Category: Analysis

For What It’s Worth.. If Only the First Half of Games Counted

Go ahead and file this under things I did just because I thought it would be interesting and I haven’t yet checked to see whether it’s predictive at all.  The table below is what 2011′s stats for receivers would have looked like if only the first half of games counted and we assumed that every player played every game of the season.  So basically, take the first half stats times two, then adjust so that every player gets 16 games.  Wide receivers and tight ends are shown together.

The only comment I have is that this isn’t a useless exercise.  Game situations often dictate how much a team passes in the 2nd half.  I’ll have to do some work to see if 1st half stats are any more predictive than full game stats (and even if they aren’t, I still find the table below to be interesting).

Player Team Standard FPs PPR Adjusted YDS Adjusted Recs Adjusted TDs
M.Colston NO 238 352 1,657 114 12
R.Gronkowski NE 226 314 1,304 88 16
S.Smith CAR 224 312 1,516 88 12
J.Nelson GB 221 291 1,254 70 16
C.Johnson DET 219 305 1,466 86 12
D.Bryant DAL 204 285 1,197 81 14
R.White ATL 195 293 1,230 98 12
A.Johnson HOU 189 304 1,774 114 2
J.Jones ATL 189 259 1,173 69 12
J.Graham NO 185 287 1,250 102 10
A.Green CIN 182 256 1,218 75 10
M.Wallace PIT 179 253 1,310 74 8
W.Welker NE 173 271 1,250 98 8
V.Jackson SD 173 241 1,368 68 6
J.Maclin PHI 168 262 1,204 94 8
V.Cruz NYG 167 233 1,312 66 6
G.Jennings GB 167 271 1,191 103 8
A.Gates SD 167 248 1,066 81 10
L.Robinson DAL 153 210 928 57 10
J.Finley GB 150 222 904 72 10
H.Nicks NYG 150 229 1,257 79 4
S.Johnson BUF 149 227 1,014 78 8
B.Celek PHI 149 223 1,010 74 8
T.Smith BAL 143 199 954 55 8
L.Moore NO 140 211 805 71 10
D.Heyward-Bey OAK 140 211 1,042 71 6
B.Lloyd NE 139 224 1,033 84 6
B.Marshall MIA 138 220 1,144 82 4
A.Hernandez NE 135 221 985 87 6
K.Britt TEN 128 235 1,163 107 2

Checking Back in With Wes Welker, Roddy White, and Andre Johnson

A few weeks ago I wrote a post for Rotoworld that focused on the older WRs sitting at the top of ADP.  Be sure to check out that post as it has some good info.  I figured I would compare the careers of the three receivers on a Fantasy Points Over Par basis.  Fantasy Points Over Par is essentially a way to measure a WR’s efficiency on a per target basis.  It’s a measure that looks at each target relative to what all targets from that line of scrimmage produced in total fantasy points.

The graphs below will show the FPOP on a per target basis for each season in a WR’s career.  Comments below the graphs.





The graphs are basically illustrating how many points above average, or Par a receiver is on a per target basis.  To illustrate, let’s take a target from the opponent 1 yard line.  On average that’s worth 3.15 standard fantasy points.  Catch the ball and a receiver essentially just picked up 3 points over par.  Drop it and the receiver went negative in terms of points over par.  We do that for every target in a season, average them up, and we have FPOP for each season.  On the graphs a bar above zero is better than Par, a bar below zero is below Par.

You can see that Andre Johnson became an extremely efficient receiver as soon as Matt Schaub got to Houston.  Schaub probably isn’t the world’s best QB, but he’s head and shoulders above David Carr.  Johnson doesn’t show any signs of losing his efficiency.  That hasn’t been his problem.  Staying on the field has been his problem.

Welker doesn’t look like an extremely efficient WR, but that could also be largely because I’m not currently separating out the value of short targets compared to deep targets.  Doing so would (I suspect) make Welker look more efficient.  In any case, it doesn’t look like Welker is really seeing any kind of age related decline yet, and that’s primarily what I’m interested in today.

Roddy White is the interesting case.  He’s been an efficient WR in his career, but not on the level of Andre Johnson.  Johnson has a few seasons that are better than White’s best season of FPOP.  White has survived more on target volume.  The ATL offense has actually been really two dimensional.  Hand it off to Michael Turner or throw it to White.  But 2011 was the first time in a few years that White turned in a negative FPOP season.  He was actually slightly below average.  He did less with his targets than you would have expected.  That’s probably not as bad as it sounds.  Consider that the average value of a target in fantasy football is influenced by a lot of things.  Some players only see the field when they’re fresh.  White and other top WRs are going to play most downs and they’re going to play against coverage geared to stop them.  If we compare the production of those #1 WRs vs. situational players, we shouldn’t be distressed if the top WRs come out looking closer to average in some cases.  A shorter way of saying that is that we shouldn’t be concerned if Jordy Nelson is more efficient than Roddy White.  Although it would be fine to compare White to Andre Johnson or Larry Fitzgerald.  Basically any player who is seeing similar situations.

