So I’m a bit in between websites right now, thus I’ll only post this here. I know this ain’t exactly peak draft time, but anyone who’s interested enough in the draft to read about prospect stats will most likely see it. What I was striving to do was combine some key college metrics for QBs to come up with a final score. Football Outsider’s does something similar except using more box score and career oriented stats with their Lewan Forecast. I’m looking at play metrics exclusively. There are some things that aren’t included in here that I’d want to include for a final draft. I’ve been toying around with the importance of 3[SUP]rd[/SUP] down conversions and 1[SUP]st[/SUP] downs, those aren’t included in any way in this specific calculation. This is a rough draft and if you’re not into draft stats, this isn’t for you. I’ve broken the metrics into three categories: Difficulty of throws, Situational success, and Decision making. Difficulty: These are aggregated scores combining four distinct metrics that I’ve deemed go into the difficulty of throws: Average Distance of Pass, Rollout Completion Percentage, 6 – 15 yard comp %, 16+ comp % [TABLE="width: 576"] [TR] [TD][/TD] [TD]Bridgewater[/TD] [TD]Griffin[/TD] [TD]Luck[/TD] [TD]Wilson[/TD] [TD]Manuel[/TD] [TD]Boyd[/TD] [/TR] [TR] [TD]Score[/TD] [TD]75.63[/TD] [TD]77.76[/TD] [TD]68.83[/TD] [TD]75.60[/TD] [TD]67.21[/TD] [TD]71.12[/TD] [/TR] [/TABLE] It makes sense that RGIII had the highest of these scores, he was superb in college at throwing the deep ball and had the second highest average distance of pass behind Wilson. Bridgewater and Wilson don’t lag far behind RGIII, Bridgewater notably getting boosts from his 80% completion percentage on rollouts and his success in the 6-15 yard range. Manuel and Luck both suffered due to frequent short passes and average showings in the intermediate, 6-15 yard range. Situational This is where the third down and first down success rates would be added. For now it just consists of 3[SUP]rd[/SUP] Down Comp % (which can be misleading due to check downs) and Red Zone Comp %, [TABLE="width: 576"] [TR] [TD][/TD] [TD]Bridgewater[/TD] [TD]Griffin[/TD] [TD]Luck[/TD] [TD]Wilson[/TD] [TD]Manuel[/TD] [TD]Boyd[/TD] [/TR] [TR] [TD]Score[/TD] [TD]74.06[/TD] [TD]76.00[/TD] [TD]77.43[/TD] [TD]86.17[/TD] [TD]64.74[/TD] [TD]68.23[/TD] [/TR] [/TABLE] Russell Wilson pretty much cleaned up in this category, both his 3[SUP]rd[/SUP] Down Comp % and Red Zone Comp % were around 75% which were the top of all prospects in both categories. Combine that with his success in making difficult throws and we know that he wasn’t just checking it down to inflate these numbers. Manuel suffered the most in this category, both percentages hovered around 57% - the lowest in both categories. Luck is an interesting case, most of the prospect’s third down comp % and red zone are similar. Luck had inordinate success in the red zone (73.8%, it helps to have big TEs), and 61.7% on 3[SUP]rd[/SUP] down, only above Boyd and Manuel. Decision Making This is probably the most intriguing one to me as it includes my new favorite QB stat – percentage of throws into coverage (passes defensed and intercepted). A cursory look at past QBs (with admittedly low sample size) indicates it good have a high correlation with success. The aggregate score also includes completion percentage against the blitz. [TABLE="width: 576"] [TR] [TD][/TD] [TD]Bridgewater[/TD] [TD]Griffin[/TD] [TD]Luck[/TD] [TD]Wilson[/TD] [TD]Manuel[/TD] [TD]Boyd[/TD] [/TR] [TR] [TD]Score[/TD] [TD]69.62[/TD] [TD]71.47[/TD] [TD]57.49[/TD] [TD]100.00[/TD] [TD]66.16[/TD] [TD]60.50[/TD] [/TR] [/TABLE] Confession