Unicorns, The Tooth Fairy, the Cowboys, and Field Goal Range

If there is one example that demonstrates the fallacy of field goal range, Sunday's Cowboys-Cardinals game has to be it. With the score tied at 13 and under a minute to play, DAL converted a 3rd and 11 to move the ball to the ARI 31.

There it is. Game over. Field goal range, right? Who needs more time when you're inside the 35?

At the end of the play there were 26 seconds on the game clock, and DAL had 2 timeouts. DAL milked 18 seconds off the clock, then spiked the ball to stop the clock with 8 seconds to go. Then, head coach Jason Garrett called a timeout as if he was icing his own kicker. (The icing effect is greatly overstated, and in many cases, it simply gives the kicker a practice kick.) The kick was missed, and the game entered the dice roll of sudden death overtime.

With 26 seconds left, DAL could have used one of their 2 timeouts immediately. They could have run two plays, even including one run, and still saved a timeout for a final field goal. With the ball at the 31, a field goal attempt is a 49-yarder, which, on average, is a 65% proposition. Moving the ball just 6 yards increases the chances to 75%.

Compare Garrett's decision to that of Packers coach Mike McCarthy, who faced a similar situation against the Giants yesterday. With the score tied and 44 seconds to play, GB had a 1st and 10 at the Giants 29. GB threw a pass for a one-yard loss, setting up a 2nd and 11 on the NYG 30, with 21 seconds left--an slightly dicier situation than DAL would have faced if they had called a timeout. GB threw the ball deep for an 18 yard gain, which led to a relative chip-shot field goal for the win.

After the game, Garrett tried to explain his clock management. "We felt like we were in field-goal range. We have yard lines that we use as guidelines before the game. We felt like we were in range at that point. Tony [Romo] had them on the line of scrimmage quickly, so we went ahead and clocked it and used that as a timeout."

First, that's not even true--it took 18 seconds to line up and spike the ball. Second, it makes no sense. With 2 timeouts in hand, why even rush to the line and spike the ball, risking a false start?

Also, note Garrett's conception of "field goal range." He has a yard line established before the game, which is the standard throughout the league. That kind of thinking assumes a yes-or-no, black-and-white idea of a field goal attempt. Either you're in range or you're not. If we're "in range", and the kicker misses, well, that's his fault. Coaches would be better off thinking in probabilistic terms.

There is no such thing as field goal range, except in the most technical and useless sense. Closer is always better. The field is one big gray area when it comes to field goals. Head coaches should announce an edict to their staffs that the term field goal be banished from their vocabulary immediately. In its place, the term field goal attempt shall be always be used.  

Send in the field goal attempt team. 

The coaches that best manage risk will win more often.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

41 Responses to “Unicorns, The Tooth Fairy, the Cowboys, and Field Goal Range”

  1. sunrise089 says:

    This sort of story makes me thing about 3 NFL head coaches should be paid ~20 million per year and the other 29 should be paid $150,000. I can't imagine replacing a guy like Garrett with a ANS-reading random manager of one of the more successful big box retail stores in its respective chain would lead to any fewer wins.

    If the NFL had a relegation model I suspect most of this idiotic gut-feel mentality would be gone in the first decade.

  2. Anonymous says:

    Great article.

  3. James says:

    Drove me crazy. Furthermore, Garrett did the same thing in the first Redskins game and lucked out, and again last week versus the Dolphins. I've now decided the only reason he continued to run the ball against the Dolphins is because he correctly understood the need to bleed the clock.

    I can't believe he understands the need to not allow the opponent to respond, but NOT that shorter field goals are always easier. He'll be crucified for "icing" the kicker but almost no one will mention the field goal was only a couple of yards short. A couple of yards Murray could have picked up if they hadn't run down the clock.

  4. Krazee says:

    "I can't believe he understands the need to not allow the opponent to respond."

    Actually, Garrett only understands what he reads from books. He has an ivy-league edu-maca-tion. The dude couldn't survive on the streets if he had to though. You think he knows how to change the oil on his car? Nope, but he probably wrote a thesis on how a combustion engine operates.Garrett reminds me of the book smart nerd boy chicken from Looney Tunes that messed with Foghorn Leghorn's mind, but without the end results.

