Breakfast Table: Number Crunching

For the last decade or so, Michael Salfino and Scott Pianowski have been putting together an email exchange centered around (but not limited to) the NFL. You might enjoy listening to them haggle. You might prefer a swift kick into the stomach. The Table isn't for everyone; we hope some of you enjoy it.

From: scott pianowski
Date: Wed, Nov 7, 2012 at 8:47 AM
Subject: big blue breakfast
To: Michael Salfino

It's been a good week for overdogs, a good week to lay the favorite. Ten of 14 NFL teams beat the spot, and the Blue States routed the Red States last night. Hat tip to Nate Silver, who went 50-for-50.

If only predicting football were so easy.

Some of the big-name algorithms have interesting views of the current NFL landscape. Football Outsiders ranks the Packers above the Bears, the Seahawks before the Texans, and the Cowboys a lofty 12th. Massey-Peabody believes in Houston but has the same Bears-Packers view. It also puts the Cowboys above the Giants.

Atlanta? The computers don't trust them at all. Indianapolis? A stone fluke, both sites say.

Who's getting the Salfino check mark these days? Who's the class of the NFC East, or the NFC North? Forget the records, I want to know who the best teams are. M-P has Baltimore over Pittsburgh, which surprised me. I wish I could delete the rest of Baltimore's schedule with a couple of keystrokes. If you had to pick a Super Bowl winner for your children's tuition, how long would it take you to say "Atlanta"?

Onto the Week 10 slate. Indy, ever the sportswriter's darling, has to play at Jacksonville on Thursday. Trap game? Does Atlanta cruise in New Orleans, or are the soap-opera Saints ready to pop Mercury's cork? Eli Manning's been in a funk of late - does that end at Cincinnati? (When's the last time the Bengals actually beat a good team? Was Boomer Esiason the quarterback?)

I'd like to talk some Denver-Carolina because I think there's a good Cam Newton debate (or argument) ready to percolate between us. Houston at Chicago should be fun on Sunday night. Everyone has permission to skip the Steelers and Chiefs on Monday. The Bart Simpson Underachiever Special matches Dallas at Philadelphia.

Week 10 Breakfast is served.

From: Michael Salfino
Date: Wed, Nov 7, 2012 at 12:16 PM
Subject: Re: big blue breakfast
To: scott pianowski

Math won on Election Night, not Silver. He has the most known model and did have the most to lose (I'm guessing), but his model was not the most accurate. Drew Linzer at Votamatic.org actually called the election right on the number -- since July!

Politics are low hanging fruit for this type of modeling. There is such a limited range of outcomes. And the polling is tethered so closely to the actual election. The genius of these models is just averaging the polls. If anything, Silver adds an unnecessary complication by weighting them and adjusting them.

So in politics you have a slightly imperfect view of the current state of the electorate that you can make much more perfect by simply averaging all the views. And that view, as you get closer to election day, is perfectly predictive of future (election day) results. In sports, we have a perfect view of the past, no doubt. But that view is not especially relevant at all to what will happen in the future, even if that future is the next at bat or play, never mind an entire game. So all the nerds who gravitated towards sports should have been going into politics where the luddites are even more in charge. That, really, is Silver's genius. Regarding PECOTA, again, you have so many unknowns and so many possible relatively minor adjustments/improvements that can so dramatically improve any player's on-field performance (especially pitchers). So I think comparing players to past players statistically, while great fun, is largely illusory. (But I reserve the right to change my mind.)

I don't look at Football Outsiders. I don't want to start another Massey-Peabody war, either. A few years ago, I wanted a rankings/prediction system that I could understand with simple statistics. Due to my relationship with the Wall Street Journal, I could have approached anyone. I asked Cade because I thought he would be the best person to do it. So when he comes through and puts something like Massey-Peabody together, I have no need to look anywhere else. Plus, I know first-hand the rigor they apply to what they do and to constantly reassessing and adjusting their modeling. But of course as I said earlier, compared to something like presidential polling, there is so much less predictive reliability even when, like Cade and Rufus, you are looking only at things that have the most predictive value. So many things that determine outcomes in football are descriptive of why Team A beat Team B but not predictive at all when it comes to forecasting how Team A will fare against Team C (turnovers, special teams play, explosive offensive plays, etc.).

So with those caveats, M-P sees five championship contenders right now: SF, HOU, NE, GB, DEN. That's it. But of course you have the Giants situation from last year where they squeak into the playoffs and then every turnover falls their way. There's no predicting those things. Another team could get on a lucky roll like that and win the Super Bowl as long as they get a ticket to the dance. It's very unlikely that any specific team will, but obviously much more likely that some team will. But I don't wring my hands over that. It's why I love sports.

Indy is the sportswriters' darling? I generally don't read sportswriters (with obvious exceptions such as yourself), so I have a poor sense of that. Plus, the storm still has NJ semi-apocalypitc (and another storm is on the way). Please, everyone, text "redcross" to 90999 because they will put your $10 to very good use here where it's so very needed. M-P says the Colts are very bad but Jacksonville is worse and that the game is basically a pick 'em. But I just can't pick the Jaguars for some reason (my inner luddite). But let me step aside now and you first crack at the games before addressing them in my next reply.