time, I had to fudge Russell Wilson’s decision making score. His throws into coverage were so much lower than every other quarterback (2.8% compared to an average of ~7.5%) that it broke my scale. His official score with this group was 118, but I rounded down to 100 to give everyone else a chance. In the other category Comp % Against the Blitz – no other quarterback broke 63.2% (Boyd and Bridgewater tied) but Russell Wilson was at 69%. Both Bridgewater and Griffin threw into coverage just under 6%, which is my unofficial qualifier for a successful quarterback (not proven in any way, just my intuition) and Bridgewater did well against the blitz. Luck takes a beating in this rating as he threw into coverage 8.66% of the time, worst with Boyd at 8.2% for second. Luck also had the worst completion percentage against the blitz at 60%. Overall: [TABLE="width: 576"] [TR] [TD][/TD] [TD]Bridgewater[/TD] [TD]Griffin[/TD] [TD]Luck[/TD] [TD]Wilson[/TD] [TD]Manuel[/TD] [TD]Boyd[/TD] [/TR] [TR] [TD]Difficulty[/TD] [TD]75.63[/TD] [TD]77.76[/TD] [TD]68.83[/TD] [TD]75.60[/TD] [TD]67.21[/TD] [TD]71.12[/TD] [/TR] [TR] [TD]Situational[/TD] [TD]74.06[/TD] [TD]76.00[/TD] [TD]77.43[/TD] [TD]86.17[/TD] [TD]64.74[/TD] [TD]68.23[/TD] [/TR] [TR] [TD]Decision Making[/TD] [TD]69.62[/TD] [TD]71.47[/TD] [TD]57.49[/TD] [TD]100.00[/TD] [TD]66.16[/TD] [TD]60.50[/TD] [/TR] [TR] [TD]Weighted[/TD] [TD]73.74[/TD] [TD]75.75[/TD] [TD]68.15[/TD] [TD]84.34[/TD] [TD]66.33[/TD] [TD]67.74[/TD] [/TR] [/TABLE] Surprise, surprise - Russell Wilson dominated. His final aggregate score nearly 9 points higher than RGIII. Even if I dropped his Decision Making score to a 71, he still has an aggregate score of 77, higher than all other quarterbacks. My metrics are a little bearish on Luck, while his situational score was solid the average showings in Difficulty and Decision Making knock him down to slightly higher than EJ Manuel and Tajh Boyd. Griffin and Bridgewater have similar scores, which makes me hopeful for Teddy. If Bridgewater had similar success in his sophomore year as RGIII did in his senior, Heisman year – that has to mean something positive. If Bridgewater puts up another year with similar numbers, I’d be willing to wager he’ll be very successful at the NFL level. I’ll put this simply and in bold. If a GM had these numbers before the 2012 draft and DIDN’T draft Russell Wilson, they should be fired. In 5 of the 8 categories I used, he had the highest metrics of all QBs I studied. When I extend to other categories I didn’t use, it’s 7 of 10. There were no indicators in any of these numbers that he wouldn’t be a successful QB. Now for random notes. I’m going to try to keep this sort of stuff coming, whether on here or other websites. It’s going to be tough, so it may come sporadically. I recently graduated college and had positive vibes from the Bucs as well as STATS that I might be able to keep working in this field. Alas I seem to be SOL on both counts, so I’ll be doing the best I can to deliver this kind of stuff while finding a job. Thanks for reading this and all my past word, I’ll try to keep getting other stuff up.
Interesting, I'd like to see the formula dated back a little further to include QBs like Cam Newton, Blaine Gabbert etc. Obviously having the best QB prospect to come out since 1983 as 4th will raise doubt on the formula, but in life we always have the exception to the rule. I would argue Luck's decision making is in reality far better than his culprits, he just has the ability to actually make throws into coverage unlike some of the others. Interesting stuff though.
