tag:blogger.com,1999:blog-38600807.post5328341139366992012..comments2023-11-05T04:16:44.937-05:00Comments on Advanced Football Analytics (formerly Advanced NFL Stats): Deep vs. Short PassesUnknownnoreply@blogger.comBlogger36125tag:blogger.com,1999:blog-38600807.post-11126320444023609572013-02-11T22:49:21.967-05:002013-02-11T22:49:21.967-05:00There are other factors not accounted for and the ...There are other factors not accounted for and the term "deep pass" is a broad one. I believe that fades, 9 routes and deep corner routes are "safer" than seam (skiiny posts) and posts. The deep side line routes I mentioned at the outset of this post are safe routes in that, provided the proficiency of the passer (and we would certainly hope that most NFL starters are proficient) they can use the sidelines as oppossed to throwing deep or even intermediate middle where there is a lot of clutter<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-23960038268427924352010-12-16T13:30:39.026-05:002010-12-16T13:30:39.026-05:00Search "Air Yards" on this very blog, an...Search "Air Yards" on this very blog, and you get lots of material on this subject. Here's a good post on this:<br /><br />http://www.advancednflstats.com/2007/08/who-gets-credit-for-yac.htmlTarrhttps://www.blogger.com/profile/14368810359650066790noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-91998631538947576682010-12-16T12:20:30.490-05:002010-12-16T12:20:30.490-05:00Tarr, I'm unfamiliar with that evidence re YAC...Tarr, I'm unfamiliar with that evidence re YAC. Can you (or anyone) point me in the right direction? Thanks.Bigmouthhttps://www.blogger.com/profile/04885083460724621786noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-10043147953422787832010-12-16T12:01:07.244-05:002010-12-16T12:01:07.244-05:00I know I'm late to the party here, but a coupl...I know I'm late to the party here, but a couple comments:<br /><br />1) Frankly, I don't think the analysis here tells us anything meaningful at all about whether teams should pass deep more often or not. As you mention, "if a team is passing deep, several things have already gone right". You need pass protection to hold up, and someone has to at least seem to be open downfield. Many short passing plays are simply check downs that result from aborted attempts to pass deep. These factors will heavily bias the result unless we are able to control for them. Including sacks in the deep pass data is far from sufficient, in my opinion.<br /><br />2) Is there anywhere where QB's "Air Yards" are compiled? The only place I can even find YAC data for receivers in a public spot is on Yahoo sports, and that's only active players.<br /><br />Since all the evidence we have suggests that the large majority of the credit for YAC should go to the receiver, not the QB, it seems like we should be able to use this as a basis for dividing EPA/WPA shares between the QB and the receiver. (Sorry, O-line, you're still out of luck.)Tarrhttps://www.blogger.com/profile/14368810359650066790noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-61753708628952938152010-12-09T12:26:57.504-05:002010-12-09T12:26:57.504-05:00I really think the sack data -- and attendant inju...I really think the sack data -- and attendant injury risk -- is a big question mark in this analysis. There's a reason Mike Martz's offense, with its ubiquitous seven-step drops and long routes, has fallen out of favor. It's hell on the quarterback.Bigmouthhttps://www.blogger.com/profile/04885083460724621786noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-53264320754815558552010-12-08T02:52:11.651-05:002010-12-08T02:52:11.651-05:00There may be a way. On a pass play, the intended t...There may be a way. On a pass play, the intended target receiver is listed in the play-by-play. If there is no target, it's generally a thrown-away ball, or sometimes a tipped pass. The 'untargeted' passes could be excluded.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-47911744693594258112010-12-08T02:30:41.190-05:002010-12-08T02:30:41.190-05:00I wonder if theres a way to control for throws on ...I wonder if theres a way to control for throws on plays that are meant to be deep throws, but are covered and then the QB has to get out of the tackle box and throw it away to avoid the intentional grounding, which should show up as an short incomplete pass. Doubt it would have too much of an effect, but the current system essentially takes that from an unsuccessful deep pass to an unsuccessful short pass, making the deep play look better than it really should be while making short plays look worse. Again, this doesn't happen enough to really explain the discrepancy, but if this happens a few times a game, could explain why it may be logical for teams to use the mix they currently do, especially since that should improve the SR of short passes relative to long.TheGreatFatMannoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-19900993347584497912010-12-06T12:01:51.121-05:002010-12-06T12:01:51.121-05:00Brian,
