RotoWire Partners

East Coast Offense: This Year's Outliers

Chris Liss

Chris Liss

Chris Liss is RotoWire's Managing Editor and Host of RotoWIre Fantasy Sports Today on Sirius XM radio.

Where Should Jimmy Graham Have Been Drafted?

Graham is just shy of the receiving triple crown right now, tied for the league-lead in catches with 37 (on pace for 118), leading it with 593 receiving yards (1898) and second in TD catches with six (19). A 118-1898-19 season from your tight end in a PPR league is easily worth the No. 1 overall pick, except in a season where Peyton Manning is on pace for 64 TD passes and 6,029 yards - and three rushing TDs. Graham and Manning are the league's only major outliers through five weeks, and while they'll almost certainly regress, there's little reason to think either couldn't remain atop his positon from Weeks 6-17.

Does that vault them over the top running backs, knowing what transpired in the first five weeks? I'd still take Adrian Peterson ahead of them - especially given the failures of other first-round backs - but it would be a tough call at pick two between Graham, Manning, LeSean McCoy and Jamaal Charles.

Some Noteworthy Facts Through Week 5

Torrey Smith is second in the league in receiving yards, but has only 27 catches and one score.

Wes Welker leads the NFL with seven TDs (on pace for 22).

Adrian Peterson leads the NFL with five rushing TDs despite already having his bye week. Knowshon Moreno and Fred Jackson are tied for second with four.

Among players with 50 or more rushing attempts, Moreno is third in YPC with 5.1 behind only LeSean McCoy and Alfred Morris, ahead of Peterson and Jamaal Charles.

Doug Martin leads the NFL in rushing attempts despite already having had his bye week.

First-round RBs Martin, Ray Rice and Trent Richardson are all averaging 3.5 YPC or less through Week 5.

Danny Woodhead leads all RBs in receptions (31), while Darren Sproles leads them in yards (308). Sproles' teammate Pierre Thomas has 28 catches to Sproles' 26, however. If you want to know why the Saints wideouts haven't done much, consider 91 of Drew Brees' 140 completions have gone to Sproles, Thomas and Graham.

The Browns lead NFL defenses in yards-per-play against (4.3), and the Jets are second at 4.5. The Falcons, Broncos, Redskins and Chargers are the four worst.

The Carolina Panthers are in second place.

The Giants have 20 turnovers. The Jets are second with 12. The Colts are second to last with only four, i.e., there's as big a difference between 1st and 2nd as 2nd and 31st.

The Chiefs and Seahawks have forced 15 turnovers each, while the Chargers have forced three and the Steelers none. Even though the Steelers defense has no turnovers and only four sacks, it's 11th in yards-per-play against (5.1).

The Cowboys pass defense has allowed 14 TDs. The Browns have allowed only two. That said, the Browns lead the NFL with eight rushing TDs allowed, while the Bucs have not allowed a TD on the ground yet this year. The Ravens and Cardinals have only allowed one TD on the ground each.

Some Considerations When Using Probability to Predict Outcomes in NFL Games

Heading into the Monday night game, the Falcons were 9.5 point favorites or roughly 83 percent (according to Vegas) to win the game. But after watching the Jets go toe-to-toe with them and drive down the field for a game-winning kick, it seems like those odds were wrong.

Of course, one could argue the odds were right, and that the 17 percent simply came in, but that's (1) not falsifiable, i.e., even if you said the Falcons had a 99.99 percent chance to win, you could argue the .01 percent came in; and (2) Watching the game, the two teams looked roughly equal, i.e., the Jets didn't simply luck into plus-3 on the turnover ledger, while otherwise getting outplayed. In fact, Geno Smith had 10.0 YPA, and Matt Ryan had 7.1.

Another argument is to say given what we knew about those teams, the Falcons were an 83/17 favorite, but that's like saying, given what we knew about a coin, there was a 50 percent chance it came up tails, when it later turned out the coin was weighted heavily toward heads. It was reasonable to think tails was 50 percent at the time, but also wrong.

Moreover, if we were to schedule a rematch of this game in Atlanta next week, it's likely we'd decide the odds were 65/35 Atlanta, or something close to that, further undermining the claim that the odds were really 83/17 as the Vegas moneylines implied.

This differs from situations like flipping regular coins or calculating poker odds where we know exactly what the odds are, no matter what the result is - assuming the deck isn't stacked or the coin isn't weighted. If I go all-in pre-flop with 6-6, and the other player has 8-8, my odds (roughly 20%) are bad, and that's so whether I end up catching one of my two outs or not. If I wind up catching, we don't then revise the odds the way we might when one football team surprises another.

