Aside from the draft itself, the NFL Combine is the gem of the offseason. All 32 teams and millions of fans tune into what amounts to a combination interview/workout—or a meat market if you're particularly disinclined to what happens—to see which players will perform the best running in a straight line, and therefore which players might have a better shot in the NFL.
For some of the more obscure positions, how do we know what's important? The offensive line in particular seems to be an enigma.
Obviously, the Combine is about more than the 40-yard dash, although sometimes it feels as if all the attention devoted to this workout is about the footrace, even for the big tanks in the trenches: offensive linemen.
Experienced draftniks will tell you that it isn't the 40-yard time that matters, though. It's the 10-yard split. Some will argue that the 10-yard split is a player's "real football speed" while others will tell you that it's a testament to strength and explosiveness. In fact, outside of workouts specifically for Pro Days and the Combine, strength and conditioning coaches are more often going to train players to run in ten yard bursts.
So if the 10-yard split matters, it might be OK to partially ignore 40-times. But does the 10-yard split actually matter?
What about strength? Arm strength (used in the initial punch in pass pro or as a run blocker) can theoretically be measured by bench reps, and leg strength can be tested by the vertical leap and the broad jump. Both of those seem important to linemen, as much of an emphasis as there is on technique.
There are a few ways to determine which workouts matter, and each have their advantages. For offensive linemen, I used three different types of analysis: first used draftmetric's excellent work on determining effectiveness through tracking the number of starting snaps a player took, the second used 2012 rookie year data culled from Pro Football Focus, and the third used PFF data for players drafted four years prior to their grade (i.e. 2005 Combine data will be matched to 2008 performance).
Draftmetric consistently puts out quality work about the draft—and I encourage a readthrough. In this case, what they have to say about the offensive line and the combine is interesting, but nothing outside of what you would expect.
What was interesting was that there was nearly no difference in 10-yard split times and the number of starts a player accumulated. In all events, there was a very slight level of improvement in the number of starts, but it was a small enough difference that they could safely ignore it.
Most importantly, they found that different things mattered for different positions along the line.
They found that there wasn't much of a relationship in the range of times and how many starts a player has had, although they did find that there were specific cutoffs in times that seemed to be important. For centers, it seemed modestly important, to post a short-shuttle time of 4.43 or better. Those players were 80% more likely start with a low shuttle time. 18% of 1-year starters had a sub-4.43 time, compared to the 10% of centers at the combine who posted that time in general—indicating a somewhat significant relationship.
A short-shuttle time of 4.51 was slightly more important, as 35% of starters recorded a time below 4.51, out of the 20% at the combine who recorded such a time.
For guards, they found the same pattern, although the results were even more modest. What surprised the folks there the most, however, were the results for tackles. While the relationship between shuttle times was stronger there than for guards or centers, the broad jump suddenly became a relevant measure.
11% of all combine participants jumped 110 inches or more, but 16% of all 1-year starters posted such a result. 35% of 1-year starters jumped 107 inches or more, while 21% of combine participants did the same.
Starts is a great way to determine ability, starting with the relatively safe but by no means 100% accurate assumption that teams are good talent evaluators (and cream rises to the top), but there are now better ways to determine exactly which physical feats match with which general talents.
While I don't yet have a method of separating zone-run blocking and man blocking (I start running into sample size problems, already an issue), I can use Pro Football Focus' grading system to match general skills with specific workouts.
Unsurprisingly, run blocking takes a different set of muscle groups (and are therefore expressed in different workouts) than pass blocking. What is more surprising is the set of workouts necessary to exceed in each.
In general, a good run blocking interior lineman will post a quick short-shuttle time and a good vertical leap. What's weird is that the more reps on the bench press a lineman took, the poorer he performed as a run blocker. This might be the result of sacrificing athleticism or skill development for strength—alone strength isn't a bad thing, but the type of players who focus on building up strength will generally sacrifice other important skills.
That's not to say the players who performed the worst in the bench press did well, either. Those with 18 or fewer reps were not long for the league. But with players like Logan Mankins (21) and Chris Myers (24) posting pedestrian numbers in this event, there's good evidence that this could be overrated. In the NFL, technique seems to be much more important than strength—a relic of the fact that everyone in the NFL is already strong.
When removing bench reps from a generic predictive formula for run blocking, the relationship between the formula and actual run blocking efficiency got stronger despite the strong correlation between bench reps and run blocking. This implies that it the "problem" with high bench reps is more likely a result of tradeoff (between skill/finesse and body-building) that occurs and is not a warning flag in and of itself. But it also does imply that bench reps are overrated when evaluating interior offensive linemen.
Taking a reverse approach—looking at current interior linemen and going backwards to find their combine results—led to different, equally surprising conclusions. It doesn't find that there's a problem with "too many bench reps" (a further indication that we shouldn't read too much into simple correlations), but it does find that the 40-yard dash is highly correlative (relatively speaking) with run blocking ability.