It’s not the case that White couldn’t bounce back from that kind of season, as Larry Fitzgerald had an inefficient 2010 and then followed it up with an efficient 2011.  It could also be possible that having Julio Jones on the field actually helps White by making sure that he’s not the one the defense is geared to stop.  In fact the two could help each other in the same way that Reggie Wayne and Marvin Harrison did for a few years.  There were years where Harrison was a #1 WR and Wayne was maybe 1b.  Having an inefficient season isn’t  death sentence, it’s just something that you need to keep in mind when you’re weighing your options at WR.

The Case of Greg Little (And Does Experience Matter?)


The obvious answer to the question of “Does experience matter?” is “yes”.  Experience certainly matters.  But when the Cleveland Browns selected Greg Little with the 59th pick in the 2011 draft, they were essentially casting a vote in favor of size and physical attributes over experience.  Little’s best college season had been a mere 724 receiving yards with 5 touchdowns.  Little was a converted running back, so his experience as a receiver is just a few years.

Perhaps an interesting comparison to make in Little’s case is Torrey Smith, who went to the Ravens just one pick before Little in the 2011 draft.  Whereas Little is a physical beast short on experience, Smith exhibited signs of being an accomplished receiver at Maryland.  Smith caught over 1,000 yards and 12 touchdowns in his last year at Maryland.  Those numbers were good for about 35% of Maryland’s yardage totals and 45% of its touchdown totals.  Those percentages are the signs of a receiver that has figured out how to stay productive even when coverage is geared to stop him.

To take the comparison a step further, let’s look at what the receivers did as rookies.

Torrey Smith Rookie Stats

Year Age Tm G Rec Yds Y/R TD Y/G
2011 22 BAL 16 50 841 16.8 7 52.6
Career 16 50 841 16.8 7 52.6
Provided by View Original Table
Generated 6/19/2012.


Greg Little Rookie Stats

Year Age Tm G Rec Yds Y/R TD Y/G
2011 22 CLE 16 61 709 11.6 2 44.3
Career 16 61 709 11.6 2 44.3
Provided by View Original Table
Generated 6/19/2012.

Torrey Smith outproduced Little while catching fewer passes.  But the interesting thing is that Little was targeted more than Smith also.  Little had 120 targets on the season to Smith’s 95 targets.  One thing I’ll be doing on a regular basis this year is looking at players in terms of their fantasy production relative to expectation.  Not all targets are created equal, so instead of looking at players in terms of Fantasy Points/Target, I’ll actually be looking at them relative to what the average receiver would have produced with a target from the same spot on the field (I call this measure Fantasy Points Over Par, or FPOP).  I’ve actually done that with Smith and Little.  Little had the worst FPOP for any receiver with at least 100 targets last year.  Smith had an FPOP that was well above average.

Here are two graphs that show Smith and Little’s FPOP based on field position.

Torrey Smith FPOP Based on Line of Scrimmage


Each dot on the graph is an individual target.  Dots above the dashed line are targets that result in Fantasy Points Over Par, while dots below the dotted line are of negative value.  The dots well above the Par line are touchdowns.  You can see from the smoothed blue trend that Smith was above average almost everywhere on the field.  He had a number of touchdowns that came from outside the red zone, but he also had a number of other catches that were well above Par in terms of fantasy points you could expect from that spot on the field.

Greg Little FPOP Based on Line of Scrimmage


You can see from Little’s graph that he was below average almost everywhere on the field.  It’s not just that Little didn’t have any long touchdowns like Smith did, it’s that he didn’t even generate very many plays that were significantly above Par.

It’s probably fair to attribute some amount of Little’s suckitude to Cleveland QB Colt McCoy.  It is true that it’s very difficult for a WR to be good if his QB is bad.  However, it’s also probably true that Colt McCoy could have had more of a chance if Cleveland’s WRs weren’t converted QB Josh Cribbs, converted RB Greg Little, and Mohamed Massaquoi.  Cleveland’s WR corps was selected almost as if the skills of being a receiver aren’t the kind of skills that require practice to be good at.

I don’t think it’s possible that Little could be as bad in 2012 as he was in 2011.  But part of what makes me say that is the idea of reversion to the mean.  Little was so bad that he has to be better in 2012.  I started this post with the question of whether experience matters for receivers.  For Little, getting another season under his belt as a receiver will surely help.

Guest Post Round-up

In Honor of the Facebook IPO: Hacking/Disrupting the NFL

gp-priceIn Game Plan I point out that while in the realm of football few decision makers are ever under the age of 40, the world’s hottest tech company is run by a guy who started the company when he wasn’t even old enough to drink.  That tech company, Facebook, is now worth more than all of the NFL’s franchises combined.

The primary difference between managing or coaching a football team, and managing/running a tech company, is that software is a domain that allows practitioners to get good on their own, while in football the only way to acquire experience is to ask somebody else’s permission to do so.  In domains where practitioners are allowed to get good on their own (poker, music, software) the age of the masters of those domains tends to be extremely young.  In domains that require that you ask someone’s permission to get good, the age of masters tends to be much older.

