tag:blogger.com,1999:blog-15377771.post115215846344453781..comments2024-03-13T03:27:50.582-04:00Comments on Smart Football: Run/Pass Balance and a Little Game TheoryChrishttp://www.blogger.com/profile/07204245083374821812noreply@blogger.comBlogger16125tag:blogger.com,1999:blog-15377771.post-32023295034672178072008-02-06T22:22:00.000-05:002008-02-06T22:22:00.000-05:00The numbers for a truly dominant running team coul...The numbers for a truly dominant running team could arguably be flipped.<BR/><BR/>Better running average, combined with more attempts, and just enough passing average to move chains more than back teams off the ball.<BR/><BR/>Depth, formations with multiple backs, this would fit into the plan without meeting the road blocks or mind sets we develop off the stats we see.<BR/><BR/>The conceptual egg is after the chicken, in either case.<BR/><BR/>"These two stats converge in the most important first and second down stats, which are average yards per play."<BR/><BR/>These or correlative indicators.<BR/><BR/>First down success means you can take more chances on second down and that the opponent has less likely odds of stopping the play.<BR/><BR/>Still, this should be an agreed upon barometer that helps measure overall success. It certainly helps draw attention to the things teams need to focus upon coming into games from both sides of the ball.<BR/><BR/>Efficiency of offense, and for the defense, a way to take the other team out of the comfort zone of play calls and situations that form the basis of their playbook.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1163135629858809482006-11-10T00:13:00.000-05:002006-11-10T00:13:00.000-05:00If I had seen this before I published this recent ...If I had seen this before I published this recent article, I would have cited you. Nice work<BR/><BR/>http://www.bepress.com/jqas/vol2/iss4/5/Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1155759073906791432006-08-16T16:11:00.000-04:002006-08-16T16:11:00.000-04:00Chris,I just picked up a copy of Football Outsider...Chris,<BR/><BR/>I just picked up a copy of Football Outsiders 2006 NFL Prospectus. One sectin of the prospectus contains league wide statistics going back to the 1940's. One interesting thing that I noticed was how constant the "passing premium" was usually between 2.5-3.0 yds. This held even when many fewer passes were attempted and when the average yards per play were lower. Thought that you might be interested in this since your article mentioned the higher pro passing premium.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1155445589576009462006-08-13T01:06:00.000-04:002006-08-13T01:06:00.000-04:00Chris,Just wanted to say that reading your stuff i...Chris,<BR/>Just wanted to say that reading your stuff is outstanding and I am a VERY regular reader of your "columns". I was wondering if there was any way that you could give some kind of update, (maybe once a week? every two weeks?) to let us know what you are working on or let us know when some new stuff will be posted. I check your blog at least once every few days and find myself saying- " Please let there be another article, please let there be another article....." Again, I love your stuff and heave learned a TON from you. Thanks.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1155053402158835462006-08-08T12:10:00.000-04:002006-08-08T12:10:00.000-04:00Chris: Just going over this post yet again, and r...Chris: Just going over this post yet again, and reflecting how your thesis not only stems from Nash, but also how the same phenomenon is reflected in the strategic world by what Luttwak calls "the paradoxical logic of conflict":<BR/><BR/>The best road between two enemy cities is often the worst road for an invading army to take, since it lies directly along the expected axis of approach that the enemy has prepared for...Ted Seayhttps://www.blogger.com/profile/01311385177075772697noreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1154358549931029782006-07-31T11:09:00.000-04:002006-07-31T11:09:00.000-04:00Chris,When's more of this coming? This is great s...Chris,<BR/>When's more of this coming? This is great stuff. Much better than "we're going to go out there and pound the footbaw...etc." Really seeing what's happening between the stats is great. I've just started reading so I don't have anything intelligent to discuss yet.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152437613990047782006-07-09T05:33:00.000-04:002006-07-09T05:33:00.000-04:00Hi Chris,This is really interesting, but as I do w...Hi Chris,<BR/><BR/>This is really interesting, but as I do with any argument I look at the initial premise because all of the evidence presented takes off from that. <BR/><BR/>I then look at what the premise is "missing" in its logic versus expected logic from actual conditions.<BR/><BR/>To cut to the chase, you very neatly wrote this while forgetting about time. It's as if you were thinking in two and not three dimensions.