FanPost

Mocking the Draft -- A Fresh Look at the Incoming QBs

Brian Spurlock-USA TODAY Sports

Those of you who have been around a while will recognize me. I don't have the time to participate often anymore, once or twice a year I manage to accidentally my way into something I think is interesting enough and valuable enough to share. Most of the time, I do this by messing around with spreadsheets attempting statistical analysis that is frankly out of my league.

And that's exactly how I ended up back here with you, today.

The draft is coming up and regardless of the 49ers needs, I always -- always -- find quarterback stuff to be by far the most interesting. Part of what makes it so interesting is that QB is the one position in football that typically has enough readily available data for amateurs like me to do interesting and vaguely reliable things with. I've also just had this sneaking suspicion that Marcus Mariota is the only can't miss QB in this draft, and that Jameis Winston -- in addition to all the other things he is -- is severely overrated as a prospect. I was waiting for Football Outsiders to do their Long-Term Career Forecast, which is a fascinating thingajig they do over there for QB prospectors like myself, and I just got tired of waiting.

Now, I like the principle behind the LCF, but I also find it a little hard to contextualize. Last year, for instance, they said the difference between Blake Bortles, who got a ~1,000 LCF and Jimmy Garoppolo, who got a ~1,500 LCF, was negligible, while all the while my puny brain was going 500 POINTS IS A LOT THAT MUST BE A LOT RIGHT? I don't know. I guess it's not a lot. It also has a few odd quirks to it, such as using games started (the Bill Parcells principle) instead of PAs (which offers bigger samples) as a measure of experience (and also not having a cutoff for games started, on the principle that too many college games started is actually an indication that a guy is not NFL desirable. This was the quirk that gave Aaron Murray the LCF win last year, with his whopping 2,000 LCF, which may or may not be a lot more LCF than Jimmy Garoppolo... I don't know...

(Or, he would have won had he been drafted in the first three rounds -- another LCF criteria -- which he wasn't.)

But I like the principle there. I also like similarity scores a lot, because they contextualize performance for me in a way that I understand. So I decided to create a set of similarity scores for basically all of the QBs eligible for this draft who I could find people saying were worth watching, based on the principles FO describes as statistically significant for college QBs. So, my similarity scores are based on 7 statistical categories:

  1. Height (in inches)
  2. Weight (in pounds) -- it is arguable I should have gone BMI like FO does, but that's a lot of conversions to make from the readily available info.
  3. Career Completion Percentage (one of the few college stats that predicts NFL success)
  4. Career AY/A (ANY/A would be a better efficiency measure, but college doesn't count sacks, so... -- I very rarely had to fudge these, as the numbers are only readily available for major school QBs)
  5. 2-Year AY/A Difference (FO only looks at the difference itself between Junior and Senior seasons, and uses an average difference for underclassmen, but I think the principle that demonstrated improvement from year to year is an indication of ongoing development, rather than peaking, is too important, so I used the difference from the last two college years. Occasionally, I did have to fudge around to make the samples stable.)
  6. Career Pass Attempts (this is my version of games played, which I like better because the samples are better)
  7. Yards per Rush (Also slightly different from FO -- this helps add similarity for running QBs, but also gives the system a way to account for sacks)

So, basically, I took down these 7 COLLEGE ONLY stats for every QB who has thrown a minimum of a single pass in the NFL since 1998 (my "modern era" cutoff because no active QB in the NFL was drafted earlier), and who also has these stats easily available, to build my bank. The idea is to compare the incoming QBs to this bank within these 7 categories to get my similarity scores. The closer a player is to any other player in each of these categories, within a viable range (for instance, no QB is 3 inches, so I count that as mathematically impossible), the closer to 1,000 the final similarity score will be.

Significantly, these similarity scores do NONE of the following.

  1. They do not account for professional stats, because I want to see how the college careers are similar. Also, the new guys don't have professional stats yet.
  2. They do not account for strength of competition or offensive system, because these are more scouting concerns than statistical ones.

Which brings me to one last but very important point -- these similarity scores are not meant to replace scouting. They could not do that if they tried. They are meant primarily to promote discussion and possibly to inform scouting, and they require a lot of interpretation besides. (This is also the point at which you'll be able to look in the comments and see exactly which people went "tfl;dfr" because they'll be the ones complaining about how dumb the statz are.)

All that said, I think it's pretty interesting how on the nose some of these results are, and how surprising some of the others are. Some recent risers, like Bryce Petty, actually do really well in this thing -- and had I done it earlier I may have even taken credit for predicting their rises. Do be aware as we go that because I only looked at QBs who actually threw NFL passes, there will be a more and more positive bias the further down we go. The bias won't be strong for Mariota, who is virtually guaranteed to throw a pass in the NFL, but it will be strong for a guy like Blake Sims, who is not.