I saw something last night that told me that the days of football being a permission based knowledge domain are coming to an end.  I saw this headline:

A new way to make six figures on the Web: teaching


When I say that football is a permission based knowledge domain, I mean that in order to acquire the knowledge to become a master at it, you literally have to ask someone’s permission.  You have to apply to become an unpaid assistant at a high school or college, and then you have to hope that someone in that organization takes a chance on you.  The internet is going to change that.  You’ll still have to ask someone’s permission to get paid to do something football related, but you won’t have to ask anyone’s permission to actually acquire football knowledge.  That difference is going to be a disruptive game changer for the NFL.  Here’s how it’s going to happen.

The article I link to above is for an online education site that allows teachers to actually charge for disseminating their knowledge.  Let’s imagine something that I’ll call “Football University”, but which is essentially just a collection of videos on a site like Udemy, the one mentioned in the article.  The key for Football University is that the teachers can teach the class just once, but can profit from it indefinitely.  They don’t have to fill up a college lecture each semester in order to make it cost feasible.  They teach the class once, then the video stays on Udemy forever and allows the teacher to continue to profit from it.

Now let’s imagine the kind of courses that could be taught at Football University, and the kind of people who might teach those classes.

Maybe former Browns scout Matt Williamson teaches a series of classes on scouting various positions.  Scouting Cornerbacks 101, Scouting Linemen 101, etc.

Maybe former Packers VP Andrew Brandt or former Broncos GM Ted Sunquist could teach a series of classes covering NFL player contracts, the salary cap, and the finance of football.

Maybe Smart Football author Chris Brown teaches a series of Xs and Os based courses.  The 46 Defense 101, etc.

Maybe the emerging group of sports analysts all have a group of courses.  People like Skeptical Sports’ Ben Morris and Advanced NFL Stats’ Brian Burke could teach classes that cover statistics as applied to sports.  Predictive Analytics 101, Sports and Probabilities, etc.

Retired/fired coaches could get into the act as well.  They could teach classes about motivation, or the organizational/managerial side of coaching.

I’ve tried to throw out a lot of examples because the internet supports abundance.  It supports the creation of every type of imaginable content.  The emergence of sites like Udemy now make the economics favorable for the people that I mention above to actually take time out of their lives to do what I suggest (and if the people I mention don’t do it, someone else will).

However, just because the internet can change that way that people teach/learn about a business like football, doesn’t necessarily mean that it will be disruptive to the NFL, which is what I suggested in the title of this post.  Except actually, it does.  As I discuss in Game Plan, the NFL is currently drawing its decision makers from tiny talent pools (ex-players) because the infrastructure to educate outsiders has never been in place.  But with the education of what I call “obsessed amateurs” through online learning, the traditional paradigm for educating people about football is going to change.  Ex-players will no longer have a stranglehold on the acquisition of football knowledge.

The disruption that I am talking about is already under way. 

One of the Fantasy Douche readers has written me a few emails to tell me his story, which as he points out, is pretty much exactly what I talk about in Game Plan.  This reader is a current college student and ex-poker player.  He took a small deposit and turned it into over $10,000 by playing poker online (before the government shut down poker sites in the US).  Now this reader is trying to get into the business of football.  His story is illustrative of why the teaching/learning system that I discuss above is an important change.

This FD reader is what I would call an Obsessed Amateur, so let’s just call him O.A. 

OA isn’t an insider to football, but he wants to become one and he’s willing to burn a lot of calories to become one.  He sends notes to me and other internet/twitter folks and asks questions about how he can break into the business of football.  He spends time going through Brian Burke’s Advanced NFL Stats, paying attention to things like the relative value of running backs compared to receivers. 

OA worked through his college honors advisor to actually get a meeting with some of the coaches of the D1 school he is attending in order to pitch them that he could be their Game Theory/Analytics guy. They were impressed with some of the things he had to say, like a suggestion that their team should rely on one of their more efficient, but underutilized goal line running backs.  But there were some red tape hangups with working with that program, so now OA is transferring to another school where he has already been in contact with the football program and has pitched them on looking at their recruiting practices from an analytics standpoint.  The football program has thus far been receptive to working with OA.

There are a number of important takeaways from OA’s story.  First, he has an analytical/obsessed mind.  He’s like a 1,000 kids who taught themselves to play poker.  He’s basically exactly the kind of person who should be involved in football.  Second, even though OA is obsessed, he still has to ask permission under the current system.  He’s waving a flag saying “Please, please let me in.” which is what any employer should want from a prospective employee.  But because he wants to get into the world of football, he faces long odds to break in.  There are a lot of people who want to work in football, and football organizations are used to turning away people just like OA.

The disruption to football learning that I describe above is the game changer for people like OA.  It’s a game changer because OA could take a number of online courses so that when he tries to break into the world of football, and the organization is trying to decide whether to hire OA, or perhaps their recently graduated safety, who was one of their smarter players, OA is now on close to equal footing with the recent grad in terms of football knowledge.

We’re running a little long here, so let me bring this back full circle.  Mark Zuckerberg learned to program under a system where he could get good on his own.  By the time he launched Facebook he was already an extremely accomplished programmer.  Innovations like Udemy mean that people who might aspire to become experts in the game of football could also benefit from learning and getting good on their own.  When (not if) that happens, the NFL will see a game changing disruption.  The change to the teaching/learning paradigm of football means that the NFL could grow its talent pool by orders of magnitude and select the best minds, not just the best minds conditional upon also being an ex-player.