  5. Michael says:

    Garrett seems to be Coaching in order to 'not lose', as opposed to playing for the win. I can understand if he was attempting to spare another freaky Romo end-of-game-gift-interceptions (and I'm saying that as a Romo FAN), but come on.

  6. Total says:

    Actually, Garrett only understands what he reads from books. He has an ivy-league edu-maca-tion. The dude couldn't survive on the streets if he had to though.

    Yeah, hard streets that NFL is. You know who else has essentially an Ivy league edu-macation? Who--by your logic--can't change the oil in his car (cause that sure is good training for clock management)?

    Bill Belichick (Wesleyan).

    So take your reverse snobbery and shove it.

  7. Anonymous says:

    Can't tell if joking or stupid.

  8. Pete says:

    I realize teams practice two-minute drills all the time, but do coaches? Do coaches practice clock management under real game conditions? Do teams have "play books" or do they have "scenario books" with plays in them? I’m really asking because I don’t know.

    In the last two minutes of the game, there isn't much time to think through decisions. There has to be a plan for how to deal with various scenarios. It seems like in the Cowboys game there are a few possibilities.*** #1. They had a smart plan and didn’t follow it, #2. They had a dumb plan and followed it. #3. They had no plan and were dumb. Garrett’s comments about the game seem to indicate a combination of #2 (field goal range) and #3 (time out use).

    The field goal range problem seems to be an example of coaches not wanting to put themselves in a situation where they would be blamed. If the team is in “range,” a fumble or a pick in an attempt to get better field position can get blamed on the coach. If a kicker misses a “makeable kick” , it’s the kicker’s fault. As dumb as this it is, it falls under the umbrella of NFL coach orthodoxy.

    The time out situation shows just blatant stupidity and lack of preparation. Some of the blame for this should go to Romo too, who should have known to call the time out. Dumb and Dumber.



    ***There's also another possibility, #4, where they had a dumb plan and didn't follow it and were dumb, but that makes my head hurt.

  9. Max says:

    When you say "There is no such thing as field goal range, except in the most technical and useless sense," what exactly us that sense? I understand that closer is always better, but the choice to kick or not to kick is ultimately a binary decision that implies there is a field goal range, and that they were indeed within it. Obviously they were dumb and should have gotten closER, but if there's 1 second left and you're 90 yards away, you don't kick even if a successful kick would win the game (unless you're playing against the Eagles, in which case all last second kicks are good...). So while I completely agree with everything you said above about stupid decision making, where do you think the REAL field goal range starts? Where do you go for a Hail Mary over a field goal?

  10. Brian Burke says:

    In the sense that, at some point beyond a 60-yd attempt, the chance of success is zero.

    My point is that FG range is probabalistic. A coach can draw a line at a certain distance, but what does that mean? >50% shot, a 75% shot?, a 95% shot?

    There is no line, just a big gray blur that gets fuzzier the further out you get. It's like quantum field goal theory.

  11. Brian Burke says:

    Sorry, Max. I meant to actually answer your question and went off in a spiral. There is an inflection in the probability curve for field goals at about the 50-yd kick distance. That's a meaningful point on the field, but I would still avoid calling it "FG range."

    With 1 seconds to go, as long as the probability of a FG att is greater than that of a Hail Mary, I'd go with the kick. But I'm not sure what that is. 3%? 1%? 5% It's based on distance to some degree, obviously.

    Probably would make a good subject for a research post.

  12. Brian Burke says:

    Pete-That's a good question. I think many coaches have a guy in the booth whose job it is to focus exclusively on end-game time management. I know all teams will do a next-day scrub of end-game management decisions and be very self-critical. Everything gets evaluated, even head coach decisions.

    Unless you're in a high-pressure time-sensitive occupation, it's very hard to understand how small your IQ becomes in situations like that. Flying fighters, we used to joke about doing math while in the air. The trouble is, flying requires lots and lots of math. Time/fuel/distance/speed/altitude/climb rates, etc. considerations are critical in all phases of flight. When I think about it, flying fighters is really just doing math (including 3D geometry) with a stick and throttle instead of a pencil and paper.

    In that light, I wonder if an end-game simulator would be useful for coaches. They could rehearse dozens of end-game scenarios, essentially giving them a lifetime worth of experience.