One quick final note: I'm listing the top 5 similarity scores for each QB, and I have ranked them based on the average draft position of those 5 similar players. This shows a decent approximation of perceived talent rather than actual talent. A guy similar to, say, Tony Romo, will thus be ranked lower, but look better. This is where that being intelligent people who can interpret complex systems in interesting ways thing comes into play.

Without Further Ado:

1: Marcus Mariota Is Statistically Most Similar To...

  1. Andrew Luck (894)
  2. Alex Smith (882)
  3. Colin Kaepernick (876)
  4. Sam Bradford (861)
  5. Tim Tebow (860)

This jibes with a lot of what I've been thinking for a while. Mariota is most similar to 3 number-one picks who all became above average starters, and one project QB who became an above average starter, and Tim Tebow. These are AMAZING similarity scores. Mariota, to me, based on this AND what I've seen, is a can't miss guy who should clearly be the top pick in the draft. That he won't be is just dumb.

2: Jameis Winston:

  1. Brian Brohm (895)
  2. Blake Bortles (870)
  3. EJ Manuel (866)
  4. Sam Bradford (860)
  5. Drew Stanton (856)

This is a tougher one to interpret. We'll see guys like Brohm and Bradford show up a few times in these scores, in part because their college careers are statistically very typical of good prospects. Similarly, we will never see Russell Wilson with a high similarity to anyone, because he's so dissimilar from so many other successful players in NFL history. A few things are clear, though: We have a grab-bag of first and second round picks with none of the obvious "top pick" clarity we saw with Mariota. We also don't see the same kind of professional results with saw on Mariota's list. We shouldn't actually expect to see that, though, except for mega-prospects, because the NFL bust rate is so high. This, ultimately, helps confirms my original idea: Jameis Winston is a mid-to-late first round pick with uncertain potential, who is going to be extraordinarily overdrafted this year by some dumb team.

3: Garrett Grayson:

  1. Kellen Clemens (952)
  2. Kevin Kolb (938)
  3. David Carr (937)
  4. John Beck (935)
  5. Matt Stafford (914)

This is a fascinating grouping to me. Grayson is one of those risers I was talking about, and this collection of similar players shows why: He profiles statistically as a second-round passer (in fact NONE of those guys made it past round 2), but looking up and down those names his ceiling may not quite be so hot. Bryce Petty, the next on our list, profiles worse than Grayson, but with a visibly higher ceiling in his comps.

4: Bryce Petty:

  1. Brian Brohm (895)
  2. Tim Tebow (881)
  3. Tony Romo (877)
  4. Aaron Rodgers (875)
  5. Sam Bradford (875)

Okay, full disclosure: Except for Brett Hundley, I know almost jack about most of the remaining guys, Petty included. I know he's flying up draft boards right now. And I know this list makes me very very interested. The successes we see here were all long-term projects, and this is amazingly exactly what people keep saying about Petty -- that a team willing to let him grow for three or four years on the bench may be richly rewarded. It's one of those fascinating quirks where my system doesn't completely fail comparison. This suggests that Petty's perceived talent is worse than Garrett Grayson's, but suggests that his ultimate ceiling may be higher even than Jameis Winston.

5: Brett Hundley:

  1. Geno Smith (926)
  2. EJ Manuel (906)
  3. Andrew Luck (906)
  4. Bruce Gradkowski (900)
  5. Brian Brohm (894)

Hundley's similarity scores are particularly weird. Luck stands out only a little less than Geno Smith, but it's also hard to square Andrew Luck and Geno Smith being posed as statistically similar. There's just, like, this massive cognitive gap that my brain refuses to leap with that one. But, hey, thems what the maths say. I think the takeaway really should simply be that Hundley is a volatile prospect, who is as enticing as he is frightening. Which also... huh... that also squares with the scouting.

6: Cody Fajardo:

  1. Andy Dalton (889)
  2. Christian Ponder (885)
  3. Drew Stanton (880)
  4. Colt McCoy (880)
  5. Bruce Gradkowski (880)

I'll have less commentary from this point on, just because I don't know these players as well. I'll be very interested to read the comments to learn more. What does this tell me, though? Hey, this guy could be a pretty solid backup.

7: Taylor Kelly:

  1. Andy Dalton (953)
  2. Austin Davis (909)
  3. Teddy Bridgewater (906)
  4. Kellen Clemens (901)
  5. Aaron Rodgers (900)

This looks a lot better than Fajardo, but that's the potential ceiling talking. The average perceived talent based on those comps is still lower than Fajardo. Even so, go ahead and rejoice Taylor Kelly fans.

8: Connor Halliday:

  1. Eli Manning (872)
  2. TJ Yates (862)
  3. Chris Redman (857)
  4. Rusty Smith (856)
  5. Tim Couch (849)

Not great, Connor Halliday. But on the other hand, I love this list. Chris Redman! Rusty Smith! Tim Couch! I gave us all a reason to think of these three people today, and this is literally something. So thank you, Connor Halliday.

