Football Frontier

In the second half of this pre-draft analysis, I’ll write about the “Combine Numbers” associated with wide receivers. As I noted yesterday, the bell cow of the Combine, the 40-yard dash, is random noise when it comes to predicting whether a receiver will be a success in the NFL.

Since some NFL athletes can run almost as fast as Olympic track stars, the score that relates most to track – the 40-yard dash – receives significant attention. However, the receiver with the fastest Combine 40 time in the period I studied, Jacoby Ford (4.22 seconds) in 2010, was selected in the fourth round by Oakland and played only four years in the NFL, accumulating 848 receiving yards. He showed some game-breaking return skills, but fumbled the ball too frequently. This is a decent, normal amount of production for a fourth-round wide receiver.

Darrius Heyward-Bey came in second at 4.25 seconds in 2009. He has stuck around for nine seasons. He accumulated 2,071 yards in his first four years, well under what the Raiders (again) would have hoped for from the seventh overall pick, and another 817 yards since. While he’s signed for just above the veteran minimum, I would not be surprised if he retires this off-season.

Up third is Jerome Mathis (4.26), drafted in the fourth round by Houston in 2005. Like Ford, he made an impact as a kick returner as a rookie, but had an issue with fumbling. He had 80 total receiving yards in a three-year career.

You get the picture. Some speedsters, like Mike Wallace, Julio Jones and Brandin Cooks, are very productive in the league. But just being able to run really, really fast is fairly meaningless on its own.

The other tests include a “bench press,” in which an athlete repeatedly lifts a 225-pound barbell until he feels he must stop. There’s a vertical leap from a standing position, a broad jump from a standing position, a 20-yard “shuttle run,” which measures the ability to accelerate and make turns quickly, a 60-yard version of the shuttle drill and a “three-cone” drill, which measures many of the same agility skills.

Players are also measured, with their height, hand size and arm length recorded within an eighth of an inch. They are also given an intelligence test, subjected to any of a number of medical examinations and (sometimes controversially these days) available for interviews from coaches whose goals are to surprise athletes with unusual questions.

Experts can make a lot of money by taking all this information and coming up with a formula for success. Unfortunately, no one has done that. Since statistical software these days is quite sophisticated, data miners can come up with a formula that fits a large number of past performers. But is that predictive? Even with my sample size of 444 drafted wide receivers, I suspect it’s limited.

A wide receiver needs many skills, and anyone who saw Jerry Rice play could tell you that his speed, while hardly close to the best on the field, was enough to get the job done. Rice reportedly ran a 4.59 in the 40-yard dash. He was drafted 16th overall in 1985. Larry Fitzgerald, a certain Hall-of-Famer, ran a 4.48 – same as A.J. Green, who is headed to Canton himself if he can maintain his production level a few more years.

What do they have in common that other receivers don’t? Fitzgerald was the third overall pick and Green was the fourth in their respective years. Scouts knew about them. Fitzgerald didn’t take part in most of the Combine events, so scouts likely didn’t have that type of comparative information down to the hundredths of seconds. Green did, and while his numbers were good, you can find better. What makes Green one of the best in his generation and Heyward-Bay, who had slightly better, but similar Combine numbers, disappointing?

I gave the 272 drafted players who ran most of the drills in the study a similarity score to Green’s. Strangely enough, one of the lowest scores came from Chad Johnson, who had an excellent career mostly with Cincinnati as well. But I look at his cone-drill time of 7.51 and his reputation as a bit of an eccentric (well, aren’t all good receivers a bit eccentric?) and I wonder if he simply didn’t feel like running the drill. That’s the worst time in that drill among wide receivers, and more indicative of a defensive lineman or tight end.

But some receivers had almost identical scores to Green’s, and some, like Michael Jenkins (2004) and Sammy Watkins (2014) went in the first round. And some, like Robert Herron (2014) and Pierre Garcon (2008) went in the sixth. Herron barely played and Garcon is one of the better 6th-round picks in recent years. Watkins was average for a number-four overall receiver pick and Jenkins was a bit of a disappointment for a late first-round pick.

There’s no magic in Green’s combination of results. In fact, the 100 receivers least similar to Green were a little bit better than the 100 receivers closest to Green. And the slightly smaller group of receivers between those groups averaged a tiny bit better than both groups. As for draft position, the top 100 averaged being picked 128th, the bottom 100 135th and the middle 71 126th.

Am I saying the Combine times don’t matter? Not even enough to be worth tracking? No. There are slight correlations between times and performance. Vertical leap seems to be the most important number, though it only correlates 15% with output. A receiver’s weight actually matters slightly more (18%).

The shuttle drill and 3-cone drill are difficult to use, because the top scouted players are far more likely to skip these drills. Why risk draft position on a poor performance (unless you’re Chad Johnson)?

I suspect that weight matters because it takes a certain amount of strength to deal with good bump-and-run coverage. And some of the low-round or undrafted receivers who have done wonders (going back to Bill Belichick’s penchant for finding Wes Welkers and Danny Amendolas) work out of the slot much of the time. If you want to find an exceptional athlete who was a little smaller but found enormous success out of the slot early in his career, Randall Cobb was picked by Green Bay at the end of the second round in 2011. More recently, the Chiefs have gotten a lot out of tiny Tyreek Hill, a fifth-rounder with elite speed, in many different roles.

That’s a lot to digest. What about a measure of simple competency? Take the 272 receivers with measurements in every category and look at those who scored in the top half in all of them. So that would mean someone who weighed more than 200 pounds, ran a 4.48 or better in the 40, a 4.20 or better in the shuttle, a 6.92 or better in the cone drill, leaped at least 36 inches and had a broad jump of at least 121 inches.

Who are these extraordinary people? The list is just 12 of the 272. Surely these are all players the scouts would agree are superstars in the making. Not so fast. There were two first-rounders (Heyward-Bey and Javon Walker, who ended up being slightly disappointing his first four years), four second-rounders – two who greatly exceeded expectations (Alshon Jeffery and Torrey Smith) and two who busted (Tyrone Calico and Chad Jackson). There was a fourth-round pick, Keenan Burton in 2008, who didn’t fare very well. In the fifth round, Roy Hall in 2007 had nine career receiving yards. The pair of sixth-rounders – Garcon and Jeff Webb (2006) were very solid picks. And the two seventh rounders, Marcus Maxwell in 2005 and Jeff Janis in 2014, were forgettable.