  13. Joshua Northey says:

    How can someone with this poor decision making skills be paid this much money?

    Yeah he can design an offense, that is great, hire him as the O-coordinator (or preferably a consultant) but for pete's sake leave the in-game decision making to someone who has a flicker of an idea that they are doing.

  14. Anonymous says:

    I would classify three important distances for FG kicking:

    #1 When it is possible to actually kick the field goal. This depends alot on your kicker and the conditions, and can range from 50 to 65 yards. This is the distance where you would attempt a FG at the end of the game or the half.

    #2 When it is probable that you make the field goal. During normal play, I would classify this as the distance where you "break even" in terms of expected points, ie, the distance when

    p_make*EP_made_FG = (1-p_make)*EP_missed_FG.

    The above distance is the practical FG range of your kicker, and anything above that distance should likely not be attempted (the actual decision tree is obviously more complicated, but I think what I said holds true almost all the time).

    During end game/OT situations, this distance is useless.


    #3 "Chip shot range". If a team advances the ball from the 10 to the 5 yard line, p_make barely increases at all. In fact, it is likely that your added probability of making the field goal is dwarfed by the probability of a big error (TO, 10 yard penalty). If the cowboys were in "chip shot range", then Garrett's strategy was correct (the Broncos were in chip shot range and correctly ran out the clock, even though they likely should have kneeled down instead of running the ball). As a ball-park figure, I would classify "chip shot range" as the distance from which you have a 95% or greater chance of making the FG, which is somewhere around the 10-15 yard line. Your article http://www.advancednflstats.com/2009/09/4th-down-study-part-2.html has a handy graph, but I suspect that a modern NFL kicker exceeds these percentages by a little bit.

    In short, I would classify Garret's thinking as utilizing the strategy for category 3 when he was in category 2....

  15. slushhead says:

    Grantland addresses this, as well as some other issues:

    http://www.grantland.com/story/_/id/7318214/icing-kicker-work

    I've got to hand it to Bill Simmons and his writers for acknowledging the importance of statistics in analyzing football decisions, but his "2 for 9" section contains at least two major flaws in logic. I just wish they'd be a bit more careful, it undermines an otherwise interesting argument.

  16. Jeff Fogle says:

    The link you posted had these quotes from Garrett:

    "You see so many situations where you have negative plays in those situations. We felt like we were in his range to give him a chance to kick the game-winner."

    "Well, we felt pretty good about where we were...Once you get to that 30-yard line we felt like that was a pretty good opportunity for us."

    Aren't phrases like "give him a chance," and "pretty good opportunity" examples of probabilistic thinking? Probability is the study of chance, and he even actually used the word "chance"! You imagined what was in Garrett’s head to back up your theory, but excluded his actual words that disproved it. He was obviously thinking in terms of probabilities based on those comments.

    The consensus seems to be that he made errors in judgment about the potential negatives. The metaphor for that would be fear of a bogeyman or something (can’t imagine where a fear of bad things happening on a Tony Romo-led team would come from)…not belief in unicorns, the tooth fairy, or some force-fed absolute about the definition of field goal range.

    Blame him for what he actually did wrong. Accuracy has to be at the heart of analytics or the field will just be a different flavor of blowhards. Examine all the evidence, don't exclude quotes that debunk your theory. Include them and come up with a better theory. He used the word "chance" and you said he wasn't thinking in probabilities....

  17. James Sinclair says:

    Just because he was talking in probabilities doesn't mean he was thinking in probabilities.

  18. Brian Burke says:

    No, Jeff, not at all. He's thinking in binary, either "there's a pretty good chance" or "there's no chance."

    Garrett's *ACTIONS*, not his words, nor my imagination of what was in his head prove my "theory" as you put it.

    Jeff, don't ever call me a blowhard again. Your contrarian know-it-all and run-on comments are entertaining on some level, but watch it. That kind of attitude is neither helpful nor welcome here.

  19. Anonymous says:

    Bottom line is that Tebow would have made the field goal.

  20. Ian Simcox says:

    "In that light, I wonder if an end-game simulator would be useful for coaches. They could rehearse dozens of end-game scenarios, essentially giving them a lifetime worth of experience. "

    When I was at university back in 2003 I knew very little about American Football. We just happened to pick up a copy of Madden 04 for the playstation for our house to play.

