; I’m just kind of re-tracing the steps of other writers.    I need to back off and run at this from a different direction.