More recently, Kevin White (the seventh overall pick in 2015), fits this category and is looking like one of the bigger busts in recent history. DeVante Parker, also a first-rounder in 2015, has been OK. The only player to fit the bill last year was second-rounder Zay Jones, who hasn’t done much. And, notably, undrafted Tyrell Williams (2015) has been quite a find for the Chargers.

At any rate, the simple every-category competency measure is a poor fit for the data. Probably because smaller players are often more agile. Indeed, weight correlates 18% negatively with the shuttle drill and 23% negatively with the 40-yard dash. The broad jump is the only drill where weight seems to help a little.

So I made one last try – finding players who are in the top third by weight and top third in vertical leap. That’s 209 pounds and 37 inches. That gave me a set of 51 drafted receivers from 2001-14. What was their average performance? As a group, they outperformed their draft position by about 5%. Of the 15 first-rounders, eight significantly exceeded expectations and only three were very disappointing. This group also contains some breakout lower-round performers like Brandon Marshall, Josh Morgan, David Givens and Marques Colston. More recently, it identifies second-rounder Devin Funchess (2015), though there hasn’t been a first-rounder meeting the criteria the last three years.

I’m content to stop there. Being invited to the Combine means scouts think a player has a potential NFL career. That implies a minimum competency. This 209/37 test looks decent for identifying additional potential. And I think data mining the rest is only going to identify stupid coincidences in the remaining data set.

Who meets the criteria in the upcoming draft? D.J. Moore, Tre’Quan Smith, Dylan Cantrell, Allen Lazard, Jester Weah, Justin Watson and Reggie Bonnafon, among those with draftable grades. Of those players, Cantrell has unusually good scores in other categories, indicating solid athletic skills. His 4.59 40-yard dash, however, might concern teams and despite high vertical leap and broad jump scores, scouts say he is not an explosive athlete. He has a third-day draft grade. I’d be interested to see if he sneaks into the second round, though.

Just to wrap this up… if you play Front Office Football, you know the Combine plays a much different role in the game. Individual measurable skills correspond to specific potential ratings. This is obviously not realistic. But the alternative is to place more emphasis on the scouting abilities on your staff – with wider variety between teams. This would essentially render the Combine scores within FOF useless.

I think that makes for a less enjoyable game. You’d be drafting purely on scouted ratings and focusing on where your expectations differ from league-wide grades. At that point, it’s just guesswork with no skill at all. Or maybe I’d add something to indicate where your staff member has a particular insight. But if you played a hunch on your own with a Combine number, you’d be making a mistake.

I may change this whole system in the future, but for now, I think it’s a good idea to stick with a simulated Combine that means a lot more than the real-life version. Watching tape and making scouting opinions is full-time work for the few who really do this well. Just as messing with the intricacies of player contracts and agents is full-time work for a GM. That doesn’t make for an enjoyable game. My philosophy is that the game engine itself should be as realistic as I can make it. But the staff experience – your job as the team’s hybrid GM/coach/owner/etc – has to be enjoyable.

The NFL Draft begins Thursday, and I’ve been taking some time the last few days to study how the draft works. In particular, I’m looking at wide receivers and Combine scores.

This first post, however, will focus on performance alone.

The Amateur Draft is an important piece of team building. Perhaps it’s the most important piece, because teams receive on average eight new draft picks every year and only 53 players are on the team during the season. Trades are still very rare in the NFL, especially involving young players.

When you draft a player, he’s yours for four years. After that, he’s more-or-less a free agent and gets paid based on a combination of experience, expertise and the role he’s expected to play. There are very few important bargains in free agency. You fill holes in free agency; you don’t build a team. And when signing contract extensions after that fourth year, you have to pay the market rate.

To analyze the effectiveness of a draft pick under a hard salary cap, therefore, you should limit your analysis to those first four years. In other sports you can factor in the market and the style of a particular owner. Even the NBA has its “Bird exception” and its “Rose rule” and other odds and ends to turn the cap into something a little different. But in the NFL, the draft gives you that young base that shapes your team.

I chose wide receivers to examine because there’s a simple statistic that defines them quite effectively – at least when applied to hundreds of players over a long period… receiving yards. While you can dig deeper with receivers – adding points for first downs earned in passing situations and for touchdowns – simple accumulated receiving yards are quite meaningful.

The scope of this study is 2001-2014. That’s the last 14 years of receivers who have reached four years of experience. 2001 is a good place to begin a study of the most recent passing emphasis within the game. It was the last year before the eight-division, 32-team format. There were 444 wide receivers drafted during that period – more-or-less evenly dispersed in the seven rounds of the draft with a slight increase in the seventh round.

For each player, I tried to find Combine data (and where that wasn’t available, data from the respective Pro Day). I also tracked total receiving yardage through year four. Eighty-five of these players never caught an NFL pass. The highest of these 0-yard picks (and only second-rounder) was Dexter Jackson, 58th overall in 2008, a shorter (5094) small-college star who ran a 4.37 40-yard dash. Tampa Bay tried him in a few games as a return specialist as a rookie, then waived him late in training camp the following year.

Sixteen of the 444 wide receivers exceeded 4,000 receiving yards in their first four years – all the way up to A.J. Green of Cincinnati with 4,874. Eight of those 16 were first-rounders (five in the top eight picks, including Green), four seconds, two thirds, a fourth (Denver’s Brandon Marshall) and a seventh (Marques Colston of New Orleans). Both Marshall and Colston were drafted in 2006.

In fact, 27 of the 43 receivers who have exceeded 3,000 receiving yards were first-round picks. Part of this is self-fulfilling – you draft a wideout in the first round and you had better feature him in the offense right from the start. But scouting receivers isn’t as hard as scouting quarterbacks and the learning curve isn’t quite as steep, so coaches have a decent idea of who will and who won’t thrive in the NFL.

Average receiving yards by round drafted: 1st – 2,708, 2nd – 1,725, 3rd – 1,139, 4th – 813, 5th – 510, 6th – 340, 7th – 293. I broke down the first round, and while the average tails off noticeably in the last few picks of the round, the sample sizes get too small to provide significant delineation.

I also took a look at how individual teams draft. I did this through assigning a score to every pick based on the expected four-year yardage totals for each round. Every team drafted between 8 and 19 wide receivers during the 2001-14 study period (average of about 14).

The “best” team at drafting wide receivers was Indianapolis by a wide margin. Part of that is likely that the Colts could probably get away with lower receiver skill sets because Peyton Manning was so remarkable. Part was having amazing longevity in the form of Marvin Harrison (before the study) and Reggie Wayne. So Indianapolis was also one of two teams with eight receivers drafted (San Diego – average in quality of picks – was also at eight). Philadelphia, Chicago, Arizona, Jacksonville, Cincinnati, Pittsburgh and Dallas made up the rest of the top quartile. Minnesota, Tennessee, Tampa Bay, Seattle, St. Louis, Baltimore, Kansas City and (dead last) New England made up the bottom quartile.