    I learned so much about end game clock management playing that that it seems impossible to me that coaches can be so unaware of the right thing to do at the end of a game.

    Things such as stopping on the 1 and not scoring the touchdown late on, letting the other team score to get the ball back, when to call timeouts etc just become second nature. I once intercepted the ball and had I run out of bounds the game would have been down to kneels. Instead I scored the TD, they returned the following kickoff back for a TD and I had to recover an onside to win. How coaches and players are so situationally unaware really does astound me.

  21. Anonymous says:

    For places on the field where field goal attempts are at the knee of a success curve and small changes in distance result in large changes in the success probability, it would be interesting to determine an optimal set of offensive play calls. Perhaps plays (and on field decision making) with very unlikely loss of yardage in exchange for reduced expected positive yardage relative to a standard play.

    You could then take a pdf of the outcomes for the selected play(s) and the updated field goal success probability for the resulting field positions to determine a pdf of outcomes to compare against the probability of success with respect to doing nothing and kicking from the current position.

    Just a thought.

    - Arthur R

  22. Jonathan says:

    I wonder how much of this is on Garrett, and how much is on Jerry Jones for telling him exactly what to do in endgame situations.

  23. James says:

    Jeff, if Garrett was really thinking in probabilities then he would weigh the likelihood of gaining a few extra yards or losing yards, and the resulting impact on the field goal attempt. Had he thought that way he would have realized that the chances and benefits of gaining a few more yards vastly outweighed avoiding a negative. THAT is thinking with percentages.

    As it turns out in this particular situation, a negative play wouldn't even have hurt the team, as the field goal wasn't going to be good regardless.

  24. Jeff Fogle says:

    James Sinclair said:
    "Just because he was talking in probabilities doesn't mean he was thinking in probabilities."

    He was explaining his thought process to reporters. That's the best evidence we have for what he was thinking.

    BB said:
    "He's thinking in binary, either "there's a pretty good chance" or "there's no chance.""

    Garrett explained that he was weighing the possibilities of what might happen if he tried to run more plays...with the likelihood of making that field goal from that distance. He decided that the percentages favored trying for the field goal. That's dealing in probabilities. You based the whole article on the premise that he wasn't dealing with probabilities...even though the piece you linked to showed that he was.

    BB Said:
    "Garrett's *ACTIONS*, not his words, nor my imagination of what was in his head prove my "theory" as you put it."

    You can't deduce from one action whether the thought process behind that action was binary or more complex. A complex (but perhaps badly analyzed) probabilistic approach can lead to letting the clock run down before trying the field goal in the same way a binary choice could. Multiple pathways can lead to the same action. His actions don't prove binary thinking. The best way to find out what he was thinking is to ask him. Which is what reporters did.

    BB said:
    "That kind of attitude is neither helpful nor welcome here."

    On a day where you called sports reporters monkeys and suggested that an NFL coach believed in unicorns and the tooth fairy.

    BB said:
    "There is no such thing as field goal range, except in the most technical and useless sense...The field is one big gray area when it comes to field goals."

    The definition of field goal range is widely understood, and it's not a useless definition. The field is not one big gray area...as there are areas of certainty in terms of not being able to reach the distance...and areas of near-certainty from very close ranges. You've documented this yourself, so it's odd to use a "one big gray area" metaphor. Garrett was in a range where he had to consider the probabilities, as he explained.

    BB said:
    "Closer is always better."

    Garrett expressed the equivalent of "further away is always worse" in the quote you excluded when he talked about the possibilities of bad things happening when you run plays (implying fumbles, sacks, holding penalties,false starts). He was in a spot where a false start penalty would have a percentage impact on the distance of the kick.

    BB said:
    "Send in the field goal attempt team. That might help straighten out their thinking."

    Garrett said "to give him a chance to kick the game-winner." He seemed to be aware that it was an attempt with that phrasing.

    BB said:
    "Your contrarian know-it-all and run-on comments are entertaining on some level."