He did end up third best out of the three who were rookies last season. At this point calling him a prospect is nonsensical. He has data now. Data that supports the hierarchy. So far at least. Sent from my GT-P3110 using Tapatalk 2
Interesting stuff. Nice work too man. Thanks NU. This is the kind of info that I'd use to either validate what my eyes see or give me a reason to go back to the film to see if I missed anything. I wouldn't use it as an end-all be-all b/c of all the different variables involved, but it's definitely worthwhile information. For instance, as far as "Difficulty" goes, I know that Boyd & Luck ran more complex offenses with greater responsibility placed on the QB's shoulders than Bridgewater, attempted more NFL caliber throws, faced slightly tougher defenses with better front 4s overall, and contributed to more points (Clemson's offense ranked 6th in scoring compared to Louisville's 51st). I also feel the "Difficulty" category should be influenced by a few things: The QB having less offensive responsibility (which Teddy obviously did) attempting a higher percentage of short throws (like Teddy's 40% coming in the 6 yard and under range). Prematurely opting for checkdowns (which Teddy often did) which contradicts "difficulty" and makes it easier to post a higher completion %. Attempting less than par amount of passes into tight coverage (which can also lend concerns for timidness or lack of trust in one's arm). This can misleadingly contribute to a higher completion %. (for instance, Teddy posted a higher redzone completion % but Boyd threw 1 more redzone TD despite 25 fewer attempts, plus had 0 redzone INTs versus Teddy's 1). You go ahead and take Teddy & his higher redzone completion % and I'll take Boyd's 21% higher RZ scoring efficiency. Not attempting many legitimate downfield passes. It's easy to notch a higher completion % in the 16+ yard range when you attempt very few passes over 25 yards. Just sayin. What does Teddy's 16+ yard percentage drop to if he attempts more of those 25+ yarders that he seemed fairly poor at? Completion percentage under pressure. The "Situational" & "Decision Making" metrics are somewhat misleading and are in need of some tweaking b/c there's one important stat missing from this breakdown- Scoring Efficiency. While Tajh's 3rd Down & Redzone completion % may be lower than Teddy's, his Scoring Efficiency Percentage [TDs per attempt] is significantly higher, throwing 9 more TDs than Teddy despite attempting only 8 more passes. 8.43% versus 6.44%. That's a 31% higher percentage. 3rd Down & Redzone Completion % is great but at the end of the day the QB's drive success rate and points he puts on the board are more valuable than the 3rd downs converted. For instance, Teddy could covert 80% of his 3rd downs but if he can't efficiently finish off drives then those 3rd downs aren't quite as significant. I'm not saying Teddy is bad at scoring efficiency but I am saying the 3rd downs sometimes need to be taken with a slight grain of salt when comparing QBs side by side. Drive success rate and the ability to string together multiple 3rd down conversions that lead to points is more important than 3rd down completion percentage alone. The "Decision Making" metric needs adjusting b/c as of now it seems you're only focusing on the negative aspects of it- passes thrown into coverage that are defended/intercepted. What about the QB's increased level of Scoring Efficiency because he trusts his arm and is willing & capable of throwing more passes into tighter windows/coverage even though the risk is higher and will inevitably lead to more passes broken up and intercepted because of it? Marino, Favre, Elway, and Manning would likely be poor decision makers according to this metric. Conversely, guys too timid to make some of throws will be falsely rewarded by this metric, and timidness or increased affinity for checking down does not = better decision maker. Furthermore, if Luck & Boyd throw into traffic 8.7% & 8.2% of the time but only amass 10 & 13 interceptions, I'd say that's an impressive testament to their overall passing accuracy, decision making, and ability to think quickly, especially when considering that 8.7 & 8.2% come with 9 more TDs than Teddy. Luck & Boyd throw a greater percentage of passes into tighter windows not because they're poor decision makers but because they trust their arms and have coaches who trust their arms & decision making ability too, so they're encouraged to take chances b/c the overall net affect is worth it (See Boyd's 1 TD per 11.9 attempts versus Teddy's 1 per 15.5). If it weren't for Boyd entrusting his arm and throwing into coverage, Clemson would've never had a 4th quarter comeback vs LSU. How many critical completions in those final 4 minutes would've never been happened if Tajh didn't trust his arm enough to throw into tight windows and was instead more concerned with INTs and passes getting defended like your decision making metric seems to reward? 3:36 mark- with pressure in his face on 3rd & 11. The TD at 4:17 might not have happened. (Bridgewater would've likely prematurely tucked and run after seeing his first read not open, just as he frequently did all season). 6:39- 4th & 16 facing a blitz. (Bridgwater faced 3rd & 26 in a similar last minute situation vs Cincy and quickly opted for the check down which resulted in the game going to OT rather than Louisville having a chance to win it in regulation with a FG). 8:38- tight window throw to Hopkins allowed for the pass interference call and 10 extra critical yards. 9:37- huge 1st down completion just over the leaping safety's fingers into game-winning FG range would NOT have happened. [video=youtube;Gw-bVCshojY]http://www.youtube.com/watch?v=Gw-bVCshojY[/video] Now, if you're Bill Parcells & Dan Henning and are looking for a game-manager at QB rather than a guy capable of carrying an offense, then I'd certainly put more merit in your current decision making metric, but as far as today's passing game goes, it needs adjusting. BTW, what about the "Ability Under Pressure" metric where you measure stuff like percentage of pressures evaded, completion % under duress, completion % on the run under duress, percentage and number of pressures evaded for positive yards [via the feet], and big plays under duress (like 3rd down conversions, TDs, and chunk yardage plays)?