Why New England? Perhaps Bill Belichick just isn’t great at evaluating college receivers. He’s a master at seeing low-round (or even undrafted) guys on other rosters who can help, but when he reaches for draftees, it’s not pretty. His highest receiver picks at New England were Chad Jackson (36th overall, second round) in 2006 and Bethel Johnson (45th overall, second round) in 2003. Combined, they didn’t reach 1,000 yards in their entire careers. New England is also the only team in the NFL that has not drafted a wide receiver in the first round in this millennium (Minnesota and Detroit are tied for first at four). You’d have to go back to 1994 and Michigan’s Derrick Alexander to find a Belichick-led team and a first-round wide receiver.

It’s not easy finding effective wide receivers, though. There is an element of lottery to it. Just look at 2008. In addition to Dexter Jackson, nine other wideouts were chosen in the second round. Four others were below 500 yards in their first four years. Four were above average for the round, including Donnie Avery, Eddie Royal, Jordy Nelson and DeSean Jackson. Of the group, only DeSean Jackson (4,085) was above 2,600 yards in four years. No receivers were taken in the first round. But second to Jackson in four-year yardage from that entire 2008 class was Davone Bess, who came out of Hawaii, ran a 4.60 40-yard dash, stood 5097 in height and went undrafted. Scouts seemed uncertain whether Hawaii’s incredible passing game in those days worked because of quarterback Colt Brennan (a sixth-round pick in 2008 who never played an NFL game) or Bess.

To wrap up this first post, I’ll talk about Combine numbers. I’ve tried to accumulate as much information as possible about the Combine tests for wide receivers. However, not all prospects run all of the drills and some information comes from Pro Days and some from the NFL Combine itself. I don’t want to throw out information, but a Pro Day might have more favorable (or less favorable) conditions. In addition, not running a test is hardly evenly distributed. Expected top picks are more likely to skip the tests, not wanting to jeopardize their standing with a poor result. And there’s less information about players unlikely to be drafted.

I have a 40-yard time on 441 of the 444 drafted players, ranging from 4.22 (Jacoby Ford, 2010, a fourth-rounder with average fourth-round production through four years) to 4.78 (Jonathan Smith, a seventh-rounder who accumulated 77 career receiving yards). There’s vertical leap data for 372 players, broad jump for 334, a shuttle drill for 297 and a three-cone drill for 288. In addition, I have height for every player and hand size and arm length for most.

I will start out by saying that 40-yard dash time hardly correlates at all with four-year receiving yard totals, despite a positive 16% correlation with draft position. That’s what we’re up against with any study of Combine numbers. More soon.

Jimmy Garoppolo, heading into unrestricted free agency after a half-season rental in San Francisco, just signed a $137.5 million deal over five years with the 49ers.

Think about that. A quarterback with seven games’ worth of experience – 183 career completions – is being rewarded at the astonishing rate of three quarters of a million dollars for every past successful throw.

Now it’s not that simple. Even at fair market rate, the Patriots wouldn’t reward Tom Brady with $5 billion for his 6500-plus completions. It’s about future expectations, and Garoppolo is expected to be a franchise quarterback. So $27.5 million per year is the new normal. Which means the next time someone like Russell Wilson or Aaron Rodgers hits a contract year, agents will be talking about $35 million, hoping to settle above $30 million. And they will, soon enough.

The 2018 NFL salary cap will be about $178 million. Let’s take a look at some ratios.

The NFL rookie minimum salary will be $480,000, which is 0.27% of the salary cap. Garoppolo’s average salary is 15.45% of the salary cap. The ratio between those numbers is 57.3.

Now, let’s go back ten years. I’ll use approximations, but the idea is the same.

In 2008, the rookie minimum salary was $295,000 and the salary cap was $116 million. The crazy-high quarterback contract belonged to Carson Palmer, who was expected to make $67 million over the next five years – or $13.4 million per year.

In 2008 the rookie minimum was 0.25% of the salary cap – that’s consistent. But Palmer’s contract was 11.55% of the salary cap and you could obtain 45.4 undrafted rookies for Palmer’s cap compensation.

What’s going on here? This is an increasing effect of the 2011 renegotiated CBA between the NFL and the NFLPA. It addressed a serious problem – that contracts for first-round picks were getting out of hand and causing the value of top picks to decline. But it created a new problem – that contracts for first-round picks who end up being terrific players eat up far less cap room.

This affected the delicate balance between being able to put together a reasonable roster under the cap without discouraging teams from wanting higher picks. It’s this balance that leads to competitive balance in the NFL. Well run teams are rewarded. Owners willing to push the limits of the cap are rewarded. But no one gains from tanking and bad teams get value from picking high in round one.

I’m not claiming the NFL was broken or is broken. It obviously isn’t. But because of savings from the rookie cap, a quarterback with seven games of experience is getting a contract that would have been impossible for any player ten years ago. And we’re seeing a race developing for franchise quarterback salaries. Garoppolo’s 15.45% means someone will get 18% in the next big renegotiation.

Will that cause harm? Probably not. What will happen is that when the CBA is next renegotiated, there will be pressure to increase the cap as there’s always more attention on the highest contracts and bringing more players up to that standard is impossible without increasing the cap. Most people just see the $27.5 million average, not what that means for cap management itself.

And in this case, San Francisco is willing to channel its savings from the rookie cap adjustment into Garoppolo. Since franchise quarterbacks are hard to find these days, it may end up being a good decision. But there will be more holdouts in the near future as agents want their premier players getting a bigger share of those rookie cap savings.

The balance is changing right now, and things will get a little uncomfortable for the next year or so as teams adjust. But for the last three years – after the crazy-high pre-2011 first-round bonuses completely left the league – there’s been a period of time when it hasn’t hurt to stay under the cap. So ultimately, the Garoppolo deal and the current quarterback race is probably a good thing for the NFL.

And in San Francisco, as long as Garoppolo is as good as he’s looked in his first seven games…

From time to time, sports fans discuss overtime.

In baseball, extra innings are easily understood and implemented. In a sport filled with tradition (except for the one bizarre exception that MLB’s two conferences play with entirely different lineup structures and rules), wrinkles like starting the 12th inning with a runner on second base don’t get a lot of support.