<BR/><BR/>In other words, runs are also called with respect to clock management and score. Coach Walsh was known for calling a variety of passes, working to establish a big lead, then running to "kill the clock." <BR/><BR/>If I look at the Niners and Stanford teams from your "two dimensional" perspective, then they look like balanced teams. But the reality is that's not true. <BR/><BR/>I think an adjustment in your otherwise well considered take is in order. I would restict my analysis to the first half of each contest to gain a better perspective of play calling patterns when the games are still in doubt. <BR/><BR/>More later.<BR/><BR/>ZZenniehttps://www.blogger.com/profile/05815765335808809176noreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152287372250130152006-07-07T11:49:00.000-04:002006-07-07T11:49:00.000-04:00Chris, this was a very interesting article that I ...Chris, this was a very interesting article that I read while sipping coffee during my vacation; very thought provoking. While your stats may indicate that there should be a premium placed on offensive balance (on the whole, something that I would agree with), I think that your numbers do not account for D&D and field position. Football is an entirely situational game, unlike baseball or any other sport. As a coach, I can say with absolute certainty that the value of yards gained in certain situations may be higher than another. For example Most statisticians will view two yards gained as two yards gained; nothing more and nothing less. However, the situation in which the two yards gained place a different value on those yards. Two yards rushing is more valuable on 3rd&1 as opposed to 3rd&9. The other situational factor left unaccounted for is one's field position at the beginning of the posession. From a raw numbers standpoint (raw yardage), a team that consistenty gets the ball at mid field or in their opponent's territory is generally going to have a lower output of yardage. This is even more pronounced in the red-zone, where the field is constricted and the advantage tactically goes over to the defense. Again, a two yard run on the +2 yardline is of higher value than a two yard run on 3rd&9 from the -28yardline.<BR/><BR/>With those facts in mind, as a coach, I am more interested in seeing how efficient my team's offensive output is relative to what the D&D situation is(run/pass separated). On 1st and 10 I want our gain per call to be at least 4 yards. On all 2nd downs, I want our gain per call to close no less than half the distance needed to convert. On 3rd & 4th down, we want to gain the yards needed for a 1st down or a touchdown. Any turnover or penalty counts as a point against. Although this analysis fits my needs from a general standpoint, I will also break this analysis into field position categories: 0 to -10; -11 to +30; +29 to +15, +14 to +6; +5 in. On the whole, an offense that is efficient in all situations is going to be on the winning side more often than an offense with a high output of yardage.<BR/><BR/>In closing, while I'd like to state again that although I thought your analysis is very thought provoking, I think that your numbers do not factor in the value of efficiency in situational yardage; and those situations are precisely what detemines whether or not an offense is contributing to a winning performance of a team.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152276664214692642006-07-07T08:51:00.000-04:002006-07-07T08:51:00.000-04:00Chris, this article got me thinking about the pass...Chris, this article got me thinking about the passing premium last night and the cost of a turn over. In your opinion what is the equivalent yardage that you should penalize a QB's passing yardage for an interception. I imagine using this number to subtract from total passing yardage to create a more apples to apples yardage comparision with running. My guess is that 50 yds is about the right deduction for every turnover. Using that number you could adjust the yardage so that the passing premium might disapear and also explain why a particular team might have a higher or lower premium given their propensity to throw interceptions.<BR/><BR/>Check out this site I found this morning. It seem to indicate that interceptions and fumbles are about equal. <BR/><BR/>http://www.sportsquant.com/turnovers.htm<BR/><BR/>It also supports my 50 yds assumption. They estimate the value of a turnover at ~4 points. Since previous work indicates that every 14 yds of field position increases your expected points on the drive by a point I get 4*14=56 as the appropriate yardage deduction for a turnover.