The @HeHaithMe Challenge

@HeHaithMe is a veritable force on Twitter.  He’s what would happen if you combined the gambling enthusiasm of Jimmy the Greek with the zen mental state of the Tasmanian Devil.  I’ve engaged in a few prop bets with HHM in the past year or so, losing two and then also having to pay him for winning the Mock Draft Free Roll.  So I’m down a few bucks to HHM.

HHM has pretty much been killing his MLB bets this year and while I don’t really have any interest in baseball, I am sort of mildly interested in the issue of whether individual bettors can outpick the books.  So HHM and I are going to do another prop bet.  I took the under on HHM’s record to date in MLB and he took the over.  He was 58% when we made the bet, so every point over 58%, I owe him $100, and every point under 58%, he owes me $100.  The kicker is that 75% of the winner’s take goes to charity.  So it’s going to be pretty difficult for this to be a positive EV bet for me.

In any case, wouldn’t you know it, HHM went like 5-1 the day we made the bet which got him up to 60% for the year.  Here are HHM’s picks from Saturday:

Then here is his update from Sunday:

Lucky for me, HHM did a little worse on Sunday and he’s right back around 58%.

Here was my thinking in terms of why I proposed this prop bet:

  1. We’re like more than 90 days from Week 1 and I am BORED.
  2. I think the most likely outcome is that little money money changes hands.  HHM has already picked 150 games, so there is a decent sample available.
  3. The bet is almost assymetric in favor of me.  I said almost.  If HHM were only picking games that were –110 on the betting sheet, then I think the bet would be in favor of me by a margin.  Gamblers have to pick those right at 53% just to do better than break even, so I would essentially be on the book side of the bet by taking the under on 58%.  But HHM sometimes takes games that are –125, so the bet isn’t as much in my favor as it would be if all games were –110.  But actually, you can see from above that HHM often picks games that he thinks will return the most for his money, so I’m not concerned that our small prop bet might influence his picks.  He places actual bets on these games, so his wins from picking good bets will have a larger impact for him than simply picking safe bets in order to win a small prop bet where the money is largely going to charity. 
  4. When I say assymetric what I mean is that my losses (if I lost) would likely be small, while my wins could be larger.  I would think that a 60% win rate for HHM would really be something, while a 54% win rate wouldn’t be odd at all on the downside.  So I guess I think that even while it’s likely that not a lot of money changes hands, my upside is probably larger than his.

If you know of a good charity, be sure to post it in the comments.  If I win, I might just have HHM donate to that charity.

I’m About 90% Sure This is My Last NFL Draft Post for the Year.. Ok, Maybe I’m 80% Sure.

Code and Football has been doing a series of posts on trades in the NFL draft.  You should check out the site if you haven’t before.  C&F recently linked to my Games Started draft value chart, and then followed that up with something of a rundown of recent discussions of draft trades that have been taking place on the interwebs.

As I was reading the C&F post, I remembered that I had meant to return to the issue of draft trades one more time.  Post-draft I was reading something on ESPN that sort of caught my ire.  John Clayton wrote a piece that was essentially a winners and losers column on the first round trades.  In that piece he included these lines:

Too often in past years, the Patriots got a little too cute. They’d trade a choice for a future first-rounder. They’d trade back and acquire more draft choices than they had roster spots for rookies.

And then this line:

3. St. Louis Rams: Trading out of the top six is usually a bad idea.

Really?  Why?

Maybe I’m getting too excited over a few lines in this Clayton piece, but here’s why I haven’t forgotten about it since I read it.  Clayton is a thought leader as it relates to the NFL.  Maybe he’s outside of the top five guys in terms of being thought leaders, but he’s on the list.  He gets paid to do this stuff.

Why is trading out of the top six “usually” a bad idea?  Even in the world where everybody believes in the Jimmy Johnson Chart, trades are assumed to be equal on each side based on that chart.  For Clayton to assume that trading out of the top six is a bad idea, is to say that the draft chart that transactions are based on isn’t steep enough.  He’s saying that the talent in a draft is even more weighted towards the early picks than everybody already erroneously believes it is.

Clayton also makes the assumption that the Patriots “got too cute” when in reality, they were simply executing a draft strategy that presupposes that football is a sport that starts 22 players who will incur a lot of injuries.  The Patriots were also excellent at taking advantage of the over-discounting that most teams do related to future year picks.  The Patriots were acting like bankers (or loan sharks), advancing near term picks at extremely high interest rates.

I would say about 85% of the post draft coverage I’ve read contains a “bold move” bias.  At the heart of the Bold Move Bias is the faulty idea that every move is a 50/50 proposition, but that the payoffs aren’t weighted that way.  For instance, the Bold Move Bias would break the RGIII trade down based on it either a) works out or b) doesn’t work out.  It would then assign a 50/50 probability to each potential outcome.  But then the Bold Move Bias says that if the trade does work out, the benefits will be outsized because you’ll have a franchise quarterback.  Under that view, every bold move has a positive return on investment.