Basketball has so much scoring that continually adding five-minute tie-breaking overtime periods has never been controversial. Double overtime is rare enough that alterations don’t generate much support.

Hockey has tweaked its rules the most. It’s hard to remember how many players play for how many minutes and how a shootout works from season to season in the NHL. There are even alterations in the overtime rules that make calculating the standings less than intuitive. Thankfully, so many teams reach the playoffs after a marathon 82-game season that casual fans are content to trust the system and wait until playoff hockey in April when everything is sudden death and six-a-side and shootouts are relegated to a distant and vaguely unpleasant memory.

Soccer embraces the playoff shootout concept because games can remain 0-0 seemingly longer than a cricket test match and overtime is largely absent from the regular season. While periodically (every four years, with the World Cup), interest spikes in the US, this might be part of the reason MLS is a distant fifth in the US. While the “Big Four” North American sports leagues are 1st, 2nd, 3rd and 5th in the world in annual revenue, the MLS is 14th – seventh among world soccer leagues. Whatever the reason, we view sports very differently from the rest of the world.

And football is our sport. The NFL is the world’s largest league in terms of revenue – about $13 billion per year. It also supports more franchises – 32 – than any major league in the world. The Super Bowl is the world’s largest single-day sporting event.

But how does football handle overtime? That’s a tough one. College football has an alternating possession rule which somewhat resembles a shootout. The NFL allowed ties in the regular season for a long time, then implemented a single overtime period, then has tweaked that rule a couple of times in the last decade. Playoff games are not limited by time, though there’s so much scoring in the game that this has never been a problem.

Recapping the rules: in 1974, the NFL added a 15-minute sudden death overtime period for regular-season games. Overtime starts with a coin toss and a kickoff. If the game is still tied after 15 minutes, that’s the final result.

People complained the coin toss was too arbitrary and the risk of scoring quickly too great, so in 2010 for playoff games, NFL teams were guaranteed a possession in overtime unless the team receiving the initial kickoff scored a touchdown. In 2012, that exception was extended to the regular season. This year, the overtime period was shortened to 10 minutes for regular-season games.

To make matters even more complicated, in 2011 the NFL moved the kickoff from the 30-yard-line to the 35 and in 2016 touchbacks after kickoffs were moved from the 20-yard-line to the 25. These are both changes that affect overtime.

That’s an awful lot of rule churn for a league that’s number one in the world by a large margin.

Does it matter? I looked in depth at the 324 overtime games in the NFL regular season and playoffs over the last 20 seasons. In terms of studying the game, I like to consider 1978 and the modifications to passing rules the birth of the modern game and 1998, when I started work on Front Office Football and Peyton Manning entered the league, as the beginning of the time period when studying play-by-play can generate results relevant to today’s game (though it’s always best, in my opinion, to focus on the most recent five years of results).

Here are some of my findings:

Since 1998, 6.0% of regular-season games have gone to overtime and 9.1% of playoff games have gone to overtime. The difference doesn’t surprise me since playoff games are limited to the top twelve teams. The average difference in skill level between teams should be lower in the playoffs.

One factor I considered is momentum in games. Announcers and fans like to talk about momentum being significant. Some of that might be more effective strategy changes during the course of games and some of it might be the mental attitude of players or maybe even the results of injuries.

I’ve often wondered if I should add a momentum factor to Front Office Football. In the original design for the college game, I drew an elephant named Mo who was going to appear on the scoreboard to indicate a team had momentum. But ultimately, I felt it was something that would annoy people more than make them feel more invested in the game. To date, I’ve never added a momentum factor or catch-up factor to either game and it’s unlikely I ever will. My design theory is that results should always be ratings driven and there should only be tangible adjustments to ratings. Most importantly, no ratings adjustment aside from injury should ever be, on its own, more than a small percentage. That keeps the game “honest” and understandable, though the “die rolls” associated with any kind of simulation mean individual play results can vary considerably and sometimes give people the impression that something else could be going on. I can only assure you that I test this kind of thing extensively and that everyone has runs of bad luck from time to time. Anyway…

I counted teams that scored last in regulation as having momentum. These teams won 54.8% of overtime games from 1998 until the two-possession rule took effect (essentially with the 2011 playoffs, as no 2010 playoff game went into overtime). With the new rule, 63.3% of the teams with momentum have won.

Out of everything I looked at with this study, this was the most surprising. It suggests that not only is momentum real, but that the rule change – guaranteeing that the team with momentum gets the ball in overtime – has made it more significant. Maybe even made it more fair if fair is defined as treating overtime as an opportunity to extend a game. Is the difference worth noting? I think so. That’s an 8.5% difference with 98 results since the change. Running a test assuming a normal distribution of results, I get a p-value of 0.079 on the comparison. This means I can’t reject the assumption that nothing has changed with a confidence level of 95%, but it’s awfully close. Another two years’ worth of data will help.

What about the role of the first possession? Before the change, the team with the ball first won 58.5% of overtime games. Since the change, that number has dropped to 53.1%. If the goal was reduce the impact of the coin toss, it looks like that has happened, too.

Before the rule change, teams won on that first possession 33.9% of the time. The goal was simply to score points. Nothing wrong with that strategy, but it heightens the impact of the coin toss. Of the 75 first-possession wins, 61 came from field goals and 14 came from touchdowns. The TD percentage was 18.7%. In 2017, for example, that TD percentage (TDs divided by TDs + FGs) was 58.0%.

Since the change, teams win on that first possession 19.4% of the time. They’ve scored 20 touchdowns and 18 field goals (the field goals allow the other team a possession). That percentage is 52.6%. So what we’re getting is a first possession in overtime that feels a lot like a normal in-game possession. Good or bad? I don’t know.

Before the change, a team’s odds of winning if it failed to score on the first possession was 38.1%. That has decreased to 34.4%. If a team scores a field goal on its first possession, it wins 73.3% of the decisions (61.1% if you factor in ties – three of the five ties in the NFL since the change are among the 18 games where both teams started overtime with field goals). Overall, the winning percentage once a team has failed to win (or lose) on its first possession is 42.6%.

Before and after the change, teams had about a 2% chance of losing the game on the first possession due to a fumble or interception return.

Another factor to examine is home-field advantage. Since 1998, the home team wins 57.9% of the time. Before the change, the home team won 52.5% of overtime games. Since the change, the home team has won 59.4% of overtime games. Because that’s a bit closer to the overall mean, it also suggests the new overtime rules are helping with the problems that motivated the changes.