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152248002306770582006-07-07T00:53:00.000-04:002006-07-07T00:53:00.000-04:00One note I'd like to make is I'm not entirely sure...One note I'd like to make is I'm not entirely sure what the correct "passing premium" should be. I just know there should be one, and that several major college teams had rather small ones. The NFL guys had premiums around 3, though the better offenses were closer to 2 yards. <BR/><BR/>At some point I'd like to hang a number on a passing play's "turnover risk" per play, where turnover risk is the risk of an interception, the higher than normal risk of a fumble, and an incompletion's contribution to a higher chance of a long-yardage 3rd down that won't get converted. Once at least a rough value for "turnover risk" can be established then I think we can get closer to figuring out exactly what the passing premium would be (another way is to just run regressions of the passing premium to total points and wins).<BR/><BR/>Sto: Unsurprisingly, I disagree with you. First, I just pulled very rough numbers and I admitted that the numbers may be off since this analysis really matters most for 1st and 2nd down. That said, I don't think the 20 3rd and shorts per season will materially affect the numbers for a team that ran as much as Minnesota, and you failed to mention all the 7-8 yard draw plays they ran on 3rd and 10. Last, the QB's lack of ability did not get translated into the stats, so maybe, just maybe, they underused him by a couple passes a game.<BR/><BR/>Second, my theory applies quite fine to the great Nebraska teams. According to cnnsi, in 1997 Nebraska had a 6.2 run average on a whopping 755 rushing attempts, and a 7.9 pass per play average on 182 attempts. A passing premium of 1.7 makes perfect sense to me for a team that ran for 6 yards a carry. I'd bet that with Frazier in '95 NU's rushing per play was even higher. NU hardly disproves this approach, in fact it reinforces it. NU was a lot better at running and offense than Minnesota was last year, and--surprise!--even more balanced.Chrishttps://www.blogger.com/profile/07204245083374821812noreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152231688740570332006-07-06T20:21:00.000-04:002006-07-06T20:21:00.000-04:00I've got to be honest, I think this data is useles...I've got to be honest, I think this data is useless. The theory itself may have some merit, but to come to a conclusion that Minnesota should have passed more due to their yards per pass play being greater than yards per running play - well that is ridiculous and does not take into account the poor abilities of the quarterback, and the fact that Minnesota would often run for very short yardage late in games to run the clock out.<BR/><BR/>Minnesota isn't the only team like this, the old Nebraska teams would be extremely successful at running the football and just good enough at passing to be dangerous. Similarly, I would expect NU's yds/pass play to be greater than yds/running play. <BR/><BR/>Another thing is that I don't think you can manipulate these numbers in order to score yourself more points per game (3, as you noted). That seems like a huge reach. We all know that defenses toughen up in the redzone and taking advantage of this information isn't going to somehow help you convert a 3rd and goal at the 1, turning a FG into a TD, for example.Legacy Userhttps://www.blogger.com/profile/00531037000620852691noreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152221035464950892006-07-06T17:23:00.000-04:002006-07-06T17:23:00.000-04:00Chris, I can see how the defenses could be a littl...Chris, I can see how the defenses could be a little lumpy and how that maybe could lead to a reduction in running, but I think that assumes that you can only play one defense every down. I think if instead the defenses response is just to increase the number of plays that it calls the run defense relative to the pass defense I think the response gets nearly continuous and a lot less lumpy. <BR/><BR/>I think you actually wrote an article about this so called "mixed strategy"<BR/><BR/>Theoretically in this situation the number of runs should increase, but in practice Defensive coordinators may take more of an all or nothing approach. If they took that strategy or something that approached it you are correct that the number of runs could also go down. A defensive coordinator employing this strategy likley would not be acting optimaly but there is nothing in practice to say that your opponent has to be fully rational. <BR/><BR/>Anyhow, not meaning to pick nits, but just enjoying the discussion.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152215458371930712006-07-06T15:50:00.000-04:002006-07-06T15:50:00.000-04:00The other missing variable is the role you expect ...The other missing variable is the role you expect your offense to take. Some offenses serve as almost placeholders aiming to hold the ball for as long as possible while grinding clock and hoping the other team makes a mistake. <BR/><BR/>Those teams should have lower per-play yields--like Urban's offense, for example. There's no way they should approach Tech's YPP on passing ever, since they're designed to pound on the ground and nibble away yards with short passing.