The unfortunate thing is that the real world doesn’t work that way.  The Bold Move Bias as applied to a roulette wheel would say that the odds of hitting on any number are 50/50 (it either hits on the number or it doesn’t) but that the payoffs are 35/1.  The Bold Move practitioners would then say “Any time you have the chance to put $20 on a single number and potentially make $700, you have to do that.” 

That type of thinking, the kind that considers only the payoffs of the bet and not the costs or the odds, is why a lot of teams spend years in the cellar.  While the Redskins keep putting their money on a number on the roulette wheel, the Patriots, Eagles, Steelers, and Packers keep putting their money in the bank.  Every once in a while it’s reasonable to expect that the Redskins will hit on their number, but that doesn’t mean their outlook is necessarily right.  It’s just bound to happen some of the time.


What Does This Poker Bluff Have in Common With Sean Payton’s Decision to Onside Kick During the Super Bowl?

Game-Plan-CoverIn Game Plan, I spend a considerable amount of time comparing decision making in football to decision making in poker.  To me it’s a natural connection. But I could see where it might not be a natural connection to everyone.

It’s a natural connection to me because coaching a football game is a series of decisions from set scenarios in the same way that poker is a series of decisions from set scenarios.  Coaches decide whether to run or pass.  Poker players decide whether to bet, check or fold.  Coaches decide what pass play to call.  Poker players decide how much to raise, or whether to check raise, or whether to slow play.  Poker players and coaches both have to anticipate what the opponent will do.  They both have to disguise what their team wants to do.

I spend a not insignificant time in the book talking about poker phenom Tom Dwan.  Dwan is interesting to me for two reasons.  First, he’s young.  He’s a lot younger than any NFL coordinator or head coach.  He’s a lot younger than even the youngest NFL coordinator or head coach.  That goes against the notion that age is an important consideration in forming expertise.  Dwan is probably one of the top five poker players in the world.  But Dwan is also interesting to me because he learned to play online, in a way that the old school pros didn’t even view as being real poker.

The poker hand that I’ve embedded below actually reminds me in some ways of the Sean Payton decision to inside kick in Super Bowl XLIV.  Dwan bluffs two extremely accomplished players out of made hands.

To me the impressive thing is not that Dwan executes this bluff.  If you bluff enough (he does) then some of them are going to work out.  The impressive thing to me is that after the hand, a number of players around the table engage in discussion in order to try to guess whether any of the players had three twos.  Dwan actually ends up engaging in a prop bet with Doyle Brunson where Dwan bets that Peter Eastgate had a two in his hand.

The only player at the table who is younger than Dwan is Peter Eastgate, and yet Dwan’s mastery of the game shows through on a number of levels.  First, he executes a bluff that, like Sean Payton’s decision to onside kick, is not for the faint of heart.  Then, after the hand he shows that he read the hand better than a much more experienced player and that he knew where he was in the hand the entire time.

I think that the most demanding thing that a football coach does is make strategy.  I don’t think that the management side of football is particularly demanding.  If I had to guess how many people in the world are better at managing employees than the NFL’s coaches, I would say “A lot”.  Managing people isn’t something that requires that you’re in the top group of cognitive performers.  But strategy, and implementing strategy within the confines of the play clock, is something that requires elite cognitive ability.  That’s the area where football coaches are similar to poker players.

When Sean Payton came out in the second half of the Super Bowl and decided to employ the high risk, high reward strategy of onside kicking, he had numbers on his side.  But he was also making a decision that, if it went poorly, would probably stain his coaching resume forever.  But he pretty much didn’t give a shit.  He had ice in his veins in the same way that 20-something Tom Dwan had ice in his veins when he put out a $100k bet and he knew he was dominated by the other players in the hand.

For the rest of my thoughts on this topic, be sure to check out Game Plan, which at just $0.99, is like half the cost of using an ATM not owned by your bank.

Will Write Guest Posts for Food.. Fine.. No Food.. Jeez You’re a Tough Negotiator

Just wanted to mention again that I’m going to try to crank out as many guest posts as I can in the month of May.  I just recently finished up a guest post on Greg Little for and last week I wrote a post about taking punters in the third round for Big Cat Country.

I’m also working on a few posts on Value Based Drafting (for fantasy football) for some other sites as well as a post on the relative value of quarterbacks and wide receivers (for real football).

Just in Case You Didn’t Think I Had Another Punter Post in Me

OK, one last post tonight to try to put this punting thing to bed.  I’ve noticed that a lot of the comments at Big Cat Country are focusing on the idea that a good punter’s longest punts are what we should look at.  It’s not that the best case scenario shouldn’t be considered.  In fact I argued that in salary cap terms, teams should think about the potential upside from their 3rd round picks.  But the best case scenario is just one potential outcome.  There are always a range of potential outcomes.  Even the best punter in the league is going to have results sometimes that are below average.

I thought I would actually look at the best punter in the league to illustrate this fact.  Shane Lechler is the highest paid punter in the league.  He has a big leg. 

But if you’re looking at Lechler, should you focus on his longest punts, or the average of his results? 

Below is a graph that shows Lechlers kicks based on the line of scrimmage when he kicked them.  I show Lechler’s net punt, and I also show the expected net punt from each yard line on the field.  As you get closer to the other team’s end zone, the expectation for a punt goes down in terms of net.