9: Brett Bonner:

  1. Mike Glennon (846)
  2. Tom Savage (823)
  3. Brock Osweiler (810)
  4. Andrew Walter (806)
  5. Ryan Mallett (806)

Tom Savage! Andrew Walter! Those are definitely people! The world is amazing.

10: Devin Gardner:

  1. Aaron Brooks (916)
  2. Ryan Tannehill (914)
  3. Craig Krenzel (908)
  4. Quincy Carter (905)
  5. Tom Brandstater (889)

Did you know that Craig Krenzel, Rex Grossman, and Drew Hutchinson made up the first ever QB black hole of Lovie Smith's head coaching career? Do you fondly remember Quincy Carter because of how bad the Cowboys were back then? Do you even know who Tom Brandstater is? Thank you for your bountiful gifts, Devin Gardner.

11: Bo Wallace:

  1. Charlie Frye (941)
  2. Ryan Tannehill (933)
  3. Austin Davis (911)
  4. Tony Pike (905)
  5. Andy Dalton (904)

Yeah, I got nothing.

12: Sean Mannion:

  1. Chris Redman (873)
  2. Dan Orlovsky (864)
  3. Nick Foles (862)
  4. Matt Barkley (844)
  5. Ryan Lindley (843)

Ryan Lindley is a really bad quarterback. Ergo, I'm much lower on Sean Mannion than I used to be. This is a list of pretty bad quarterbacks.

13: Taylor Heinicke:

  1. Colt McCoy (915)
  2. Chase Daniel (900)
  3. Derek Carr (890)
  4. Bruce Gradkowski (884)
  5. Drew Brees (874)

This is another one of those lists that's very interesting thanks to its combination of perceived value and potential ceiling. We're still just looking at guys at this point who are hoping to be backups, but maybe Heinicke is that 6th round pick with the potential to be a starter down the line?

14: Blake Sims:

  1. Connor Shaw (932)
  2. Troy Smith (910)
  3. Johnny Manziel (894)
  4. Tyrod Taylor (874)
  5. Aaron Rodgers (868)

I am a huge Troy Smith fan. That is all.

15: Shane Carden:

  1. Derek Carr (959)
  2. Geno Smith (933)
  3. Austin Davis (920)
  4. Matt Barkley (919)
  5. Chase Daniel (917)

Shane Carden, everybody. He's here all night.

16: Jerry Lovelocke:

  1. Kevin O'Connell (909)
  2. Shaun Hill (897)
  3. Ryan Tannehill (891)
  4. Tom Brandstater (885)
  5. Tony Pike (880)

Lovelocke is another one of those guys who is rising right now, but unlike Grayson or Petty, the comps don't support it. He's really low on my mathematically flawed list. But I do look forward to him starting off and on for 49ers in about 10 years because Kaepernick's successor is struggling to develop quickly enough for everyone's tastes (because Shaun Hill! -- I love math).

17: Bryan Bennett:

  1. Tee Martin (896)
  2. Bradley Van Pelt (893)
  3. Aaron Brooks (890)
  4. Tyrod Taylor (888)
  5. Marques Tuiasosopo (885)

What is this, a collection of San Diego Padres relief pitchers? (That one's a special inside one for all of you -- you know who you are -- who also read this and got this far). Holy moly, Tee Martin. That's not a person.

18: Brandon Bridge:

  1. Sage Rosenfels (912)
  2. Randy Fasani (876)
  3. Kevin O'Connell (872)
  4. Drew Henson (854)
  5. Shaun Hill (854)

Maybe this guy is more of a... bridge to nowhere? YEEEEEEEEEAAAAAAAAAAAAAAAAAAAAH.

No, that CSI joke is not old and tired. That CSI joke is fresh and hilarious.

19: Nick Marshall:

  1. Matt Mauck (897)
  2. Connor Shaw (887)
  3. Tyrod Taylor (878)
  4. Stephen McGee (869)
  5. Troy Smith (869)

Sorry, Nick Marshall fans. He's the last guy on the list. He's probably really bad. But, hey... He made the list! And Troy Smith! Yay, Troy Smith! Amazing. So in a way, if you think about it, he's tied for the best comp of them all.

Okay, that's my ENTIRE body of work on this project. It was kinda a lot of work. Is there anybody I missed here you want me to run a comp on? Where do these go amazingly right? Unbelievably wrong? Remember, this is meant to encourage discussion and hopefully reveal interesting player parallels to scouting types, not to suggest ACTUAL talent. It was fun, though, and I want to know what you think.

This is a FanPost and does not necessarily reflect the views of Niners Nation's writers or editors. It does reflect the views of this particular fan though, which is as important as the views of Niners Nation's writers or editors.