    It takes two words to say "great article." Dissent takes more words. The goal is not entertainment (and if there are times when it seems I'm using terms in absolutes, I'm happy to debate the use and make corrections if needed). Bill James inspired the field of analytics roughly 30 years ago. I was among the inspired. I'm the Indian in the commercial crying at all the pollution 30 years later.

    First Jamesian principals...statements are either true or false. Try your best to figure out if they're true or not. A headline of "Unicorns, the Tooth Fairy, the Cowboys, and Field Goal Range" and the ensuing article didn't represent the true realities in play in my view. I made my case. The essence of peer review. Thanks for the opportunity to post comments and have discussions at your site.

  25. Anonymous says:

    Jeff, your comments don't fit with the observations.

    Garrett's decision making and subsequent explanation clearly demonstrate that

    a) Garret didn't feel that advancing the ball improve the chance of making the FG

    b) Due to a), the only worthy metric to evaluate running another play is "what can go wrong".

    What he CLEARLY didn't do was consider the advantages of advancing the ball further (i.e., improved chance of making the FG) vs the negatives of potentially having something go wrong (penalty, turnover, etc). So I really think you barking up the wrong tree here.

    But, having said that, I would be very interested to determine, using the correct thinking, as to when teams in this situation should

    a) sit on the ball and kick the FG as time expires.
    b) Run more conservative plays (i.e., run it up the middle, short roll out passes, etc.) that minimize really negative outcomes.

    I hypothesize that a) occurs around the 10 yard line, and b) around the 20, but would love to see Brian run the numbers.

  26. epv says:

    Also keep in mind that they burned probably 30-40s not playing in a 2 min drill prior to actually making the kick. The Cowboys have had issues with clock management all year. There is never any urgency, or indication that they have the ability to play with urgency.

  27. Jeff Fogle says:

    anon, appreciate your comments.

    I don't think Garrett's actions and comments "clearly" demonstrate that he "didn't consider the advantages" of moving the ball. It could also be interpreted as something like...the chances of the percentages getting a lot worse outweighed the chances they would get a little better. He considered them, and decided the probabilities favored kicking now.

    Bailey's career long is 51 yards. Looks like he was 75% this year at 48 yards or more.

    Let's assume Bailey would fit the standard BB suggested of 65% at 49 yards rather than 75% in the short sample size. A five-yard penalty for a false start (in front of a large and very loud road crowd) would move that back to 54 yards. He's never kicked a field goal that long in game action. Run plays, and you might move 65% up to something better, but you might fall back past the edge of Bailey's trustworthy range. (Oh, Romo had already been sacked 5 times too).

    I think Garrett's actions are more in line with being overly intimidated by these probabilities than they are with believing in unicorns or the tooth fairy. He was right on the edge of his kicker's range. There wasn't much margin for error when running additional plays.

    Agree with you that more study of the issue would be very interesting...

  28. Anonymous says:

    At best you can say that Garrett employed overly risk adverse thinking, like as an example, a poker player who folds A-A because he "always loses with Aces".

  29. Jeff Clarke says:

    I think I have to agree with the other Jeff here. Garrett's mistake was misapplying probabilities not refusing to apply them. It sounds like he thought the odds of a sack or a penalty were considerably higher than they were and the difference between a 49 yard attempt and a 35 yard attempt was considerably smaller than it was.

    If I was a reporter, I'd never settle for an answer like "pretty good" anything. I'd demand that they quantify it. If they don't think they can quantify it, I'd demand that they estimate it.

    Perhaps coaches can blow reporters off but owners really should demand to know their thought process on things like this. Why don't owners ask them these questions in interviews? I know darn well that some coaches could actually quantify it and those coaches would be better than the ones that use terms like "pretty good chance".

    Sooner or later an owner is going to hire a coach that thinks like we do. That owner will be more than satisfied with the results.

  30. Jeff Clarke says:

    BTW, to a certain extent, I think we are all blowhards here. There really is nothing to be gained from anybody calling anyone else names.

    I'm stopping to think about whether I think Garrett saw it as a binary or whether he misapplied probabilities and I guess its a little of both. But that really doesn't excuse the mistake and either way, it was still a huge mistaike.

    The bottom line is that he didn't quantify it and when you don't quantify it, its really difficult to understand whether its worth the risk to move from "pretty good" to "really good" chance.