That doesn't mean much—it could simply be an indicator of work ethic (it takes a lot to change one's footwork to properly accommodate the 40-yard dash), a result of a habit to drive one's legs, a product of weight, or simply access to a good training facility (itself likely an indicator of high draft stock).
In any case, it isn't useful. Adding it to the larger predictive formula makes it less powerful. Instead, factors like the vertical jump, the short-shuttle, and the 10-yard split are the most useful factors into any formula to guess at how well an interior lineman will perform at run blocking. In fact, it will explain ten percent of performance regardless of draft order, technique or NFL line coach.
These Combine statistics do not work both forward and backwards, however. In the NFL, successful run-blocking linemen generally put up better vertical jump numbers than their less successful counterparts. But posting better numbers in this statistic in the Combine doesn't translate in the reverse.
The reason for this is because there seems to generally be a ceiling in how well the vertical leap determines run blocking ability. Anything better than 29.5 inches is generally the same, be it 30 or 35 inches of elevation. Below that, there is a strong relationship between one's success as a run blocker and how well one can generate lift from a standing start.
The most consistently predictive measure (even if it is weakly predictive) for interior run blockers is the short-shuttle. The top ten percent of guards and centers post times 4.4 and below, but generally look for them to do better than 4.65. In the vertical leap, the best performing guards jump up 31 inches or more, but so long as they get 27 inches off the ground, they should be fine. Only one of the top 15 run-blocking performances between 2008 and 2012 had a guard with a vertical leap below 27 inches (Nate Livings in 2012 did very well as a run blocker and had a vertical leap of 23 inches).
In between both of those in predictive capability is the broad jump. It is useless going backwards—knowing how good a player that plays often will block for the run does not give any indication as to how they will be as broad jumpers. But the best performing broad jumpers often perform worse when blocking for the run. Some of this is a function of weight, some a function of "finesse" vs. "power" guys and some of it is a function of the fact that guards who get moved around easily themselves can move around easily. That does not mean a good broad jump is necessarily a bad sign, but it could be reason to emphasize other statistics.
The formula for "predicting" run blocking performance from the Combine is relatively simple, although again it should be cautioned that these will explain 10% of performance at best, and should generally be used as a tiebreaker or as answers to questions about "competition level". In full, the equation is:
The second part just means the maximum value will be 29.5. The equation above should give a general score from 0 to 10 (it is possible to go higher or lower, but very, very unlikely) that will rank an interior lineman's combine indicators in terms of run blocking. Any score below a 2 and any score above an 8 merits special attention. The best run blocking seasons tend to come from centers and guards who scored above a 7.5, while the worst ones generally come from interior lineman who scored below 3.5.
There are always exceptions at the edges (Nick Mangold scored 6.2 and has been consistently top-tier as a run blocker while Adam Snyder had the same score but has had an abysmal time creating lanes) , but I have yet to see exceptions at the extremes (barring injury).
In pass blocking, the most correlative Combine statistic was again the vertical jump, but the second most correlative statistic was time in the 40-yard dash.
While probably another surprise, it's not a meaningful input into any predictive formula and could be dismissed as noise (or more likely a confounded variable—one already affected by a related skill that has already been factored in). Instead, a number of other factors provide a good preview into how a pass blocker might develop in the middle of the line.
It was difficult to retrodict any data—once players passed a certain physical threshold and established themselves as pass blockers, it seemed as if there wasn't a significant difference between the best and worst pass blockers in Combine data. Once again, the only (baffling) difference makers were 40-yard dash times and vertical leap.
This makes more sense if we continue to see the 40-yard dash as more of an indicator of other desirable (or noncontrolled) traits, like a strong work ethic or access to a good training facility. The vertical leap makes some sense, as it is not just a measure of leg strength but also balance and control.
Like the 40-yard dash, bench press reps were a bit confusing without larger context. The bench press generally correlated negatively with pass blocking capability for guards and centers, and this may once again be due to a tradeoff effect.
Once established as a pass blocker, the three-cone score had some, but not very much, capability predicting the general order of pass blockers in terms of efficiency.
Combine scores seem to determine about 30% of an interior lineman's pass blocking efficiency—a shockingly high amount given the cerebral nature of the passing game, even in the trenches. Likely this comes from the fact that lateral agility is tested at the combine and that the best performers at the combine are often the hardest workers. The predictive equation that is most useful includes an "agility score" that by itself does a lot to predict one's NFL career (as well as any workout can).
The agility score is simply the shuttle time divided by the athlete's weight, which results in a much more useful indicator than either metric taken alone, although adding in the independent shuttle score once again strengthens the relationship between Combine numbers and the ability to block interior rushers.
The formula, using Combine scores, to predict one's ability to protect the passer is as follows:
After the Combine, I'll compile a list of prospects and their respective pass blocking and run-blocking scores. The fact that shuttle times were important in DraftMetrics' studies of the interior line and very important in predicting both pass protection and run blocking speaks to how important some of the relatively poorly covered events are at the Combine.
For tackles, the story doesn't change much, except that all of the relationships between performance and workouts get a little weaker. Knowing a prospective tackle's shuttle score will tell you a little bit about his ability as a run blocker, although your best bet is the broad jump score.
