I am returning to the draft discussion today, because I believe there is still useful information to be gleaned from the event.  Today, we’re throwing positional impact away and focusing purely on the order that players were drafted and how they compared to others within their position group.

Before we get started, I want to refer us back to last year so you can get a sense of why this type of analysis is useful.  It’s too early to judge last year’s draft class, but we definitely have a sense of each player and what the draft order would be if it was re-done with current information.  Here, for example, are the top DTs from last year:

The “reach” measure is the same thing we looked at last week; a negative number there means the player was chosen EARLIER than “expected”.  However, I want you to look at the “Pos % Score” column.  Within the TPR model, each position has a different “maximum score” depending on the impact values I derive from the salary cap data.  This

“Pos % Score” column tells us what percent of that “max score” each prospect attained.  Basically, it’s an easy way to back out the positional impact adjustments and focus purely on the question “how good is this prospect?”.

There are a couple quick caveats.  I don’t have the NFP ratings from last year, so those don’t figure into the scores.  The TPR model does not account for “fit” or “role”, so a NT and a 4-3 DT will be compared against each other.  However, since all we are trying to do is identify the “best potential player(s)”, regardless of role, I’m not too worried about either of those.  In the future, I will try to increase the positional resolution of the model (I did so this year by splitting OLBs and MLBs) to better account for the “role/fit” issues.

Looking at the chart above, we can see that Fletcher Cox graded out as the best DT prospect, yet he was chosen AFTER Dontari Poe.  At this moment, it looks like that was a big mistake by the Chiefs.  Also note the very low rating for Derek Wolfe, who had a very weak rookie year.

Now here is a chart of the OLBs:

Notice Lavonte David rated significantly higher than the 3 LBs chosen before him, and his rookie season bears that rating out.

Here are a couple more positions from last year, then I’ll move to this year. Notice both the order and absolute difference between players’ Pos % scores:

Above, we see the Brandon Weeden received an almost identical score as Brock Osweiler, but was taken 35 picks earlier.  Also, we can clearly see the “tier” separation between prospects.

In this WR chart, compare Rueben Randle’s rating with that of either A.J. Jenkins or Brian Quick.  While 49ers fans were probably surprised Jenkins couldn’t get on the field, this chart suggests he was drafted ahead of better prospects.

Ok, you get it.  What about this year?

Lets start with the CBs.  Please note that the only players I included are those that made the original TPR top 137:

As you can see, there is some serious deviation from the pre-draft rankings.  Darius Slay, in particular, looks like he was chosen too high, since both Banks and Taylor carry SIGNIFICANTLY higher grades.  Obviously, these are not full-proof, but as we saw above they suggest the Lions (Slay) might have made a costly mistake.

Here we have the DEs, and the variation is not nearly as severe as in the CB group.  The only one who really jumps out is Margus Hunt, drafted above Damontre Moore.  Hunt is a very high-risk/high-reward player, but note that the TPR rankings suggest this difference was due to Moore falling rather than Hunt being “reached” for.  In fact, Hunt was drafted exactly where the TPR system rated him.  Moore, however, fell 40 spots.  William Gholston, near the bottom, looks to be a bit of a steal, especially compared to Alex Okafor, but since his grade is just 71.7%, he doesn’t project to be an impact player anyway.

The DTs are notable for their surprising LACK of deviation.  Jesse Williams “fell” a lot, but as I said last week, that’s due to injury concerns and likely reflects a medical risk that can’t be quantified.  Bennie Logan, unfortunately, rates as the biggest “reach” of the group, but note that his positional score is fairly close to the DTs drafted after him, meaning within the position group, he wasn’t a terrible pick.  This would seem to suggest that the Eagles, despite their claim to draft pure “value”, likely made this pick based on “need”.

Looking at the Guard position, we see evidence of the Bears’ perplexing draft strategy.  While Kyle Long “projects” as a potential OT in the future, this suggests that regardless of position, Larry Warford is much more likely to be a good player.  The fact that he was also available more than 40 picks later is more damning evidence against the Bears’ perceived “value” in this draft.  Again, Kyle Long might become a great player, all I’m saying is that the odds of that happening are less than the odds of Warford becoming a big contributor.