What is the tradeoff? More ties in the regular season, undoubtedly. We’ve seen ties in 5.2% of regular-season overtimes since the change. That’s compared to 1.0% of overtimes before the rule change. Now that overtimes are reduced to 10 minutes, I expect even more ties. That didn’t happen in 2017. But since 11.7% of overtimes since the rule change last four possessions and 12.6% last five or more possessions, I wouldn’t be surprised if a high percentage of those games end in a tie. My guess is the tie percentage will go to about 10% of overtimes in the long term. It’s hard to figure in strategy changes, but that seems like a reasonable estimate. The net result will be an average of about 1.4 ties per season. That’s not going to break the NFL.

Another topic I examined was weather. Is it ever to a team’s advantage to kick off to begin overtime? Between 1998 and the rule change, teams chose to kick off twice after winning the overtime coin toss. In 2000, in a snowstorm with winds gusting to more than 40 mph in Buffalo, the Bills kicked off to New England. The strategy should have worked as the Patriots failed to score and Buffalo drove to the 12 on its first possession. But a 30-yard kick was blocked, then New England drove down the field and kicked the winning 24-yard field goal with just a few seconds left on the clock. Then in 2002, Marty Mornhinweg pretty much sacrificed his job as Detroit’s head coach by choosing to kick off to Chicago in overtime, only to see the Bears kick a field goal on their first possession.

One problem with using weather is that the NFL doesn’t consistently report data for each game. For instance, the stories I’ve read about the Mornhinweg game indicate winds were at 17 mph when he made that fateful decision. My research suggests that wind significantly reduces offense, making it particularly hard to pass. But the NFL reported a wind speed of 4 mph for that game, which shouldn’t trouble anyone. Using the reported wind speeds for games is obviously tricky. How do you determine what it must have been like on the field? Unfortunately temperature, where we have more accurate data, doesn’t have much effect on NFL games. Mostly it’s heavy snow and wind that makes life difficult on the offense.

Since the rule change, four teams have chosen to kick off to start overtime – all in games with high winds reported. The receiving team scored on the first possession in one of those games, and won two of the four. That’s far too low a sample size to make any conclusions, but since Bill Belichick has been the coach to make that choice in two of those four games (winning once), the reaction to this decision in the future will not be anywhere near as strong as it was when Mornhinweg tried it (and the rules made it much more of a losing proposition).

I scored games with a wind speed of 11 or more mph and a couple of other games with heavy snow as having difficult weather. That was 25.1% of all overtime games. About 4% of overtime games were marked as having extreme weather. What I found was that with no weather issues, teams scored on their first possession 36.4% of the time. With difficult weather, that dropped to 29.6% and with extreme weather, 15.4%. However, I found no significant weather difference in the percentage of times teams scored on the ensuing possession.

Overall, since the rules change, teams are 40-32-5 with the first possession in games with no weather marked and 12-14 in games with difficult weather. Of the nine overtime games with extreme weather since the rules change, the team with the ball first is 4-5.

I think in extreme cases, the decision to kick off in bad weather might help, but not much. And in difficult weather, at best the coin-toss advantage is removed. So my advice to coaches is that move should only be considered if you’re Belichick and you’re so far ahead of any living being in terms of understanding the game that this trivial analysis couldn’t possibly help or if conditions are absolutely brutal – which is something you probably have to determine from being out there on the field during that game.

This brings me to my last set of observations. Since I suggest that weather shouldn’t alter this decision unless your smaller players are having trouble remaining upright in the wind, is there any condition to examine? I didn’t want to data-mine the spreadsheet, because the results resemble astrology if you go in that direction. Obviously, if you look at enough samples, you’ll find something that passes a high significance test simply because of the nature of confidence intervals. Perhaps you should always kick off in Sunday night games in the NFC in odd-numbered years. Or always bet on horses with 12-letter names that start in position five. Data mining can bring endless entertainment in any sport.

I looked at one factor that might be a result of bad weather or, more importantly, why it might be worth kicking off in bad weather games. What is the score going into overtime?

Obviously tied. But if it’s 3-3, maybe the field position you potentially gain with a three-and-out on defense is more important than the advantage of going first. The average score going into overtime is 21.8 points per team. Are the odds of scoring on that first possession significantly worse in low-scoring games?

Unfortunately, scores are so tightly clustered at 17-all, 20-all, 23-all and 24-all (combined 55 out of the 103 overtime games since the rules change) that sample sizes are very tiny. All I’ve noted is that teams have scored on their first possession in only three of the 12 games that were tied at 13-all or lower headed into overtime. However, that low ratio didn’t carry over to the 1998-2011 period for low-scoring games. I can’t conclude anything based on the score itself. I’ve also noted that in games that are 30-all or higher since the rules change, teams have scored on their first possession in five of 18 games. Both high-scoring and low-scoring games aren’t favoring the first possession. In between, teams have scored on that first possession in 30 of 73 games.

And while the winning percentage of teams possessing the ball first (3-8-1) in games at 13-all or lower is unusually low since the rules change, that also doesn’t hold true for the 1998-2011 period (24-18-1).

All of this seems like low-sample-size gibberish to me, and perhaps looking at the weather is gibberish as well and you should simply take the ball first no matter what. Hard to say.

Whenever the situation arises in the NFL, the media likes to repeat the cliche that it’s unusually tough to beat one opponent three times in one season. We hear the usual pseudo-analysis – that if a teams beats an opponent, it will likely stick to that game plan because it was successful while the opponent gets to try new things. And that effect would be magnified in a third contest.

But why would a coach give away a game plan? Why wouldn’t multiple results against the same team have more to do with match-ups than anything else? Is there some rule that a team that wins a game has to stick with the same planning concepts?

It’s relatively rare to see teams play three times in one year. On an average of once or a little more per year, there’s an intra-divisional playoff game. Top teams often split home-and-home games during the regular season. So it turns out that only 19 teams have ever had the opportunity to take that third victory over one opponent.

It looks somewhat likely a 20th team will get that opportunity this year, since New Orleans has beaten Carolina twice and the schedule seems to favor a third game as the NFC 4/5 wild card match-up this season. Since the three-win opportunity hasn’t occurred since 2009, it will get more attention than usual.

Naturally, pundits will spend the week explaining that Sean Payton can’t possibly come up with a third way to beat the Panthers.

So, what’s the reality of the situation? Well, the home team has won 12 of those 19 games, and the team with the two prior victories has won 13 times. Myth busted.

I’ll also take the opportunity to point out that the last team to win that third game after losing the two regular season games was the New York Giants going out on the road as the fourth seed at 13-3 Dallas in the 2007 playoffs. Those Giants ending up beating the only 16-0 regular-season team in NFL history in the Super Bowl.