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152209268814470982006-07-06T14:07:00.000-04:002006-07-06T14:07:00.000-04:00Brad: Thanks for the kind words. I agree with your...Brad: Thanks for the kind words. I agree with your stance, and to some extent I wrote about it in the way I did for rhetorical shock value to get people thinking. That being said, I do think it is possible for a team to get better at running and wind up running less, both from a game theory and a practical standpoint.<BR/><BR/>Most defenses cannot scale their focus continuously, and instead shift discretely. For example, imagine a defense had three options on first and 10: balanced (base 4-3), run defense (8-9 man front), and pass defense (7 man front). The team that averages 6 yards a rush and 7 yards a pass may face the balanced defense. Then, when the rush improves to 7 yards a carry, and the offense begins to run more and more, the defense then stacks the front and the O now must face the run defense.<BR/><BR/>The expected values against the run D, even with the improved run game, could end up being something like 5 yards per rush and 9 yards per pass, in which case they would pass more and run less than they did when their running game was not as effective. I suppose one could kind of label this a theoretical approach to "take what the defense gives you."<BR/><BR/>Anyway, your point is well made though and I do agree, and is still fairly counterintuitive: if your run game improved by 10%, you'd run more, but by some percentage less than 10%.<BR/><BR/>Anon: I briefly looked at some NFL teams and their passing premiums range quite a bit higher, but I think there are reasons for this. On a cursory review of the playoff teams, the premium appears to hover around 3 yards, with 4 being the high (Colts) and 2 being the low (Seattle). This does imply that NFL teams undervalue the passing game, which is consistent with what some of the other forward looking stat people have been saying, like Football Outsiders, etc. <BR/><BR/>That said though, I cross checked a couple stats. First, it appears passing is riskier in the NFL from an interception standpoint: NFL teams had interceptions per pass ratios of around .04, give or take, while even teams like Texas Tech in college had ratios around .02, or half that. Also, injury problems at the QB position plague NFL teams. More passes exposes QBs to more hits and more injuries, and this alone might be enough to tilt the playcalling to running.<BR/><BR/>Nevertheless, the most balanced teams had the best offenses, like Seattle. The outlier was the Colts, but their passing game has been so good they score plenty of points, they won so many games and put teams away early (which is rare in the NFL), and seemed to be preparing all year for a possible cold/bad weather game in NE.Chrishttps://www.blogger.com/profile/07204245083374821812noreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152199253628458662006-07-06T11:20:00.000-04:002006-07-06T11:20:00.000-04:00Hey great article Chris. I was wondering if you'v...Hey great article Chris. I was wondering if you've looked at how these premiums would compare to NFL teams?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-15377771.post-1152198467729718022006-07-06T11:07:00.000-04:002006-07-06T11:07:00.000-04:00Chris, as always I love the analyis and have alway...Chris, as always I love the analyis and have always thought about balance in this way.<BR/><BR/>I do have one slight issue/correction in your analysis.<BR/><BR/>If a teams running game improved from one year to the next they would not in equilibrium reduce the number or running plays. They would run more and pass less, but both their average yards per pass and run would increase.<BR/><BR/>Lets say one year you ran 25 times and passed 25 times and averaged 6.0 and 7.0 yards respectively 1 yd pass premium.<BR/><BR/>The next year you have a very strong running game but you are still the same at passing. ie if you ran the ball 25 times this year you would average 7 yds per carry and if you passed 25 times you would still average 7 yds per carry.<BR/><BR/>Now lets assume that every additional time that you run the ball beyond 25 your yds per attempt goes down 0.10 and every carry less than 25 your yds per goes up a similar amount. Let also assume that this same relationship holds for passing.<BR/><BR/>To get back to the 1 yd passing premium you assuming that the number of plays is constant you would need to run the ball 5 more times and pass 5 less. In this case You would averge 6.5 yds running the ball and 7.5 passing and run 30 times and pass 20. The key is you allways will run more.<BR/><BR/>Intuitively this is why Texas Tech passes more and Minnesota runs more but they both have have pass and run averages close to each other. The reason is that TT has is better overall at passing were they both to runn the same number of passes.<BR/><BR/>This same logic carries over from one season to the next.<BR/><BR/>This may all be obvious to you, but I thought you seemed to be saying that a team thats running game improved would counter intuitively run less. I appoligize if I misunderstood.<BR/><BR/>Anyway I love the article and the site. Keep it coming!Anonymousnoreply@blogger.com