At the OT position, we can see why the Fisher/Joeckel/Johnson triumvirate was so sought after.  Those three players represent the clear “top tier” at the OT position this year.  Nothing to note after that, as the OTs were drafted in almost the exact order they “should” have been.  However, we do see a big value difference if we look at DJ Fluker and Menelik Watson.  While both players graded out similarly, Watson was taken 29 picks later, meaning he was a significantly better “value”.

Not too much to say here that hasn’t already been said.  Mike Glennon is a very peculiar choice, but EJ Manuel represents the biggest risk.  Either Manuel was rated by several teams to be the only QB worth a 1st round pick, or the Buffalo Bills were bluffed into making a poor decision.  With a GM of questionable judgement and a rookie head coach, it’s likely they just screwed up.

This chart also throws the breaks on the Barkley hype.  Again, the story with Barkley is “great value”, NOT “great QB”.  His score is OK, but nothing special.  FOr reference, it’s roughly equal to Brandon Weeden and Brock Osweiler from last year (though notably it’s much higher than Nick Foles’).

Here are the TEs, and they tell a similar story as the DTs (as far as Eagles fans are concerned).  I have Ertz rated as a slight reach, but within the TE class, he was picked where he should have been (though Escobar is clearly the better value).

Without comment, below are some other positions.  In general, it’s important to remember the larger point here.  When a team “reaches” for someone, they are essentially saying “My evaluation of this kid is better than everyone else’s (or almost everyone)”, usually, they’re wrong.

Today I’ll cover the players that “fell” in the draft and see if we can identify any potential “steals”.  However, I want to start by doing something I should have done yesterday.  I have ESPN ratings and NFL.com ratings for several past years (I do not have NFP ratings or NFL.com ratings for 2011).  Therefore, I can run the TPR system through the 2010 and 2012 drafts and see if it correctly identified potential busts and steals.  I showed this analysis once before, but have since updated the model (and its more interesting now anyway).

As I explained yesterday, the model should be more successful in identifying busts than it is in identifying steals.  If a player falls dramatically, there is usually a reason for it, and one that can’t be quantified and put in our model.   However, players that go well ABOVE their TPR rankings, are usually just indications that teams did in fact “reach” for a player.  For 2010, this is what the biggest “reaches” list looks like:

You’ll notice that, compared to this year, the “reaches” in 2010 were not nearly as significant.  Still, in the above chart, we’ve done a good job of identifying the players who were not “worth” their draft spot.  It’s not perfect, obviously, but the “bad” selections far outnumber the “good ones.

Consequently, I’m comfortable saying that when a team “reaches” for a player, rarely do they “know more” than everyone else.  In fact, the few successful cases may just be the result of luck (if you reach on enough players, a couple of them are going to work out).

Now you can go back and view yesterday’s post with a bit more evidence behind it.

The “Steals”

Unfortunately, as I mentioned above, the “steals” are not nearly as easy to identify.  For example, here is the same 2010 draft class, with the “value” picks shown:

There are definitely some players in there that qualify as “steals”.  However, they are mostly concentrated in the 8-12 region (that’s draft spots below TPR ranking).  I’m not entirely surprised.  Jimmy Clausen, for example, rates extremely high on the TPR system, due to his impact position and high consensus rating from ESPN and NFL.com.  Clearly, though, teams saw something about him that is not reflected here. They seem to have been justified (though he hasn’t been given a great chance).

So what can we learn?  I’m not sure, but perhaps we can be skeptical of anyone who fell “too far”.  That’s not as scientific as I’d like, but we have to start somewhere.  If a player fell more than a full round (30+ picks), we can assume that almost every team passed on him, meaning there’s likely something about that prospect that our model isn’t picking up.

Here is the chart for this year.  I’ve cut the sample to players drafted in the top two rounds (for reasons I explained yesterday) plus the 10 remaining players with the highest TPR rankings.

Several of these players have “health issues”.  Jesse Williams and Keenan Allen are both reported to have injuries that raised flags with most teams.  That’s obviously not what we’re after here, but if healthy, those two are pretty obvious candidates for “value”.  After reviewing the 2010 data, we should be skeptical of the other players that “fell” more than 30 places (roughly a full round).  As we saw with Jimmy Clausen, it’s likely teams saw something in Barkley, Nassib, and Wilson, that can’t be represented quantitatively.   I obviously hope they’re wrong about Barkley, but that’s probably not the case.