Yesterday’s episode of Monday Night Football had more than its share of unpleasant moments. Starting with an accidental, but particularly scary back injury to Ryan Shazier. That transcends the game and I’m sure every player in the NFL, even those who really don’t like the Steelers, shares genuine concern here and hopes for a full recovery.

Unfortunately, the game quickly devolved, mostly between Pittsburgh’s offense and Cincinnati’s defense. The Bengals’ George Iloka looked like he was going after Antonio Brown whenever possible, and finally was flagged after a touchdown. He was suspended for a game (strangely, his suspension was the only one retracted this week, since his behavior seemed the most calculated). The Steelers’ JuJu Smith-Schuster head-hunted Vontaze Burfict during a play, then stood over him and received a taunting penalty. He was also suspended for a game.

These incidents, combined with the Patriots’ Rob Gronkowski’s disturbing behavior on Sunday, have the NFL once again at the top of the sports news cycle for all the wrong reasons.

Ben Roethlisberger, the Steeler quarterback with two Super Bowl rings to his name and a Hall of Fame resume, was asked about the dirty play after the game. His response? “AFC North.”

That got me thinking… are some teams more prone to dirty play or is it just perception? Is the AFC North some sort of special haven for teams that can’t help but goon it up against each other? How would you study this?

Given that dirty play stemming from high emotion is fairly easy to spot, my assumption is that penalty yardage would correlate to these games. So I constructed a spreadsheet with some penalty numbers from 2013-2017. This covers 1,260 games, including playoffs. I also separated out all the games involving two AFC North teams – a sample of 58 games.

Among these AFC North games, the 239 penalty yards yesterday was the most in a single game. The 173 from Cincinnati was second only to Cleveland’s 188 against Pittsburgh in their first matchup of 2015. Third place – and the only other +200-yard combined penalty performance was the infamous 2015 playoff game between Pittsburgh and Cincinnati.

I think NFL fans remember that game. Cincinnati looked like it had completed a fantastic fourth-quarter comeback. With Pittsburgh ahead, 15-0, Roethlisberger was sacked and injured on the last play of the third quarter. The Bengals scored two touchdowns and a field goal to take the lead with 1:50 remaining. Landry Jones promptly threw an interception and it looked like two decades of playoff futility had finally ended for Cincinnati. To that point, Pittsburgh had been penalized 142 yards to Cincinnati’s 49.

But the Steelers still had time outs, so the Bengals needed one more first down to secure the victory. Jeremy Hill fumbled on the next play. Still, Pittsburgh was back at its own 9. Roethlisberger returned. He moved the ball downfield quickly, but time was running out. He threw a long pass for Brown, maybe their last chance, and it fell incomplete. But Burfict was penalized for a nasty hit on Brown and Adam Jones drew an unsportsmanlike conduct penalty, and all of a sudden Pittsburgh was in position for the winning field goal in the closing seconds.

The way that game ended – with Cincinnati losing simply because their defensive players couldn’t control their emotions – cemented the Bengals’ reputation as an undisciplined team and lends a lot of credence to the claims Roethlisberger made yesterday.

Is all of this true? Here are some numbers:

Over the last five years, NFL teams average 56.2 penalty yards per team per game. Cincinnati has averaged 56.0 penalty yards per game. So, no, the Bengals are not a particularly high-penalty team. Teams range from Carolina (49.0 yards) to Seattle (66.3 yards). Baltimore, at 60.8 yards, is the only AFC North team in the top quartile.

Are AFC North games particularly penalty-prone? AFC North teams, overall, average 57.4 yards in penalties per game. However, divisional games average 58.3 yards in penalties. That’s not a huge difference, but Cincinnati’s 70.6-yard average against Pittsburgh (not including yesterday, it’s 57.0 yards) is the highest team versus team average.

The numbers really aren’t all that notable except for one total: in the 30 games against Pittsburgh, opponents are averaging 66.4 yards of penalties while in the 52 games against Pittsburgh played by the rest of the league, they’ve averaged 55.4 yards.

Now, one thing I haven’t done is split all divisions in this manner (I don’t want to turn this into a major project), but Roethlisberger’s perceptions seem valid (53.7 out-of-division committed by Pittsburgh, 58.2 in-division), though that experience does not hold true for the rest of the division. So, over the course of the last five years, Pittsburgh’s AFC North games have averaged about one major penalty per game more than you’d expect based on team averages. That seems significant and worth some extra attention from the NFL.

You have to feel for Eli Manning. If he weren’t the brother of a quarterback whose face would be on a Mount Rushmore of quarterbacks, should such a concept exist, he’d be better loved. He has 118 career wins, including playoffs, and a pair of Super Bowl MVP trophies.

No modern quarterback with that many wins isn’t in the Hall of Fame or a lock to get there (like his brother). No player with multiple Super Bowl MVP Awards isn’t in or a lock to get there (like Tom Brady).

Yet somehow, it doesn’t add up. Recently, he passed the elite 30-mile club in career passing yardage. In the modern era, that club includes Brady, Drew Brees, John Elway, Brett Favre, Peyton Manning, Dan Marino and Ben Roethlisberger (also there this year). That’s it. More than 40 years of the modern passing rules, and he’s one of seven to throw for more than 52,800 yards (Philip Rivers might well make it eight late this season or in the playoffs, should the Chargers qualify).

We’re in the age when the definition of great quarterbacking is changing, and perhaps expectations are too high. Or perhaps there was a run of a few years when random luck produced a group of top quarterbacks. Hard to say. A few months ago, I tried to come up with a Bill James-style formula for calculating quarterbacks reaching the Hall, and Eli is borderline. Only Ken Anderson has more points in this formula of those who aren’t in or still playing (or like Peyton, not yet eligible), but the list of current players is impressive.

I think I’ve made my feelings about Ken Anderson’s qualifications well known (darn it, Canton, it’s time to fix this). But ultimately, I think Eli won’t make it because his brother will, and Brady and Brees and Roethlisberger and Aaron Rodgers are dead locks to make it. We’ll have quarterback fatigue.

We also have to consider quality of play. My quarterback metric is designed to analyze this across generations, and the fact is that Eli is an average-minus quarterback statistically. His career average in the metric is 49.8, a number that would and should have coaches wondering if they should draft a replacement. Only two of the Hall quarterbacks are below 56 career – Warren Moon (53.8) and Elway (52.7). And Elway is one of a handful to reach 150 wins while Moon had to prove himself in the CFL for years when he should have been in the NFL. None of the other current quarterbacks I’ve mentioned here are below 58 average for their careers (Peyton is at 61.2).