From: scott pianowski
Date: Thu, Nov 8, 2012 at 1:07 PM
Subject: dogs with different fleas
To: Michael Salfino

Ah yes, Math. It does a mind good.

I don't look at Massey-Peabody and Football Outsiders every week. But maybe once or twice a month I take stock of their charts and use it as a reality check for my own analysis and opinions (you always want to use numbers and scouting in tandem, even though they start from much different areas). I don't know all the special sauce to their formulas, but I do find each set of rankings useful. Of course, it's important to recognize the types of normal outliers that might be worth discounting (for FO, this usually involves the Eagles; the David Bush/Javy Vazquez team of the NFL.)

But this stuff is always evolving, of course, and the game constantly changes as well. I find it interesting that Massey-Peabody just recently started including weight for teams off a bye or a semi-bye (a game after a Thursday game). Teams off a bye against a non-bye opponent win about 54 percent of the time, per Spreadapedia. Not stop-the-world stuff, but something worth considering. You take your edges where you can get them.

I suppose I judge pet sportswriter stories from Peter King (who I generally skim), Sports Illustrated, Sports Center, that kind of stuff. I rarely go A to Z on any of their stories. Not that they don't have good reporters, I just feel like there's better places to get the type of writing, opinions and breakdowns I want. Is Peter King watching hours of game tape at Starbucks? He always struck me as a guy who had no problem cultivating an opinion based off a handful of plays (see Chan Gailey, curiously named King's Coach of the Week, after the curious win at Arizona in mid-October).

And hey, we're all spoiled in today's world of resources. We can rewatch any game we want, from any angle we want, in glorious HD. Pro Football Focus is breaking down every play and putting it into digestible terms. When King puts Tony Gonzalez on his midseason All Pro team (over Rob Gronkowski, blocking overlord), we can quickly prove that it's absolute hogwash.

Pittsburgh deserves to be on anyone's list of championship contenders, along with the Fab Five M-P listed (Niners, Texans, Pats, Packers, Broncos). Everyone bags on the Steelers offensive line but how come every back they use seems to run effectively? Roethlisberger is remarkable on third down, and so many times he defeats a pass rush on his own. The pass defense is somehow elite again, despite a number of key defensive injuries. I'd follow Mike Tomlin into a burning building.

I don't know what to make of the Bears because I can't see Jay Cutler playing 2-3 strong games in a row against quality teams. His mechanics aren't consistent. I don't like the way he commands the huddle, either. How many great quarterbacks didn't have complete respect of the locker room? It's hard to find someone. Montana, Young, Marino, Elway, all the greats today, Aikman, Staubach, Namath, Starr, Unitas, those guys all had (or have) the makeup you need. The only name that comes to mind to me is Terry Bradshaw, but those Steelers teams were so loaded, Jo Jo Starbuck could have quarterbacked them to four championships.

All this yammering from me and I've yet to touch a game. Looks like I'll be driving thru, just like King. I don't like road teams in these Thursday games, and Luck's development is shielding flaws in the Colts. I definitely like the Jags to cover, and possibly steal the game. I picked the Panthers plus the spot but I still don't trust Cam Newton; I know the YPA is shiny, but the rating is mediocre and in particular craters in the red zone. Does QB rating, the stat everyone loves to bag on, correlate to wins and losses pretty much as well as YPA? Cold Hard Football Facts was discussing that this week.

Houston and Chicago strike me as similar teams, though the Texans have better balance and no obvious flaws. But Chicago's a tough prime-time location when the Bears are nasty again. Give me Chicago there. The Patriots will probably crush the Bills, as they always do.

Sign me up for the home Seahawks in a romp, too. Russell Wilson, to almost no fanfare, has a 120.2 rating at home and the most touchdown passes of all the rookie quarterbacks. He's not asked to throw a ton, but they take regular shots in the middle of the field. He also uses his mobility to set up the pass, not to take off and run - that's obviously a big plus. And no one will ever worry about his arm strength. He's reasonably accurate, too.

And I can't help but root for the story with the Jets - I want to see Tim Tebow start eventually. I don't care if it's a sideshow, I don't care if it's a mess. Tebow is the type of guy who will run through walls to succeed, and if Mark Sanchez had 10 percent of that, he'd be a lot better quarterback. I just want the experiment to play out, something different.

Colin Cowherd (completely an acquired taste but I'm starting to enjoy his show for what it is, a wide-audience hit) talked about Sanchez being "too casual" today, an angle and phrase he got from Bill Romanowski. I can see that. Sanchez never had much to worry about at USC. Good weather, clean pockets, receivers who are wide-ass open. No wonder Sanchez, and Matt Leinart for that matter, came to the NFL and had no idea how to be a pro or succeed from a messy pocket.