I'm gonna spend a few hours tonight or tomorrow reviewing every 3rd Down pass attempt by Bridgewater & Boyd to see what percentage of those attempts led to scoring drives (because a 3rd isn't very effective if the drive itself ends in a punt). A few interesting stats omitted from the metrics of the opening post but do influence those metrics: completions of 25+ yards: Boyd- 48, #1 in FBS (Teddy had 34). Tajh attempted more downfield passes so it's understandable that his 16+ yard completion % is barely higher than Teddy's and his overall completion % is lower. That's just the nature of the beast. Scoring Efficiency [TD per attempt %]: Boyd- 8.43%. (31% higher than Bridgewater) Redzone scoring %: Clemson w/ Boyd- 94.64%, #1 in FBS (25 TD by Boyd). Louisville w/ Teddy: 92.31%, 6th. (17 TD by Teddy) Total points per game: Boyd- 21.54, #2 in FBS. (Bridgewater- 12.92, 39th) Total 3rd Down Conversion % (passing & rushing): Boyd- 48.4%, 76 1st downs, 14 TD. (Bridgewater- 48.1%, 64 1st downs, 5 TD) Total 1st Downs: Boyd- 213. Bridgewater- 183. Rushing yards: Boyd-514 (Bridgewater- 26) rushing 1st downs: Boyd-50 (Bridgewater- 18) rushes of 10+ yards: Boyd-27 (Bridgewater- 7) The rushing stats for some reason get omitted in these evaluations despite this ability being a factor in the NFL. It's significant in this case b/c Boyd is anything but a run-first QB, and these stats support how effective Boyd is at evading pressure and creating something from nothing, where as Teddy has mobility but lacks consistency in using it under pressure, and the film supports what these stats show. Personally, I could care less that Tajh threw 8.2% of passes into coverage or whatever the stat is. That stat is insignificant in Boyd's case b/c he's operating within the confines of the offense and coach's expectations, and that 8.2% didn't contribute to blown games; however, having a QB with the confidence & ability to thrown into tight windows did contribute to more wins.