Today, New York Giants coach Ben McAdoo announced that Eli won’t be starting at quarterback the rest of this season. He offered him the opportunity to play a series or two, then come out, because Eli’s streak of 210 straight starts is second all-time among quarterbacks to Favre (though still a few years short). Manning wisely declined, understanding this kind of record is only valid when you’re playing in the fourth quarter.

Is it time for the Giants, 2-9 and mathematically eliminated from playoff contention, to consider alternatives to their 36-year-old leader? Manning’s average score in the metric is down to 44 this year, 27th among NFL starters. There’s reason to wonder if it’s time to retire. The Giants invested a third-round pick this year in Davis Webb, who is rather raw coming off his graduate-transfer year at California, but is huge and has a gifted arm. But Webb won’t start, either. That goes to Geno Smith, who has an 11-19 career record with the Jets and a 42 average in the metric. Smith is in his fifth year, and will be a free agent next spring. I’m not sure he’s worth the look if a look has to be taken. I’m not sure that’s a great decision, given that the Giants will draft high next year and really need to know all they can about Webb whereas the odds are low Smith is worth starting next year.

But decisions have a way of becoming necessary at inconvenient times, and the worry is that Webb isn’t ready or he would get the longest look. Since this is likely the end of Manning with the Giants, fans can appreciate what his long career has brought to New York, even if it’s just a little bit short of Hall quality and he has struggled along with the rest of the team this year.

As the Cleveland Browns stumble toward what could be the second 0-16 season in NFL history, I’m reminded of this date, Halloween, in 1999.

Halloween was unkind to the Browns this year. It’s clear that DeShone Kizer is not ready to lead an NFL team. He’s very young, and many scouts say he has talent. But starting a 21-year-old second-round draft pick at quarterback is not the NFL norm, and Kizer is really struggling.

The trading deadline was at 4:00 today, and the Browns traded for Cincinnati backup A.J. McCarron with minutes to spare. The Bengals called it in to the league office. However, for some reason, the Browns did not beat the deadline. That’s just one of those stories you can’t make up.

Anyway, McCarron, like Jimmy Garoppolo (who the Browns apparently wanted to trade for in the off-season), is in his fourth year and stuck behind an established franchise quarterback. This means he would go into next season restricted as a free agent. So the Bengals would have liked to get something for him. I’m not sure why they’d be willing to trade him in-division, but perhaps that’s another slight against the Browns.

Colin Kaepernick fans should take note of this, as an aside. If there’s one situation in the NFL where Kaepernick fits, it’s this one. As long as he’s willing to play mentor to Kizer and accept fringe starter money (this runs about $6 million for a full season these days), this would be a great road back to the NFL. I can’t say what’s in the heads of NFL GMs and owners, but this seems like a good idea while some of the other openings (backup in Tennessee, more recently) have not seemed like good fits.

But let’s flash back to 1999, the first year of the expansion Cleveland Browns (the “old” Browns became the Baltimore Ravens in 1996). In return for not causing trouble, the city of Cleveland was promised an expansion team no later than 1999, plus the expansion team would “own” the Browns’ history and team colors.

For some reason, and this was apparently unrelated to the move, the Browns also fired their head coach at the time. He found a new gig relatively quickly and has since won five Super Bowls with his new team, but that’s another story so I won’t mention his name here.

The Browns began play again in 1999 and were fairly bad. They headed into their Halloween matchup at New Orleans with an 0-7 record. As NFL games go, it was exciting. The Saints took a two-point lead with 21 seconds remaining in the fourth quarter. But quarterback Tim Couch, the first pick in the 1999 draft, was up to the challenge.

Couch completed a 19-yarder to the Browns’ 46-yard line. Time out with 0:02 on the clock. Couch then rolled right and threw the ball as far as he could. Kevin Johnson fought through a crowd in the end zone to give the new Browns their first victory.

Incidentally, as of the last time I messed with these numbers, Couch has by far the highest rate of late fourth-quarter game-winning drives in modern NFL history. No, don’t reserve a bust in Canton – his career quarterback record was 21-36. But 11 of those 21 wins came on plays just like that one (well, not quite just like that one).

Now there’s some connection between this New Orleans team and Front Office Football. For years, it was a running joke that there were too many Billy Joes in the game. When I designed the original game (and perhaps this is still an issue, because the design in this case is still quite similar, but I may have removed the name), any first name that was shared by more than one NFL player was placed on the list of more frequently used first names (IIRC, there were about 500 names on that list).

The Saints had two primary quarterbacks – Billy Joe Hobert (4-8 career record) and Billy Joe Tolliver (15-37 career record). It’s a name that stands out, I guess. In 1999, they shared quarterbacking duties. Both had 1-6 records as a starter that year. Hobert started this Halloween game. He was either hurt or pulled after throwing a second-quarter interception (he is listed as making the tackle, so I’m guessing hurt). Tolliver relieved.

Cleveland ended up winning a second game that season. And that was the Browns’ worst record (new or old) until last year. But they have finished last in the AFC Central/North in 14 of their 18 seasons since expansion and look certain to make that 15-for-19. They have a 0-1 playoff record since expansion and their last playoff victory was on New Year’s Day in 1995 against New England, who was apparently impressed by their head coach because… well… still not mentioning his name.

Finally, back to winless seasons. The Browns won in week 16 last year, beating the Chargers as their kicker missed a game-tying 45-yarder as time expired. Now in Los Angeles, the Chargers have played a bit better (3-4 so far, as opposed to 5-11 last year). They lost their first two games this year essentially on missed game-ending 44-yard field-goal attempts.

This meant YoungHoe Koo, their new kicker this year, lost his job. Koo, by the way, is only the fourth player born in Korea to play in the NFL (and the second with a father not an American stationed with the military). Since one of those four players is Hines Ward, South Korea has by far the highest percentage of native-born Super Bowl MVPs in NFL history. In case you’re interested, the Browns visit the Chargers in week 13.

It probably hasn’t escaped most NFL fans that something unusual is happening in week seven. Three different teams have been shut out.

The Cardinals were beaten, 33-0, by the Rams. In addition to treating London fans to a performance that could, on its own, cause serious harm to the NFL’s popularity overseas, Arizona lost quarterback Carson Palmer to an injury that may end his season.