Lechler is actually better than average if you take his net punt and then subtract out what the average punter gets from each yard line on the field.  But he’s only 2.5 yards per punt better than the average expectation.  This analysis adjusts for field position in the way that the BCC commenters are saying is important.  Lechler also isn’t significantly better when he’s closer to his own end zone.  He’s still just about 2.5 yards per kick better than average.  So in the places where the BCC commenters are saying Lechler can “flip the field”, Lechler is still just 2.5 yards per punt better in terms of net punt than what we would expect.

The average is the central tendency.  For every punt that Lechler kicks that might go 70 yards, he is balancing that out with a shorter kick so that his average is close to what the average kicker yields (although slightly above).

Psychologist Daniel Kahneman has done a number of experiments looking at whether humans can make good judgments about the most likely outcome.  It turns out that we can’t.  We often focus on the best case scenario.  We’re drawn to the idea that the best punter we can find might be able to outkick the next guy by 10 yards per kick.  But that’s not realistic.  It’s not realistic for the same reason that municipal projects never finish on time or on budget.  The projects are always bid with the best case scenario in mind.  Just like the Jaguars probably do feel like they took a guy who can net them an average 50 yards per game in field position.  But a quick review of historical outcomes shows that those expectations aren’t reasonable.

Fooled by Randomness in the Punting Game?

For my guest post at Big Cat Country I looked at whether it made any sense to take a punter in the third round of the NFL draft.  Go check out that post if you’re interested in punters at all.  In the comments of that post a number of people suggested that I had ignored the value that a punter might have in pinning the other team deep in it’s own territory.

I thought I would look at that issue.  First I had to create a formula to figure out whether a punt should or should not be a touchback.  I used results from 2000-2010 to create this graph that breaks down punts by likelihood of becoming a touchback, depending on field position.


I can then use the formula from that graph to analyze actual punts and see whether punters have repeatable ability to avoid touchbacks (when controlled for field position).  When I do that, I am basically calculating Expected Touchbacks vs. Actual Touchbacks.  Doing that, I get the following graph.


The trend line explains about 60% of the variance in touchback results, which is to say that field position explains about 60% of the variance in touchback results.  But that’s not 100% either.  So is the actual punter responsible for causing or preventing the touchbacks that can’t be explained by field position, or is it randomness at play?

It’s probably randomness.  The graph below shows an X, Y scatter where prior ability to prevent touchbacks is the independent variable.  It doesn’t have any explanatory power over future ability to prevent touchbacks.  Just because a punter may have had less touchbacks than you would expect based on field position in the past, doesn’t mean that will continue.


This is something of a “fooled by randomness” issue.  Sometimes it’s easy to mistake randomness for skill.  Some might remember the performance of San Diego punter Mike Scifres in the 2009 playoffs.  Scifres was lauded for his performance against the Colts.  Below is an account from the game:

Scifres, passed over again this year in Pro Bowl voting, booted the ball six times last night for a 51.7 net average – an NFL playoff record for a punter with five or more punts. All six were downed inside the 20-yard line, also an NFL playoff record. Five times Scifres pinned the Colts inside 11 yards, and Indy had 6 yards in returns. Scifres had one booming drive of 67 yards.

But over his career, Scifres has about as many touchbacks as you would expect, based on field position.  Over the long term he hasn’t shown any increased ability to pin the other team deep and avoid touchbacks.  I have Scifres calculated for 44.05 “Expected Touchbacks” and he has 44 “Actual Touchbacks”.  The skill he showed in the playoff game against the Colts may have just been randomness.

In all of this analysis, I only found one punter whose results looked like they deviated from expectation significantly.  That was Shane Lechler, who causes touchbacks a lot more often than should be expected.  Based on field position you would expect that he would have caused about 75 touchbacks in his career.  He’s actually caused 129.  The interesting thing is that Lechler is really bad at that part of the game and yet he’s the highest paid punter in the league.

A Quick Rundown of the First Round Trades

If you’re a fan of watching the ways that markets work, then the NFL draft pick trade market can be fun to watch.  I sort of got interested in the topic when it occurred to me that the way that the NFL values picks is not probably very efficient based on a reasonable expectation of the distribution of human abilities.  Basically the NFL works off a trade chart that assumes the existence of super human individuals.  I think it’s unlikely that those superhuman individuals actually exist and I would argue that actual player results back me up.

I thought it might be fun to look at the draft day trades and compare the trades on the basis of the Jimmy Johnson Chart and also my chart that focuses on how many games started you can expect to get out of each pick.  As a refresher, here’s my draft pick value chart shown versus the Jimmy Johnson Chart.


In putting this together I relied on this recounting of the trades. 

Here we go!

Browns move up to No. 3: Wanting to secure their top choice, the Cleveland Browns moved to the No. 3 pick in a deal with the Minnesota Vikings. Minnesota acquired the No. 4 pick, plus three additional draft choices ( a fourth-round pick-No. 118 overall, a fifth-round pick-No. 139 overall and a seventh-round pick-No. 211 overall). The Browns moved up one spot to select Alabama running back Trent Richardson, while the Vikings grabbed Southern California offensive tackle Ryan Kalil with the No. 4 pick.