Of course this formula is imperfect. It's a rough run at some of the more interesting metrics I had. I also omitted some if I thought receivers could influence the outcome drastically. For instance I left out third and first down stats because I couldn't account for receiver YAC (right now). To use those, I'd want to chart every throw to determine how much impact receivers had on conversions. Obviously when you have Sammy Watkins averaging an obscene 8 yards after the catch, it'll affect the numbers. Same for Kendall Wright/ RGIII, etc. Some interesting stats and thoughts, I have some comments on a few. "Tajh attempted more deep passes than Teddy" >> Tajh threw 88 passes, 16+ yards down the field. Bridgewater threw 81. Tajh's completion percentage on 16+ throws is extremely impressive at 48.9% and higher than Teddy's at 45.7%. That was accounted for just as strongly in the formula as any other factor and helped Tajh immensely. Tajh's main problem were on the intermediate throws 6-15 yards where he only completed 60% of his passes compared to an average of 66% against the other QBs. "Scoring Efficiency" >> An interesting train of thought. In the past I have ignored scoring in many of my metrics. With CBs, RBs, QBs I've thought that touchdowns (gained or given up) are aberrations. Whereas you might have 100 passes against the blitz you only have maybe 15-30 TDs. That doesn't mean I'm right, just how I've tended to think. I might consider merging it with another statistic, "successful" throws or something similar. I'll think on that. "3rd Downs and 1st Downs" >> I don't find total 1st downs or 3rd down conversions convincing on their own. Like I said earlier, I would really like to add these in some form or another in the future. I think they're incredibly important, but I don't want to add them until I have full down and distance details, YAC, throws beyond the first down marker, etc. "Rushing" >> Rushing is interesting. I don't know how comfortable I would be putting it into the formula because sacks are counted as negative yardage. Certainly big plays of 10+ yardage are intriguing. The one thing that needs consideration are designed runs. A player running the read option as a designed run play are effectively taking a designed run play from a RB. If a QB doesn't receive that a play call (Brandon Weeden?) to run they are effectively penalized for running a specific system. Every stat I have so far is mostly system adjusted. The deep completion percentages are weighted by their usage rate. The data to system adjust for read option/ scrambles/ etc just doesn't exist at this point in time. There's certainly some things to think about. Most notable scoring efficiency and down conversion rate. Appreciate your thoughts and thanks for reading.
Does that make the decision right? If you switched the order of the draft maybe not. I would probably take Russell Wilson. Sent from my GT-P3110 using Tapatalk 2
Good stuff, NU. I like where your heads at how you're trying to interpret the data. That's what I was hoping you were doing but wasn't fully certain until this post. I know what you're trying to accomplish so I wasn't actually knocking on you per say but rather providing more food for thought considering how big of a task it is to accurately use stats to assess an NCAA's QB potential. I think you're on the right track if it's any consolation. When I mentioned Tajh attempting more deep passes than Teddy it was in reference to the true downfield passing stuff where he led all FBS with 48 completions of 25+ yards. That makes his 48.9% completions in the 16+ range even more impressive, and based on the 6 games I watched of Teddy, if he attempted as many downfield throws as Boyd his 16+ yard completion percentage would be noticeable lower than 45.7%. well, some sort of metic needs to be used to assess a QB's affinity for stringing together drives and putting up points. I've seen every Tajh Boyd snap the past 2 years, live at that- so I've been fully aware of the exact particulars of the game upon each snap, and what I can tell you from this is that neither myself nor any knowledgeable Clemson fan was surprised at what Tajh did in the final 4 magical minutes of the LSU game. Was I excited as heck? -definitely. But not surprised. It was just Boyd being Boyd. If some type of scoring efficiency isn't attached to Boyd's metrics then it's not an accurate metic. I'm not saying that because he scores a lot; I'm saying that b/c he has a gift for making things happen and the mental strength to remain zeroed in throughout each drive. Good to know. In full agreement here. I'd like to see the percentage of 3rd down conversions that result in successful drives [FG & TD]. right, and QBs adept at avoiding sacks will negate some of that negative yardage, as well as pick up more 1st downs with their feet while under duress. I agree with you about the read option stuff however; that needs to be adjusted for. Boyd had a handful of designed runs, but much of what he did was on his own. He was near Russell Wilson-like at evading pressure and then using his feet to either keep plays alive for his arm or picking up positive yards with his feet. The 1st Down & TD stuff have an application b/c in the NFL they have an application too. If the QB's feet & mobility are a legitimate part of his makeup then the feet need to be accounted for in the metrics used. For instance, if you watch Bridgewater's highlights you'd be inclined to think he's great at avoiding pressure and then subsequently using his mobility to extend plays b/c he's athletic and had some impressive plays doing so. However when you watch Teddy's film (or at least the 6 games on Youtube) you see there's some inconsistency to it. One play he'll look like Houdini, and the next few he might look instinctively poor at it, doing stuff like turning right into the face of pressure while under duress even though an opportunity was present to make something of it. His inconsistency left me wondering if Teddy was just flipping a coin as to which way he'd move to avoid pressure rather than actually relying on instincts, where as Boyd seems to have eyes in the back of his head. To me, the disparity of their rushing stats validates what the film shows. Boyd might not be quite on Russell Wilson's level of mobility/escapability but it's close enough to say he's got special feet, and an exclusion of this from metrics meant to assess QB potential makes the metrics incomplete IMO, especially in today's game of mobile QBs. No problem. I'm glad you're aboard making the content of the draft forum all the better.