The Colts fell to the Jaguars, 27-0. Jacksonville’s defense is one of the positive stories of this season – a unit that could drive a playoff run for a team that probably wouldn’t make the playoffs at all based on offense alone. The Jaguar defense has taken all the pressure off of a difficult quarterback situation.

The Broncos, suddenly looking quite bad, lost 21-0 to the Chargers, playing in front of a crowd that seemed made up mostly of family and friends of the players (I exaggerate).

Not only have three teams failed to score this week, but they lost to three teams that had a combined total of 12 wins in 2016.

Is this unusual? The three shutouts brings the 2017 total to five. That’s two more than we had in all of 2016. Since 1978, there have been 345 shutouts (8.6 per year). The most in a season was 17 in 1992. The least was two in 1994. We’re in a down-cycle lately, probably because scoring is up. Just 37 in the last eight years.

Do they come in bunches? Shutouts don’t happen all that often, so one could perceive a bunch just because they’re notable. The last time there were three in a week was week 15 of 2012. There were only two other shutouts the entire year. Three has happened a few times. The last time there were four shutouts in a week was week 12 in 1983. And the only other time there were more than three in the modern era (I go back to 1974 when the passing rules were changed) was the opening week of 1977 with five.

So I don’t think this is a sudden and notable occurrence. Odds are good we won’t have more than a handful of shutouts the rest of the season.

What about scoring in general? Is it down this season? So far this year, NFL teams have averaged 21.9 points per team per game. That’s the lowest since 2009 and down from last year’s near-record of 22.9 (a half point lower than 2013’s record total of 23.4). But 21.9 matches the most from 1968-2007, so it’s not an unusually low total. And prior to this week, teams were averaging 22.2 points, which is certainly in line with recent seasons.

My sense is that scoring is a bit down this year. If I had to put my finger on it, based on a cursory look at statistics, I’d say improved pass rushing is making teams decide to throw a tiny bit less – perhaps that’s also because we have some more inexperienced quarterbacks this season or the league is allowing a little more contact from defensive backs. It’s very hard to tell without a full season of data.

From time to time, I like to take the opportunity to challenge my own perceptions about football. One strategy that’s often debated is whether teams should attempt a two-point conversion after a touchdown.

Decades ago, legendary coach Dick Vermeil, then a coordinator at UCLA, created a chart that’s still in popular use today. It’s not a chart I used when developing Front Office Football, but it’s not that different. Some of it is obvious; for instance if you trail by two after scoring a touchdown, you should try and tie the game. But even then, when in the game should you start consulting the chart?

Some of it is complex, or even controversial. You’re supposed to go for two when you lead by four after scoring. The idea is that if you make it, a subsequent field goal will give you a two-score lead. But if you miss, a subsequent field goal will leave your opponent the opportunity to tie on a touchdown without going for two.

Since the two-point conversion rate in the NFL is about 45% and the extra-point rate with the new distance rule is about 97%, I’d hesitate to start using any chart until well after halftime. Just take the point.

A few months ago, when working on a percentage win calculator that I’ve yet to put into any product, I analyzed a few seasons’ worth of play-by-play data and compiled a chart that could be useful for making these decisions. As an aside, I do this kind of thing a lot. Most of what I discover on these odysseys never amounts to meaningful work within my products. This might well fall into that category. But it could also be useful in compiling a more fine-tuned chart – one that even incorporates time remaining.

Today, I watched a good part of the Lions’ opener against Arizona. With 3:07 left in the third quarter, the Lions scored to cut the Cardinal lead to two. Some might say it was too early to start using a chart, but no one would question the wisdom of the decision in the closing moments. The Lions tried the conversion and failed. How did that change their win chances?

I have my play-by-play data broken into 100-second increments. To take a broader brush to include as much as is reasonable, I used the categories from the end of the third quarter to 6:40 remaining in the third quarter. There have been 2,797 plays undertaken with the score tied and a team with the ball, 1,031 plays with a one-point lead and 397 with a two-point lead. Not an overwhelming amount of data, especially since you can’t assume a reasonably uniform distribution of field position within that small a data set with the two-point lead. But in 1,629 of the tied scenarios (58%), the team with the ball won (possession matters – Arizona would have possession after the ensuing kick). In 574 of the one-point lead scenarios (56%), the team with the ball won and in 280 of the cases with a two-point lead (71%), the team with the ball won.

This is a great example of where sample size lets you down. The reason I’m writing about this is to give you some insight into my process when examining a particular question. When do you make a conclusion and when do you accept that you just don’t have enough information? This is several seasons’ worth of complete play-by-play data (350,000+ plays broken down by lead and time to go blocks), and the raw data set still isn’t good enough to make solid conclusions.

What the above numbers suggest is that the value of the extra point is immaterial, but there’s a 10-15% game-win cost in going for two and failing. That just doesn’t feel like a reasonable conclusion. When I was doing the initial work with this particular data set, I ran a whole series of rolling averages to come up with a win-percentage chart that supported the data with less precision, but a consistent set of percentages that required the least amount of intervention on my part (making “decisions” about interpretation and then using those decisions to influence how the rolling averages were applied). I feel more confident in presenting that chart as reasonable, at least in the sense that it could be used to help with this kind of decision. If this work results in use in any product, I would use that more processed chart.

Going back to the analysis of the first Lions decision: I come up with a 55.8% Arizona win percentage when tied during that time block, 61.1% with a one-point lead, 64.3% with a two-point lead. So, -3.2% for Detroit with a failure, +5.3% with success. Assuming 45% success on two-point tries and 97% success on extra points, the decision to go for it, in itself, raises Detroit’s win percentage by about a half a percentage point. I think it’s reasonable to conclude that it was a good decision. In general, my data supports this case up to about midway through the third quarter. Earlier than that, I would advise against ever going for a two point conversion except when desperate and in need of multiple positive results (let’s say you score and trail by 18).

The second Lions decision came with 9:27 remaining in the game. This time, they led by four. Vermeil’s chart says “go for it.” I don’t know if Jim Caldwell uses this particular chart, but he went for it. The Lions failed. The value of success and penalty for failure is explained at the start of this article. Now to break down it down using the data…

If the Cardinals gain possession with a six-point deficit (Detroit makes the conversion), they have a 29.2% chance of winning. With a five-point deficit, that’s 36.0% and a four-point deficit, that’s 41.0% percent. Factoring in analysis of success and failure, that amounts to, again, about a half-percentage-point increase in the Lions’ win chances. This surprised me a little, but, as it turns out, a six-point lead is quite a bit better than a five-point lead, even relatively early in the game.

Vermeil’s chart may be old, but it holds up even in the modern game.