This is so because the cards have fixed values, and we can repeat the experiment over time. Players' and teams' values fluctuate, and no game can be repeated under the exact conditions as a previous one. So instead of merely concluding
the unlikely possibility came in, we also have to consider whether we misunderstood the values of the teams. That those sixes were actually upside down nines.

I suppose you could decide whether a team was similar in most material respects to a set of prior teams and treat, for example, the 2012 Cleveland Browns like the 2013 Rams and figure out their win probabilities as if they were both a particular (crappy) poker hand. But it's hard to decide what differences between them matter and what don't. And the performance of comparable teams might converge against some opponents, for example, and diverge against others.

I'm aware some people have had success using similarity scores with players and teams to predict likely outcomes of games, and I'm open to learning more about how it would work. In some ways such a model mimics the way our minds work when picking players or handicapping games, e.g., maybe we don't like laying points on the road against a team with a tough defense because we've seen home dogs with good defenses make things tough on top teams before. Or maybe we didn't like Reggie Bush coming into this year because older, injury-prone backs off decent workloads playing on turf typically get hurt. We're constantly selecting certain criteria we deem relevant in comparing current situations (and players) to prior ones even if there are many obvious differences.

So if you were to figure out what makes players or teams similar generally and apply those criteria across every game, you're doing the same thing only without relying on your faulty memory or hidden biases. (Of course, your biases might show up in the criteria you choose).

In the end, while using similarity scores to convert teams into poker hands (so to speak) seems like an oversimplification, it's clear we're stuck with using probability somehow to predict games. After all, whatever the precise probability, if there even is one, that the Broncos beat the Jaguars this week*, almost no one would argue it's 50/50. Actually, a friend of mine, (who incidentally smokes prodigious quantities of marijuana), *would* argue that, saying "it'll either happen or it won't."

*According to Vegas, it's -8000/+4000, the average of which is 6000, or 60 to 1 which is 98.4 percent.

Week 5 Observations

As an owner of David Wilson, I honestly have to say: "**** fantasy football."

It's one thing to lose Michael Vick for a half, or Darren McFadden early in the first quarter of Week 4 - you're upset, and the timing is rarely good, but those players were discounted appropriately. With Wilson, you endured his benching in Week 1, terrible blocking and play calling for four games, ineptitude by Eli Manning that shelved the running game and ended drives, a called-back TD on a bogus holding call and a timeshare with scrubs like Brandon Jacobs and Da'Rel Scott. Finally, Scott gets cut, the Giants draw a great matchup, and Wilson scores early. Here it is - that monster game there for the taking, a reward for those who showed patience or bought lowest. And he leaves with an ****ing neck injury.

Serious question: What would Eli Manning or Tom Coughlin have to do to lose their jobs. I think Coughlin should go, and Eli should stay, but regardless, what would the Giants record/performance have to look like before management made changes? Would 0-8 be enough? What about 20 picks in half a season for Eli? I'm just curious where the line is, or whether the team can go into a bottomless abyss before a single change is made. Hell, Alex Smith lost his job, and the team was winning. Cam Cameron lost his offensive coordinator gig, and the team was at least average (and went on to with the Super Bowl partially as a result). What's it going to take before there's accountability in New York? Other than David Wilson's benching, of course.

Speaking of Alex Smith, it's funny Jim Harbaugh replaced him with Colin Kaepernick because the latter gave the 49ers a chance to win the Super Bowl. Only remember, the 49ers lost in the championship game with Smith in overtime largely due to Kyle Williams' fumbles, while Kaepernick got to the Super Bowl because Matt Ryan came up 10 yards short on a play that could have been called pass interference. But Kaepernick has much more upside, obviously - he can throw down the field, and he can run! But Smith has seven more rushing yards than Kaepernick this year (as well as 233 more passing yards) and two more completions of 20-plus yards. How is this possible? Because like Smith last year on the 49ers, Kaepernick is playing the role of caretaker, not playmaker. In fact Kaepernick is 22nd in passing attempts, but behind the pace of players like Cam Newton, Terrelle Pryor, Jake Locker, Christian Ponder, Brian Hoyer and Josh Freeman who have played fewer games.

The Giants defense is actually not that bad - 12th in YPA allowed (7.0) and 14th in YPC (3.9). This despite giving up 36, 41, 38, 31 and 36 points in their five games this year. Given the Giants lead the league with 20 turnovers, that four of the TDs scored against them were either on special teams or offense and that their offense is last in the league with 25:59 minutes of possession per game, that disparity isn't too surprising.