  31. Anonymous says:

    The Cowboys have obviously let their extremely fluky losses to the Jets and Lions influence their overall decision-making. You can never become a championship team if you stop trusting your best players.

    The Packers sustained all kinds of unlikely losses on their way to an 8-5 start last year. It would be interesting to peak into a parallel universe where they missed the playoffs instead of winning the Super Bowl and see if that led Mike McCarthy into all types of sub-optimal decisions (beyond his normal game management gaffes).

    It's impossible to know for sure, but I doubt he would. The Cowboys - and a lot of other teams - won't achieve their goals until they eliminate the culture of decision-making cowardice.

  32. Pete says:

    Jeff(s),

    Garrett’s words are inconsistent with his own actions. If he thought the risk of losing yards outweighed the benefit of trying to gain yards, why did Romo line up and spike the ball, risking a fumbled snap or false start?

    Maybe Garrett wanted to preserve the two time outs and run more plays and then changed his mind after Romo spiked it. Then why did it take Romo so long to spike it?

    Maybe Romo did this on his own. That means Romo was unprepared, which reflects poorly on Garrett.

    There is no way to interpret what happened that makes Garrett look good, but his explanation at the press conference made him look even worse.

    It’s just shocking to me that, after 90 years of professional football, plus all the stats now available, plus all the personnel dedicated to designing plays and watching film, that there are still teams that haven’t figured out how to run the two-minute offense—the most important set of plays in the game.

  33. Anonymous says:

    Garrett's action reminds me a lot of the blackjack player who elects to stand with 16, even though the dealer is showing a King. The odds say hitting is the correct play, but the player's simply too afraid of busting to risk taking one more card. Just another example of the better-to-lose-slowly mentality, I suppose.

  34. Anonymous says:

    thought it might be useful to have blowhard definition handy :)

    blowhard 63 up, 194 down
    As mentioned in another definition, a very braggy and pretentious person, overly prideful and arrogant. See egotist.

    Also could be an insult to anyone; the definition of this other insult is totally unrelated to boasting or bragging. We all know what blow means.
    You blowhard!

    You blow hard!

  35. Anonymous says:

    "With 1 seconds to go, as long as the probability of a FG att is greater than that of a Hail Mary, I'd go with the kick. But I'm not sure what that is. 3%? 1%? 5% It's based on distance to some degree, obviously."
    Speaking of which, has anyone tested the success rate of a hail marry play vs one of those razzle-dazzle lateral plays?

  36. Max says:

    I have agree with everything said. Whenever the coverage team shows a line to say where they have to get to to get into a guy's range it bugs me. Just because he hit there once in Denver with the wind doesn't mean he'll ever do it again.

  37. Rick Desper says:

    @Total
    As a Wesleyan grad, I'm offended that you would equate our fine university with the nonsense that goes on at Ivy League schools.

    Jason Garrett is a Princeton grad. Now if only he'd had the sense to go to a proper New Jersey school like Rutgers (as I did for graduate work), he might have learned something about probability.

    Hmm..
    to be fair, Princeton is reasonably good at math.

  38. Anonymous says:

    Funny how BB stopped responding after our hero Jeff put him in his place for a second time.

  39. Brian Burke says:

    huh? This is from, like, a year ago, dude.

  40. Anonymous says:

    I play golf with a retired, top-notch NFL kicker. He said the way to ice him was to NOT call time out. Everyone is expecting the time out. Do the unexpected.

    Also, closer IS better except perhaps on very short angle kicks. I'd rather have a 30 yarder in the center that 19 yarder on the hash.

    But, "centering the ball" for kicks is highly over-rated beyond 30 yards. A 40 yard kick from the hash is just as easy as 40 from the center. The angle is irrelevant except in close.

    Also, the angle is a much bigger factor in college with the wider hash marks.

  41. Matthias says:

    "Either you're in range or you're not. If we're in "range", and the kicker misses, well, that's his fault. This is a failure to think in probabilistic terms."

    Great point. Same problem in baseball with bunting. If the player blows the sac bunt, all the blame goes on the player. Sac bunt are generally bad decisions, in part because they're only successful 70-80% of the time. Coaches that pull this "he should have made the play" bullshit are stubborn blobs who need a lesson in probability.

Leave a Reply