So where does that leave us?

I’d suggest that we focus on Arthur Brown and the players below him on the chart.

– Arthur Brown is, today, widely regarded as having been a great “value” pick.  I have no idea why he fell so far, but there have been no reports of injuries or off-field issues that I have seen.  However, pointing to him is cheating a bit, since every “draftnik” is already calling him a “steal”.

– Similar story with Sharrif Floyd.  While the TPR system did not rate him as highly as some “experts”, he was still graded a top 10 pick, and he fell 13 spots to #23.  Obvious candidate, and you didn’t need to come here to know that.

– The rest, however, I’ll take credit for, particularly those at the bottom.  Notice that Star Lotulelei, John Cyprien, Bjeorn Werner, and Cordarrelle Patterson all carry relatively high “positional scores”.  That means, ignoring of positional impact, they are good prospects.  Star, though he only “fell” 8 spots, jumps out due to his 92%+ score, but the others are clear “value” picks as well.

I was hoping Cyprien would be the Eagles #35 selection, but I was obviously not the only one who liked him (he was the first pick on day 2).

– The other players I’d point to are Jamar Taylor and Menelik Watson.  Each of them fell a significant amount (more than 20 picks), so it’s possible there are some behind-the-scenes issues with both players.  However, given the significant rankings deviation we saw throughout the draft, it’s also possible these guys were just overlooked.

Neither projects to be an “impact” player, but both have solid scores.  Additionally, both players play positions with relatively high historical “hit” rates, meaning their positional risk is less than most other players.

– Larry Warford I’m not too interested in, though Guards have a very high “hit” rate.  His positional score is relatively low, and it’s near impossible for a G to become an “impact” player anyway.  He might have a good career, but it’s going to be difficult for him to become good enough to count as a big “steal” in the draft.

– Tank Carradine is another player with injury issues.  It’s interesting, though, that he did not fall nearly as far as Williams or Allen.  That tells me he received a much better report from team doctors than the other players did.  If Tank is healthy, he can absolutely be an “impact” player.  He’s not a great fit for the Eagles, so I don’t mind passing on him, but he can also be looked at as an “impact” player that was taken in the 2nd round.  However, since the medical risk is real, even if he pans out it will be unfair to say he was a “value” pick.  Just because a player avoids the risk associated with him doesn’t mean the risk wasn’t real.

I’ll be closely tracking the progress of these players throughout next season.  While one season won’t be enough to judge each pick, we’ll likely be able to knock a few off the list one way or another.  Hopefully, after watching this class play out, we can adjust the model or create a new measure that will help us identify “steals” as easily as we’ve identified “reaches”.

Now it’s time to go back to our TPR rankings and see which players represent the biggest potential “reaches” and “values”.  Today I’ll do the “reaches”.  A couple of notes before I start:

– The TPR system is only designed to analyze the first 2 rounds of the draft.  Therefore, today I will only be looking at players drafted in those two rounds.

– In general, I think we’ll be much more successful in identifying “busts” than we will be in identifying “values”.  If a player falls dramatically, my first assumption is that the league knows something that we and the media analysts do not.  It could be due to an injury risk or personality defect; neither of which we can measure.  The “busts” however, represent the reverse, where a single team picks someone well above their perceived value.  In this case, it’s likely that the team is being overconfident in its own assessment.

So who were the biggest round 1 and 2 reaches?

Below is a table illustrating the players that were most heavily “overdrafted”.  The column with red text represents the prospect draft pick minus their TPR ranking.  So if prospect X has a score here of -30, it means he was drafted 30 spots earlier than the TPR ratings suggested he should be.

Due to team differences in scheme and the relatively close ratings of a lot of prospects, we aren’t really concerned with small differences.  Large ones, however, should be very informative.

Overall, there was significantly MORE deviation than what I expected to see.  Before I break down the chart above, let me advance a theory about this draft:

It might be a lot worse than we think.