Congrats to Matt Schaub who broke Peyton Manning, John Elway (and many others') record for consecutive games with a pick six. Sadly, fantasy owners and Texans fans won't be able to spew their most hateful thoughts to him on Twitter anymore.

Tony Romo threw for 506 yards Sunday, the 15th time a player eclipsed the 500 mark. But Romo did it on only 36 attempts, the fewest ever. (Y.A. Tittle was the only other player to do it with fewer than 40). In other words, it was the most efficient 500-yard game in NFL history (14.7 YPA). Yes, Romo threw a costly late pick, but let's cut him some slack the way we would a pitcher with a perfect game who gave up a home run in the ninth to lose 1-0.

And while Romo throwing a pick to a diving linebacker hurt the Cowboys' chances to win, nothing sealed their fate like Jason Garrett and Monte Kiffin not instructing their players to let the Broncos score rather than allowing them to have a game-ending first down. The odds of missing a chip-shot field goal are about one percent. The odds of driving down the field with a minute or more left and scoring a game-tying touchdown (in a 55-48 game, no less) are substantially greater. That everyone on Twitter (and eventually even the genius that is Phil Simms) came around to this obvious conclusion and the Cowboys coaching staff did not is disturbing. It's one thing to bitch out on 4th and 1 from mid-field in the first half, but quite another to be out to lunch when the options are so clear and stark. I sincerely hope a reporter grills Garrett et. al. about this during the week. (I'd also like to hear John Fox say whether the Broncos had instructions not to score).

I'm beside myself for not getting Keenan Allen is some key leagues after I cited him as a player here last week. I was outbid (I think) by Mark Stopa in the Stopa Law Firm League, but I know I outbid Jeff Erickson in our Steak League, ($29 to $10), but the bid fell through because it wasn't in a separate tier from another apparently higher-ranked bid on MyFantasyLeague. I say "I think" Stopa outbid me because that league's on the same service, and I also erroneously put all my bids on one slate and thought they'd take. (Apparently, you need to start a new series of bids for each player you want). In any event, it's now obvious he's the team's best and most explosive receiver and might have been a first rounder in the NFL draft had he not had a foot problem.

Pick up Roy Helu last week. Or do it now at least. Major upside, and Alfred Morris is nursing a rib injury that could get aggravated even if he plays.

In my 0-4 14-team NFFC main event league, my first pick was Calvin Johnson. When I found out he was scratched, I was forced to put in my only other available receiver, Rueben Randle. I would not have started him otherwise. Between Randle and Austin Pettis, I was up by a sizable margin heading into the afternoon games, and I pretty much booked it as a win. Then I checked, and I was somehow behind by 13 points. He had Romo go off for 56 (6 points per pass TD). So I needed decent games from Colin Kaepernick and Vernon Davis. But the 49ers were winning in a blowout, and neither player was doing anything. I tweeted a complaint about it, and not 10 seconds after I sent it, Kaepernick hit Davis for a gratuitous 64-yard TD. I still lost when Jeremy Kerley went off on Monday night, but I got my money's worth.

The Browns are a good set-up for a quarterback, so it's too bad Brian Hoyer's out for the year. Josh Gordon is a rising star, Jordan Cameron is a prototype pass-catching tight end, Davone Bess is a good slot option, the offensive line is good and the defense will usually keep it close. Even Brandon Weeden can succeed at times, but he stands in the pocket with the urgency of a DMV clerk.

Teams with bad offenses and good defenses are death on your fantasy teams. The Bucs, Jets and Bengals qualify.

Andrew Luck has only 7.3 YPA this year after getting 7.0 last year (league average is 7.3 in 2013). But he's 15-6 in his 21 starts, and leads the NFL in come-from-behind fourth-quarter wins over that span. (Sure, Tom Brady and Peyton Manning are usually leading in the fourth). So is Luck the next great NFL QB as his record and dramatic comebacks seem to indicate, or one of many decent young QBs as the numbers seem to say? Here's an argument for the former.

The Lions never win at Lambeau Field, and when Calvin Johnson was scratched, I switched my survivor pick from the Rams to the Packers (at least my official one as my pools have different rules). Grantland's R.J. Bell recently wrote about how Vegas moves the lines in the absence of key players, and Johnson clocks in at only 1.5 points, well below even marginal quarterbacks like Sam Bradford. But I think Vegas is wrong here. In fact, I'd give Shaun Hill a better chance to beat the Packers with Johnson than Matthew Stafford without.

The Chargers-Raiders game should have been played so late that no one watched it. It was Thursday night bad. Terrelle Pryor might not be polished but he sure makes great plays under pressure. He deserves the season to develop.

You can follow me on Twitter here