The 80s version of Gordon Gekko wanted underlings who were poor, smart and hungry. I can see that. I can get behind that. Wanting to be great is often as important as the greatness itself. I'll take Larry Fitzgerald eight days a week over Dez Bryant.

You get the idea Gekko wants to see Indianapolis and Pittsburgh collide in the playoffs. Blue Horseshoe loves Anacott Steel.

From: Michael Salfino
Date: Thu, Nov 8, 2012 at 5:01 PM
Subject: Re: dogs with different fleas
To: scott pianowski

It's a 1.5-point edge in M-P now for teams off byes/Thursday games.

I don't read much because if I do, I feel like all the good ideas are off the table plus I have to worry that I plant a seed somewhere that grows later after I forgot where I picked it up.

The key thing we get now is the coaches films, which was the holy grail for me. But real good data that we can interpret in creative ways is woefully lacking, especially in football. Baseball is pretty good but not great by any means. For example, why don't we have hitter splits vs. groundball and flyball pitchers? Why not velocity splits by pitch count? Why do I need a physics degree to understand pitch movement charts? If I got started on the obvious things we need in football, this reply would be about 5,000 words. But here's a real easy one -- yards per tackle. ProFootballFocus and the Pro-Football-Reference stat databases are great, but they cost money. This stuff should be free so that more people can get a crack at making good use of it and sharing that with the rest of us. Obama is going to fix this.

The Steelers are coming on in my defensive fantasy rankings after I thought weeks ago we could write them off as average at best. Fantasy rankings aren't the best barometer of defensive efficiency but I throw in red zone possessions and, basically, TD rates. That improves the correlation with reality, I think. (These rankings as you know are for how defenses stop fantasy players, not how they score in fantasy.)

I always defer to M-P for power rankings. But what about the back-of-the-envelope, net-YPA (sack-adjusted yards per pass attempt gained minus allowed) for old time's sake. Here's the top five there: Denver, SF, Houston, Pittsburgh, Carolina (!!). All those are at the plus-1 or better threshold that you generally have to win at least 10 games. Only Denver is at 2.0, the minimum threshold for a certain Super Bowl contender. Bottom five: KC, Jacksonville, Tennessee, Buffalo and New Orleans. All other teams in that barely-good, barely-bad range. Here's a link to a tweet (@MichaelSalfino) where I have the full chart.

So cross-referencing, I really, really like Denver rest of season and beyond into January and February. And we see again how Cam Newton and his Panthers are so very weird. There's no explanation really for the Panthers losing so much with him. Newton is eighth all-time in YPA among QBs with 700-plus passes. He's 8-16. The rest of the group, 684-391. You can point to his sub-par QB rating but the QBs who run for scores get screwed in passer rating because their rushing TDs don't count for anything AND they take TD pass opportunities away and thus swing the ratio of TD and INT rates the wrong way. So let's forget passer rating with these running QBs who score so many TDs. Let's just look at regular YPA and also at INT rates.

I think Cutler is an ass but he seems like someone who commands the huddle. He's not taking any crap out there, that's for sure. He'll clearly point fingers, an attribute of many leaders. Now maybe half the team would like to shoot him when he turns his back, but that's reportedly the case with Ben Roethlisberger, too. Phil Simms also famously pointed fingers at teammates and went into grand theatrics whenever a receiver dropped a pass, too. Namath was a different animal. He'd pat you on the back when you dropped a pass and often call the same play the next chance he got (Richard Caster famously dropped two perfect bombs in a row like this once).

I know you think Arian Foster is overrated, too, but I'll say it here and take the insults from the Foster army who will surely come after me like I just insulted their mother. (Enjoy the TDs, kids, and the unexpected volume, but he's not a great back this year.)

I say the floor for New England this week is 40 points (that's with the Bills scoring zero).

I agree with M-P that the spread in the Jets-Seahawks game is too high. These teams are close to even.

You have to be kidding me with Tebow. The guys Jets fans should want to see is this guy. I figure you'd be a sucker for my research about lower round picks who excelled when finally given a chance. No matter how hard Tebow works (and he barely knows the few plays he runs given how he has to look at his wristband like he's Sarah Palin whenever he's in the game), he can't throw the football to save his life. So any snaps you give to him are opportunity costs you should be spending elsewhere. Sanchez not working hard is beside the point, though I'm not stipulating that this is even true. He also can't throw the ball well enough, it's now clear. (Though he throws it a million light years better than Tebow.) I'm not saying that Sanchez is garbage, either. I'm just saying he's very unlikely to ever be very good. When you conclude that, you need to move on I think.

Cowherd is way too casual a thinker for my taste, but the type who's convinced he's thought everything through perfectly. That, I understand, is sort of the nature of the sports radio business. And when it comes to radio hosts, Cowherd is a Rhodes Scholar, for sure.

Are you saying that Tebow is poor, smart and hungry? Hard to say that about a guy who flies into training camp on a private jet. But at least we know he'll be flying out soon enough.