The table below breaks down the value of the picks exchanged.  image

Note that the JJ Chart is denominated in points, while my chart is denominated in “Games Started”.  So you can expect to get about 85 career games started out of the fourth overall pick.  The JJ Chart says that Cleveland won the trade by a pretty wide margin, and the FD Chart says that Minnesota won the trade by a pretty wide margin.  Note that if I were to include position specific data, Minnesota would have blown Cleveland out of the water on this one.  Offensive linemen tend to start a significant number of games more than a running back.  Then if you consider salary cap issues and the fact that Minnesota is saving a lot of money on a left tackle and Cleveland is saving less money on a running back, it gets even worse for Cleveland.

Moving on.  From the trade summary on MassLive.

Jaguars go up to get Blackmon: The Jacksonville Jaguars traded up to the No. 5 overall pick to select Oklahoma State wide receiver Justin Blackmon. Tampa Bay acquired the No. 7 overall pick and pick up a fourth-round pick from Jacksonville (No. 101 overall). With the No. 7 pick, Tampa Bay selected Alabama safety Mark Barron.


Again, the Jimmy Johnson Chart says that the team trading up won, my chart says that the team trading down won.  My chart says that the expectation for the games started difference for the 5th and 7th picks isn’t that great.  However, if you look at the actual picks, I think you could make the case that Jacksonville did alright with this trade.

Moving on.  From the trade summary on MassLive.

Cowboys jump up eight spots to No. 6: The Dallas Cowboys have moved up eight spots to the No. 6 pick in a deal with the St. Louis Rams to select LSU cornerback Morris Claiborne. The Rams acquired the No. 14 pick (which St. Louis used to selected LSU defensive tackle Michael Brockers) as well as a second-round pick (the No. 45 pick) from the Cowboys.


The JJ Chart says this one is about even.  My chart says that the team trading down (STL) won, and by a sizable margin.  The 2nd round pick that the Cowboys gave up is worth quite a bit.  In terms of the actual players, I do think there’s something to be said for getting the best corner in the draft due to positional importance.

Moving on.  From the trade summary on MassLive.

Eagles move up to No. 12 for defensive tackle: The Philadelphia Eagles traded up three spots to acquire the No. 12 pick from the Seattle Seahawks to select Mississippi State defensive tackle Fletcher Cox. Seattle picked up the No. 15 pick (which Seattle used on West Virginia outside linebacker Bruce Irvin) plus two additional picks (a fourth-round pick-No. 114 overall and a sixth-round pick-No. 172 overall)


From a pick value standpoint, my chart says that SEA won.  But from an actual pick standpoint, it doesn’t look like they really capitalized on the opportunity.  PHI got Fletcher Cox, who many considered to be a top 10 quality player.  To figure the pick value on this one, it might actually be appropriate to think about Cox in terms of what his inherent value is, not what other teams assigned to him.

Moving on.  From the trade summary on MassLive.

Bengals deal out of No. 21 pick: After taking Alabama cornerback Dre Kirkpatrick with the No. 17 pick, Cincinnati traded out of the No. 21 pick in a deal with New England. The Patriots gave up the No. 27 pick as well as a third-round pick (No. 93 overall) to move up six spots. New England selected Syracuse defensive end Chandler Jones with the No. 21 pick. Cincinnati took Wisconsin guard Kevin Zeitler with the No. 27 pick.


The only comment I have here is that we were talking last night about CIN and how they seem to be increasingly making smarter decisions.  I’m not even saying that they sharked NE here, but between their draft picks last year, the Carson Palmer trade, and then this trade, they are increasingly operating like the smarter franchises.

Moving on.  From the trade summary on MassLive.

Patriots move up again: The New England Patriots moved up for the second time tonight when they acquired the No. 25 pick from the Denver Broncos. Denver acquired the No. 31 pick and a fourth-round pick (No. 126 overall). Denver later dealt those picks to Tampa Bay. The Patriots went defense once again with their draft selection, selecting Alabama linebacker Dont’a Hightower.


I think there’s actually an important point to be made here.  When NE trades down, they often are just swapping picks with a 5 or 6 pick difference, and they will sometimes get a future year number one pick in the deal.  When they do that, they are taking advantage of teams discounting future year picks too heavily.  But even when they trade up, they do in with terms that no other team gets.  In terms of Games Started (my chart metric), this is the first trade where the team trading up got the better end of the deal.

Moving on.  From the trade summary on MassLive.

Vikings jump back into first round:The Minnesota Vikings picked up a second first-round pick in a deal with the Baltimore Ravens. Minnesota acquired the No. 29 pick and in exchange the Ravens received a second-round pick (No. 35 overall) and a fourth-round pick (No. 98 overall). The Vikings selected Notre Dame safety Harrison Smith with the No. 29 pick.

I really have no comment on this one.


Moving on.  From the trade summary on MassLive.