Curious whether you've expanded that study out to come up with scores for QBs other than the six guys in question. Would you post them?
I don't, unfortunately I don't have access to the data I would need to finish it up. However, if you check Club - I have posted a decent amount of data on QBs from the 2012, 2013, 2014 data classes. The data may or may not be interesting to you, it includes QB Incompletions and zone accuracy for the 2012 and 2014 classes. You won't get things like comp % against the blitz, but some interesting stuff nonetheless.
So, where does Tannehill rank? You included 3 of the QB's from his class, so I assume there is data on him?
Great work but there are two major things that are not accounted for that really bother me even though theres no way to truly account for them. One is Luck was heavily penalized for playing in a pro style system. That he put up those numbers in an offense that is far more advanced then anything the other qbs sans Wilson played in is more impressive than the stats. Two Luck was in a lot of close games while contending for a national title. That means he almost always took the other teams best shots. You (to no fault of yours) cant possibly cant account for that. But those are two major x factors that you cant sit and say Wilson should automatically be the best QB based on your stats is something I cant agree with.
Appreciate that kind of work Just from watching Wilson's reps he was every bit as good as any Qb in the draft, Luck with the edge just based on size, still, the height did not affect my draft value of him like it did for 32 GMs..which I'm still trying to understand.
Me too. You know what's alarming? E.J. Manuel has like a 96 passer rating through two games (higher than Tannehill). He has completed 68 percent of his passes and if you count drops that number goes to like 73 percent. He has only taken 1 sack. He has scrambled 6 times for 37 yards including that 9 yard scramble that put the Bills 2 yards away from the end zone in time for the game-winning touchdown against the Panthers with only a few seconds left on the clock. He came 2 points away from beating the Patriots, who had Danny Amendola playing in the game and entered the season favored to win 11 games. He upset the Panthers. Surprised there hasn't been more talk about him. That's about as good a start as it gets for a rookie unless your name is RG3. I had Manuel and Barkley as the only two QBs in the draft I would take and satisfy myself that I have my QB of the future...so I can't say I'm TOO surprised...but still!
To me it's: Bridgewater (some space) Tajh Boyd Zach Mettenberger (everyone else) Still not entirely sure about Zach Mettenberger. Right not it's directly underneath Boyd but with the caveat that if I see his TCU and Kent State games it could go lower. I am not a fan of Marcus Mariota. I don't see it there. I have never been a fan of Johnny Manziel. Brett Hundley could absolutely climb to the top level but the fact of the matter is he's a redshirt sophomore in his first three games of his second year playing and he's just not there yet, IMO. He's got talent. But he's also making brain fart decisions.
I didn't see it last year. But watch this and you might start to. [video=youtube;9P_vZ0oy75g]http://www.youtube.com/watch?v=9P_vZ0oy75g[/video]
that's why I posted to you saying "E.J Manuel #1 overall " b/c I thought he was heavily underrated but worthy of being the top QB taken.
.... which is stupid b/c Boyd is a strong vertical passer and IMO would fit Turner's offense well and definitely better than some names they mentioned.
To me it's still Boyd first, however if Teddy keeps up the improved play he's shown so far this year I'll have no problem putting him #1. I just wanna see more consistency first in the areas he was inconsistent in last year.
How did you feel about Boyd's showing last night? I thought NCSU gave him a little more trouble than he should have had. In fact if Brandon Mitchell had been playing in that game I wonder if the Wolfpack might have gotten the upset. It's obvious Pete Thomas is just an absolutely atrocious fit for Dave Doeren's offense, which produced juggernaut pass/run threats like Chandler Harnish and Jordan Lynch at Northern Illinois.