why don´t you invest 200 $ to buy the play...Brian,<br /><br />why don´t you invest 200 $ to buy the play by play charts from FO? They started this game charting in 2006 i think. They have all passes with length of throw and YAC.<br /><br />BTW, i think deep passing sacks can be calculated to a certain point. If you look at stats when teams are behind in the 2nd Half, they throw more deep because they are sacked way more often. Or you look at 3rd and Longs, where sacks are also more up.<br /><br />Greetings from Germany, Karl.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-9003392942148948372010-12-06T10:01:13.290-05:002010-12-06T10:01:13.290-05:00"I think one of the most underused tactic on ..."I think one of the most underused tactic on 3rd and long (or very long) is to throw it up for grabs and let your WR make a play for it."<br /><br />Really interesting idea. We're conditioned to think "interception = bad," but in some circumstances not that different than an incomplete.Guynoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-38497610427386905062010-12-06T01:10:43.085-05:002010-12-06T01:10:43.085-05:00I'm curious to see what happens if third downs...I'm curious to see what happens if third downs are removed. Like a poster above said I imagine a lot of negative EPA for short passes comes from dumpoffs on third down. <br /><br />But maybe EPA is also proportionately negative for long passes that occur on third down. It would be interesting to know. <br /><br />Also I think the success rate of long passing would take a fairly big hit if the majority of the sacks were attributed to long passing. <br /><br />I agree that throwing it up deep on 3rd and longs might be slightly underutilized but the defense is also defending against this fairly often.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-86164359718443123452010-12-05T04:47:31.454-05:002010-12-05T04:47:31.454-05:00I don't disagree. However, the result could ha...I don't disagree. However, the result could have been the reverse, but wasn't. The data could say that deep passing was less efficient and the added complications of deep attempts would probably give them net negative value. <br /><br />Also, the data show that short passing is not much more effective than running. The main conclusion of the article is the finding that the known EP advantage of passing over running is from the deep attempts, not all passes. That's substantially more than "throwing for long yardage is better than throwing for short yardage.<br /><br />Additionally, this analysis includes penalties, such as holding or PI on both sides. It also accounts for down/dist/yl situations. I think one of the most underused tactic on 3rd and long (or very long) is to throw it up for grabs and let your WR make a play for it. A very deep int is not much worse than an incomplete pass, a sack, or a checkdown that has little chance of success. Plus, there is the opportunity for a PI call.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-90871095175205651862010-12-04T22:04:19.964-05:002010-12-04T22:04:19.964-05:00Fair enough: maybe 20% or more of sacks were shor...Fair enough: maybe 20% or more of sacks were short plays. But against that you likely have a huge number of intended long passes where the QB couldn't find a man open and threw short instead. And what about the many passes where the QB, unable to find an open receiver, deliberately throws it away? Won't almost all of those be recorded as a short attempt? <br /><br />The fact that the completion% is basically the same tells the story: it can't possibly be true that completing long attempts is as easy as completing short attempts. So many, many intended long plays are clearly not visible to you here. (And given the number of plays where the QB probably has both a long and short option, maybe this isn't even possible to analyze in these terms.) I'm not sure this data is telling us anything more than "throwing for long yardage is better than throwing for short yardage."Guynoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-88827182121039934862010-12-04T17:04:38.231-05:002010-12-04T17:04:38.231-05:00Certainly short pass plays have much greater than ...Certainly short pass plays have much greater than zero percent of the burden of sack risk. Whatever that burden is, it subtracts utility from the short category and adds utility to the long category.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-23488341948248104922010-12-04T12:56:19.342-05:002010-12-04T12:56:19.342-05:00Why is the long/short pass divide at 15 yards? Wou...Why is the long/short pass divide at 15 yards? Would it make more sense to group the quick-out WR screens with runs instead of passes? To me, defensing a 5 yd pass would have more in common with defensing a run that gets past the line of scrimmage than defensing against a 10 yd pass. Based on your comments, I'd like to see WPA/EPA that categorizes a play as in front of or behind the linebackers, regardless of whether it is a pass or run.<br />I would also surmise that a defense's success against the long pass is more an element of random luck (as long as the QB has time to throw)than success against the run or short pass. There is simply too much field to defend.