Broncos move back again: The Denver Broncos, who traded out of the No. 25 pick, also traded out of the No. 31 pick in a deal with the Tampa Bay Buccaneers. Denver acquires a second-round pick (No. 36 overall) and a fourth-round pick (No. 101 overall). Tampa Bay picks up the No. 31 pick, to select Boise State running back Doug Martin and a fourth-round pick (No. 126 overall).

No comment here except that Tampa Bay moved up in order to select a running back.


Visualizing the Results from the Mock Draft Contest

I figured I would bang out a few posts with some graphs based on the results of the mock draft contest.  Here’s a graph that shows the top 20 players from the contest based on average pick number that they were selected at (not the total times they were drafted).

You can actually see some tier groupings from this graph.  You’ve got the Luck/RGIII tier, the Kalil/TRich/Claiborne tier, and so on.


Final Pre-Draft WR Rankings

Translating wide receiver success from the college game to the pro game is sort of an analyst’s dream, primarily because there is so much work left to do.  A model based only on draft position explains some amount of wide receiver success.  If you add in some variables that I use, you can explain a little bit more of WR success.  But even then you’re still a long ways from having a model that gets every receiver right.  So there is still a lot of work to do.  It’s a great challenge.

Below I show two different groups of wide receiver rankings.  The only difference between the two rankings is that in one group I have taken out the impact of draft order.  In the first group I show the impact of draft order because it does matter.  Draft order is going to dictate opportunities at least, and it’s also likely that draft order reflects some amount of player evaluation that can’t be covered by the measures that I use.

Without further ado, here are my WR ranks based on some simple assumptions that I use for draft order.  The variable that I solve for is a wide receiver’s percent of pro team fantasy points.  The reason I do that is because receivers can end up in different situations.  For instance, Larry Fitzgerald, Andre Johnson, and Calvin Johnson all varied in the fantasy points that they compiled, but in their first three years in the league they all caught about 35% of their teams’ fantasy points.   So basically my dependent variable has been adjusted for pro team passing offense.


Some notes:

  • Market Share of YDs = the player’s share of college team passing yards
  • For my draft pick assumptions I just tried to be accurate within about 5 picks for the first round guys and within about 20 picks for the later round guys.
  • Proj. = the player’s projected share of pro team fantasy points.  To get a sense as to how this year’s crop stacks up, consider that Calvin Johnson would have been projected to produce 35% of his pro team’s production.  My top forecast player this year is Justin Blackmon, who is roughly projected in the Hakeem Nicks/Greg Jennings/Roddy White range.  Which is to say that he is a good prospect, but located well below the Calvin Johnson/Larry Fitzgerald level of prospects.
  • Even if I assume that Michael Floyd is going at around the 10 pick, I still have him rated as the fourth WR.  Floyd could certainly prove me wrong.  He has some conflicting signals in that his 2010 season yielded more touchdowns.  However, I only count last season of college production because I’ve found that to be the best predictor.
  • I don’t really know how accurate the predictions for Reuben Randle and Alshon Jeffery will be.  I missed by a pretty wide margin on Julio Jones last year and I think part of that might be related to having to play SEC cornerbacks.  But even that shouldn’t affect the % of team yards that they caught.  On that measure Randle looks very good at 39% of LSU’s yards.  Jeffery doesn’t look so good on that measure.
  • Brian Quick’s receiving numbers are actually made up.  I didn’t want to put his ASU numbers in there and I haven’t looked at small school receivers (Garcon/Colston/Cruz) enough to really try to translate small school receiver success.  But I also didn’t want to leave him off the list.  So I made shit up.


If we want to see what the list looks like without the impact of draft pick position (so basically just my raw ranking), it would look like this:


Some notes:

  • Jordan White probably isn’t the best WR in this class, but he’s also probably a lot better than teams will give him credit for.  Consider White’s games against Big 10 opponents: Illinois – 132/1, Purdue – 265/1, Michigan – 119/0.  Those are pretty impressive.  I could see him being a sort of Austin Collie-type guy.  Collie also flashed pre-draft with huge touchdown numbers.  I would seriously keep an eye out for White over the next couple of years and look for him to end up in a situation where he’s the 3rd WR who might see soft coverage. 
  • Primarily the difference between the two groups of rankings comes down to a player’s share of college team yards.  That’s a measure that is often overlooked by teams when they draft WRs.  My analysis shows that % of college team yards has a low correlation with a WR’s draft slot, but a higher correlation with pro success.
  • You might notice that all of the WRs have a higher “Proj” in this list.  That’s because when I take out the impact of draft order, I’m removing a variable that subtracts from each player’s projection.  Every 25 or so draft picks that a player drops subtracts about 1% from the projection of their share of pro team fantasy points.
  • You might look at the difference between the two lists as the value difference.  You’ll be able to get the guys rated highly in the 2nd list for cheaper than you’ll be able to get the guys rated highly in the first list.
  • I’ve been working on an adjustment for QB quality that I actually left out of this iteration.  However, that adjustment would raise the ranking of Mohamed Sanu for instance.  Sanu played in an offense that completed about 53% of its passes, which no doubt hamstrings receivers in that offense.  But the flipside of that is that when you start adjusting Sanu (and Stephen Hill) upwards, you have to adjust Kendall Wright downwards.  Anyway, I’m still working on that.