In the first round, 4 of the biggest “reaches” were for interior offensive linemen.  What might that tell us?  Well if the ENTIRE draft is weaker than we suspect (not just weak at the very top), then several teams might decide to make their picks purely based on risk.  If the risk/reward tradeoff for every player is skewed towards the risk side, then it would make perfect sense to “reach” for a relatively low-risk player like an OG or C.  By doing this, you will appear to have passed on the opportunity to select an “impact” player.  However, if there aren’t any (possible, though unlikely), coming out of the draft with a decent starter at a low impact position isn’t a bad outcome.  I hope this is not the case (and I don’t think it is), but it would explain a lot of the perplexing decisions made on day 1 and 2 of the draft.

Now back to the chart.  Here are my takeaways:

– In my opinion, the Bears had the worst draft (see the next two bullets).

– The biggest “reach” of the entire draft, based on TPR, was Jon Bostic, an ILB from Florida.  The Bears drafted him with the #50 pick, and I did not even have him in my top 137 players.  Since he’s a LB, the Bears will get the benefit of the doubt, but if you’re looking for a potential bust, he’s a very strong candidate (though in the second round it’s not as noticeable or meaningful).  He may turn into a good player, but at the very least, it appears as though the Bears should have waited a round (or two) to take him.

– Kyle Long was one of the aforementioned 1st round interior linemen “reaches”.  He was drafted #20 overall, despite a TPR ranking of just #94.  1st round Guards have a very low miss rate (Danny Watkins was a rare exception), so Long will likely have a productive career.  However, the Bears probably passed on several better prospects at more impactful positions.  This is an under-the-radar reach, since the player will probably contribute, but it represents terrible value nonetheless.

– That’s twice in the first two rounds that the Bears “got their guy” regardless of the actually value of each player.  Either the Bears know something nobody else does (or very few teams do at least), or they just screwed up.  Time will tell, but I know which side I’d bet on.

– Two RBs are near the top of the list, Christine Michael and Le’Veon Bell.  Both were taken in the 2nd round, so the bust potential is somewhat limited.  Running Backs, though, are terrible “value” picks near the top of the draft.  They’ve been proven to be, for the most part, interchangeable.  Neither of these guys (nor Bernard, also on the list), projects to be a LeSean McCoy-type impact RB.  If that’s the case, it would have been better to draft a higher rated prospect and search for a RB later.  In my success-odds table/database, second round RBs became starters just 21% of the time, well below the odds of prospects at several more impactful positions (for example, 50% of 2nd round CBs became starters).

– The Bills might have screwed up big-time with the EJ Manuel pick.  Not only was he a big “reach” by our rankings, he was also not even close to being the top QB on the board. I believe two things happened here:

1) The Bills had Manuel as “their guy”, which as you know, is a dangerous place to start from.

2) The Bills were likely bluffed into taking Manuel much higher than they needed to.  I know there have been rumors that several other teams (including the Eagles) also wanted Manuel, but given the way the rest of the draft went, it’s more likely that the Bills fell for the smokescreens.

The fact that the Bills have a less-then-sterling record when it comes to QBs only increases the inherent risk of this selection.  (Bad GM Theory)

– Everyone (including me) gave the Cowboys shit for drafting Travis Frederick early, but note that he is far from the biggest reach in these rankings.  Still, it looks like he was picked about a round too early.

– Matt Elam, picked by the Ravens, will be an interesting prospect to watch.  Since it was the Ravens and Ozzie Newsome that made the pick, everyone assumes it was a good one.  However, I’ve got it as a big reach.  I should note that Elam’s TPR ranking is lowered significantly by his NFP grade.  If I had to bet, I’d certainly pick Newsome and the Ravens over the NFP.  Regardless, given that they traded up to get him, the Ravens are representing to the world that they are extremely confident in Matt Elam being significantly better than the other safeties on the board (which they could have selected had they not traded up).

– The Eagles do make an appearance on this list, at the very bottom.  Zach Ertz was selected 15 spots higher than his TPR ranking.  As I explained on Monday, I’m not overly concerned by this, since 15 spots in the second round isn’t a MAJOR deviation.  However, it is entirely possible that Chip Kelly’s desire for a TE led the Eagles to make a poor “value” selection.  I’m betting on Chip here, and think Ertz will be a significant contributor.

For what it’s worth, here are the highest ranked guys (in TPR) that were NOT selected in the 1st two rounds:

I’ll cover the “value” picks tomorrow, but it’s safe to say the guys listed here are potential “steals”.