Multiplying the broad jump by the 10-yard sprint time produces a "burst score" that correlates incredibly well to run-blocking prowess if you're only given Combine numbers. Working backwards (and with a much, much larger sample size), however, reveals that the simplest equation for retrodicting (and theoretically predicting) run blocking performance is just dividing the shuttle time by the weight of the player.
The weak conclusion that you can draw from all of this is that the type of players who earn starts in the National Football League typically are players who have skills that parallel well with the broad jump and to some degree initial acceleration off the snap. Once on the field and regularly earning snaps, however, the skills associated with shuttle speed determine one's performance.
In order to check that theory, I instituted a cap on all broad jump performances of starters who played regular snaps. Everyone with a score of 106 inches or more had their score changed to 106 inches even. If the conclusion is correct that run blocking performance is not dependent upon talent related to the broad jump, then the raw broad jump score should produce little to no correlation with run blocking talent and an artificial barrier (like a cap or a floor) would increase the correlation.
Both turned out to be true in this test. There was very, very little correlation with raw broad jump numbers and run blocking, but the relationship became very significant when instituting a cap of 106 inches. The numbers remained meaningless when implementing a floor, which implies that up to a certain point, associated skills for the broad jump are still important.
The equation for getting on the field is simple.
Once on the field, the equation changes to something else.
Like every other equation listed, it should give a rank from 1-10, with 10 being the best indicator for success and the 1 being a red flag. In the sample (players who played tackle for four games in 2008 and/or 2012), no one scored above a seven and received a negative run-blocking grade. Only one player scored a 2 or lower and received a positive grade.
Pass blocking is much, much more difficult to predict for tackles given combine scores. This perhaps rightly implies that playing tackle in the NFL requires pass protection duties beyond those measured in workouts or is much more cerebral than other skills at other offensive line positions.
There wasn't a lot of things coming out of the small sample set of Combine data that looked promising projecting forward, but there was strong data coming from tackles who had already established themselves in the league. The strongest relationship between Combine numbers and pass blocking performance turned out to be dividing a player's shuttle time by their weight.
Overall, a somewhat more complicated formula fits the performance of successful pass blockers better.
All this implies is that a fast shuttle time, a quick 10-yard split and a fast three cone score are all important, but more important than the raw data isn't much without the context of how much weight the tackle is moving around. Being hard to bull rush is helpful, and that's why weight is a good thing to put on, but footwork is critical. More than that, additional reps beyond 28 don't really mean much for pass protectors.
It's not a surprise that the shuttle drill has a strong relationship with pass blocking—there's a lot about it that simulates pass blocking in itself.
I was concerned about overfitting the equation, but randomly splitting the offensive tackles into two separate groups several different ways showed similar correlations, so it's not as big a problem as you might think.
It's even possible to add run blocking and pass-blocking metrics into an overall grade for offensive linemen—one that has a stronger relationship with the overall ability of an offensive tackle than either metric does with pass protection or road grading.
As always, the result of the above equation should produce a result from 0 to 10 (except in extreme circumstances), with 10 being the best. Any grade above a 7.5 or below 2.5 is notable and should shift an evaluator's opinion on a potential prospect.
- UPDATE: Looks like the equation I had put in place here wasn't producing useful 1-10 results, so thanks to Grime and Tony Wiltshire, I took a look again and replaced it with a better (useful) equation.
All of these formulas were run against a generic test against where a player was picked, both in overall draft order (Pick #3, Pick #8, etc) and against other linemen (first center off the board, second off the board, etc.), and each of these formulas outperformed where a player was picked in the draft in terms of overall performance.
That's not quite a fair assessment, as players picked lower in the draft don't start as often and are therefore not included in the dataset, but even when compared to the top five linemen picked in each position, the formulas performed favorably. This implies either the counterintuitive notion that talent evaluators actually undervalue the Combine or, more likely, they value the wrong workouts, at least as far as offensive linemen are concerned.
This isn't to say that on-tape performance isn't important—if anything, this analysis has proven that on-tape performance is significantly more important in evaluation than raw physical ability. It's difficult to find strong statistical relationships anywhere in workouts or college game that predict NFL performance. Tape reigns supreme.
In addition, the drills that players perform are more important than just the results. On the shuttle drill, for example, evaluators will look at how a player's knees, hips and ankles flex. The three-cone drill will test a player's flexibility and dip. The number of falls a player has when engaging in the broad jump will test their balance. All sorts of information doesn't get captured and published, and will be important for scouts and managers.
These results also imply that the type of skills that get a player onto a field (like ones associated with the Broad Jump) don't always translate into the types of skills that make them excellent performers once the bright lights turn on.
If you were interested in the Combine, you should now have some idea of which workouts mean the most. Take particular care to look at the shuttle times, broad jump and three cone drill and pay some attention to the 10-yard splits. Most of all, controlling for a player's weight will tell you more about how they will do than the raw numbers. Later, we'll look at other positions.
Up next is wide receiver, which should be loads of fun.