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-54613410466032934342010-12-04T09:50:44.549-05:002010-12-04T09:50:44.549-05:00Brian: If you assume the sacks all count against ...Brian: If you assume the sacks all count against deep attempts, then 95% of the positive value of deep attempts is gone. Yes, it's still a net positive, but a pretty small advantage. Why do you feel so confident that long attempts have a much higher payoff? <br /><br />In fact, it's easy to imagine that the remaining advantage is explained by plays where the intent was to go long but the QB didn't see an opportunity and so settled for a short attempt (or short run) instead. It seems possible, maybe even likely, that if we knew the intent of every play before the snap, long attempts would have no advantage at all. On the flip side, some short attempts are probably a response to a blitz when the intent was a long attempt.Guynoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-17284892883240839232010-12-03T23:44:17.081-05:002010-12-03T23:44:17.081-05:00The aborted plays makes sense. Thanks for the exp...The aborted plays makes sense. Thanks for the explanation.Jeff Clarkenoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-36066920954692750592010-12-03T23:34:46.307-05:002010-12-03T23:34:46.307-05:00That makes sense. I'll do that.That makes sense. I'll do that.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-64531126574036250132010-12-03T23:25:39.638-05:002010-12-03T23:25:39.638-05:00Could you look separately by down? I'd guess ...Could you look separately by down? I'd guess that short passes on 1st and 2nd down have positive EPA; it's just the 3rd down short passes that have negative EPA because of all the 3rd & long dumpoffs.Vincenoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-90541245095082214662010-12-03T22:23:36.662-05:002010-12-03T22:23:36.662-05:00Mike Wallace as Steelers MVP comes to mind on this...Mike Wallace as Steelers MVP comes to mind on this one.Dean Jacksonhttps://www.blogger.com/profile/05862185746791036769noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-47560572676287061252010-12-03T17:23:01.202-05:002010-12-03T17:23:01.202-05:00Speaking of PI penalties.. I recall a quick and di...Speaking of PI penalties.. I recall a quick and dirty study of drives on which defensive PI penalties occurred, which found that drives on which a PI penalty occurs are MUCH more likely to end in a score, and disproportionately, in a touchdown. IIRC the avg NFL drive results in a score just under 40% of the time and in a TD about 20% of the time. Throw in the defensive PI penalty and those numbers double (actually the TD likelihood goes to just under 50%, more than double).<br />I guess my question is, if you factor OUT the defensive PI penalties, how much EPA does the "long pass" strategy lose ?JJBhttps://www.blogger.com/profile/10207257133625490532noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-47696350858725878652010-12-03T17:11:27.693-05:002010-12-03T17:11:27.693-05:00Have you looked into whether success on deep passi...Have you looked into whether success on deep passing plays increases the rate/extent of success of run plays and short passing plays?<br /><br />Deep passes often force "off" coverage and safeties out of the box. That might mean that not only are deep passes more effective, but successful deep passes improve the offense overall.John Morgannoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-5597841274840638872010-12-03T16:44:03.177-05:002010-12-03T16:44:03.177-05:00Yes, this includes all downs.
Chris-Yes, penaltie...Yes, this includes all downs.<br /><br />Chris-Yes, penalties are included in the analysis.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.comtag:blogger.com,1999:blog-38600807.post-53370446860626411312010-12-03T16:31:30.473-05:002010-12-03T16:31:30.473-05:00Hey Brian, did you include plays with PI penalties...Hey Brian, did you include plays with PI penalties to your analysis? If not, that would further skew the advantage of going deep.<br /><br />Great study by the way.Chrishttp://nfl-forecast.comnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-86481620398259349002010-12-03T16:15:59.723-05:002010-12-03T16:15:59.723-05:00Actually, I think the reason that the run and pass...Actually, I think the reason that the run and pass EPA don't add to zero is simply because of whats not included, aborted plays (penalties) and fumbles. Those are huge negative EPA plays and, when all is said and done, should bring everything to zero.<br />I am more wondering about is specifically the passing .08 number; if our data set has all plays that turned out to be passes included, shouldn't we be able to add all the EPAs for those plays up and get back to .08?<br />Thats why i was wondering if maybe this is for all downs instead of just first down (because i think .08 coincides with the 1st down passing number you came up with a while ago).<br />Thanks again!<br />AndyAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-38600807.post-44881383240733243672010-12-03T16:05:31.860-05:002010-12-03T16:05:31.860-05:00Net pass EPA will add up to be positive and run EP...Net pass EPA will add up to be positive and run EPA is net negative. There are also plays that are classified as neither run nor pass, mostly 'aborted' plays, which are typically negative.Brian Burkehttps://www.blogger.com/profile/12371470711365236987noreply@blogger.com