Zeroing in on the Jets' draft tendencies

Prior to the draft I wrote the following article, looking into trends and tendencies from the Jets' draft classes since Mike Maccagnan has been the general manager.

Analysis: The Weighting Game: A Filtered Jets Board

Now that the draft is done with and we have a three-class sample to review, we can re-visit some of the assumptions made in that article and try to refine the model.

If you re-visit the first article, you'll note that the Jets' eventual first pick - Jamal Adams - exhibited all three of the traits we identified as possibly being things the Jets were looking for. He was a younger prospect (21), with good length and no off-field concerns.

How did the rest of the draft pan out though? Does it confirm any of our main theories?

Let's first reiterate what those theories were:

Length is something they seem to be targeting;
They tend to avoid players with off-field red flags;
They target younger prospects; and
These rules become less important as the draft goes on.

Re-read the previous article for discussion about why they might consider these things as important.

Applying these rules to this year's class

If the above rules were all true, you'd expect to see a lot of green, especially at the top, and not much red, which would mostly be near the bottom.

From this, we see that - other than the first pick - any patterns are not particularly strong.

In the previous article, I treated "off-field issues" as a binary issue with a yes/no answer, while acknowledging that the cut-off point is difficult to define. Coincidentally, three of the first five picks were all suspended for one game each in obviously unrelated situations and for undisclosed reasons. While I wouldn't have considered that enough to generate a red flag, I've marked those as lighter green just to account for it.

Widening the sample

We'll learn more by looking at the larger sample covering each of Maccagnan's first three classes. Adding this year's results in gives us the following outcome:

Already we can see some problems with our initial hypothesis. Prospect age seems to really matter in round one, but not after that. At all.

That makes sense though, because your first round pick will get a fifth year option and a possible franchise tag, so it's advantageous to have a younger prospect who should, as a result, still be in their prime for the duration of their second contract. Non-first rounders only get four-year deals, so the benefit is lessened.

Avoiding spending a draft pick on a player with off-field issues remains a trend, but it's interesting to note that this doesn't extend to undrafted free agents. Jalin Marshall and a few other unrestricted free agent had some red flags last year and there are three of this year's reported class who also appear to have some highlighted concerns.

In terms of length, it looks like there's not much correlation here either, but when you review the numbers (which, by the way, are percentiles based on their position via Mockdraftable) you can see that almost everyone has above average length.

Tweaking the model

Let's see if we can play around with the results to generate a better result. This may seem like confirmation bias, but it's important to remember that we're not necessarily putting forward a theory which can then be proved or disproved. We're aiming to refine a model by which we can get some idea as to what these tendencies might be.

I arbitrarily used parameters of 0-25 = red, 26-50 = pink, 51-65 = white, 66-80 = light green and over 80 = green for length. If we simplify these parameters, it will better represent the relationship. Now I'm simply going to use 0-25 = red, 26-50 = pink, 51-75 = light green, 76-100 = green.

I'm also going to treat age after the first round as not relevant and arm length for players in positions where it arguably doesn't matter (specialists, quarterbacks, running backs) as not applicable.

While this gives a more satisfactory result, another trend immediately becomes apparent. Nearly all of the outliers are from the 2015 class.

So what happens if we eliminate the 2015 class from our sample?

Suddenly we seem to have a pretty accurate representation of the needs Maccagnan and his team seem to covet.

But we can't just remove the first year from the results because it produces a better result, can we?

Actually, we can justify this. Let's not forget that the scouting department was completely overhauled after the 2015 draft. And it's this current scouting department whose tendencies we are investigating.

Leonard Williams was really the only pick from 2015 which met the criteria and that would have been one on which Maccagnan's influence was greatest. It stands to reason that the scouting department would have been relied upon a lot more in the later picks.

I also thought about introducing some more variables, but there weren't any obvious ones that produced a pattern. Whether or not a player had an official visit with the Jets was one possibility but accurate data was difficult to find.

It looks like each of the six players selected in the first two rounds, other than Devin Smith, had an official visit, but the majority of the picks after the second round did not have an official visit reported.

An Imperfect Model

Ultimately, this is an imperfect model for several reasons, of which the fact Maccagnan could still have sought the players with these traits even when relying on data from his old scouting department is just one.

Something else that needs to be questioned is whether the team are looking specifically for length or are they just looking for good players with length happening to be one trait that many good players typically share?

Assuming length is important, Dylan Donahue stands out as pretty much the only outlier in this category. And if recent suggestions that they drafted him to be a special teams monster rather than a position player, then perhaps we should treat him as a specialist. If he's just playing special teams, then he's going to be largely operating in space where his lack of length might be a non-factor.

The final flaw in this analysis is that while Leonard Williams and Jamal Adams fit the model perfectly, nobody really expected them to fall to the Jets. Would they have been compelled to trade up for either? Were there similar players available if Adams or Williams didn't fall and would they have targeted those players? We don't know.

Re-writing the rules

Now that we have this information, let's recap what seems to be important:

They target players with good length and avoid players with below average length unless they play a position where length is unimportant;
They tend to avoid players with off-field red flags with their draft picks, unless the issues are relatively insignificant; and
They target younger prospects in the first round but age is a non-factor after that.

Although we acknowledge the flaws within this model, it's still going to be interesting to apply this data to prospective targets next year and to further add to the sample after the 2018 draft.

That, of course, assumes that Maccagnan will still have the job...