Monday, December 9, 2013

Sometimes I'm Wrong...

Speculating about the future success of college football players is a questionable endeavor.  Having your speculation based almost entirely off of the players' combine data and college stats is even more questionable.  Sometimes I'll be right, and sometimes I'll be wrong.

Being wrong doesn't necessarily bother me too much, so long as it doesn't happen too frequently.  My main goal is just to explore some ideas on how to optimize a team's chances of success in the NFL Draft, based on more objective data.  I can't pretend to have watched the majority of the players when they were in college, and honestly doubt that doing so really is a very reliable method for identifying talent.  For the most part, I'm just curious as to whether draft selections chosen by a computer can approach or exceed the results made by NFL GMs.  If, indeed they are experts in this field, I should fail horribly.  If the GMs are blindly guessing, while posturing as highly paid experts, maybe the results will be surprising.  Personally, I suspect that the average fan with a draft magazine and a dart board, could probably do just as well as the best experts, so I aim to encourage a bit of doubt when it comes to blindly trusting the people in charge.

While I have plenty of stupid ideas, the real core of my idiocy probably gets condensed down to my annual Ozzie Newsome Challenge.  While teams talk about taking the "best player available", I think it can generally be agreed that all such claims are blatantly bullsh*t.  Frequently this familiar line gets modified to "best player available at a position of need", which is more honest, though probably still a bit misguided.  It's kind of like going to a party, in search of your future wife/sandwich maker, and saying that you are seeking the "best wife available....with exceptionally large breasts".  The added qualifier "with exceptionally large breasts" sort of reveals what your real motives are.  For instance, do you really think that Matt Elam and Arthur Brown (the Ravens' first 2 selections in the 2013 Draft), were the best players available?  Or, were the Ravens blatantly trying to fill holes left by the departure of Ed Reed and Ray Lewis?  In the Ozzie Newsome Challenge, we admittedly don't exactly go for best player available either.  Instead, we try to select the player who we think is the "best player available..who we think will be gone in the next 32 picks (or however long it is until our next pick)".  Occasionally, our earliest picks aren't the ones we are most interested in acquiring, but we figure we can wait to take particular players who we anticipate will be overlooked by NFL teams.

It will be quite a while before we can judge the results from the 2013 Draft, and there are one or two selections that the computer made that I wish I could change.  Still, amongst the players that the computer selected, and who have been given  an opportunity to prove themselves, I am generally quite pleased with the results so far, and wanted to take a brief look at two of the more interesting results that came from late round selections.

First of all, we have Chris Jones, defensive tackle for the New England Patriots.  His initial selection at the end of the 6th round (198th pick overall), already shows the degree to which teams weren't too intrigued by him.  Then, the team that drafted him (the Texans) chose to cut Chris Jones before the regular season began, another vote of no confidence.  He was promptly picked up by the Buccaneers, who also quickly dropped him before he ever played a single game.  Once again, things were looking rather bad.  Then, he gets signed by the Patriots, shockingly is allowed to actually play, and accumulates 45 tackles and 5 sacks in his first 8 games started. 

I don't want to engage in a subjective analysis of how well he has played, since I don't think that will get us anywhere.  Loosely throwing around opinions is pretty useless.  However you look at it though, I think it is probably fair to say that Jones' performance has likely exceeded his draft position, and thus the expectations of NFL GMs (experts!).  Instead, let's just ask the question "How unlikely was it that Chris Jones would start at least 8 games as a rookie?", a mark he just reached yesterday.   If we look back at data from past drafts (1992-2011) we can calculate how unlikely it is that such an occurrence is just a fluke (using historical data from Tony Villiotti at To even out any irregularities at a particular draft slot, I looked at all the players taken during this 20 year span, that were selected between the 188th pick and the 208th pick (ten spots prior to, and subsequent to where Chris Jones was chosen).  In total, there have been 420 players selected in that range in the past 20 years, and only 20 of them managed to start at least 8 games in their rookie year.  So, the odds were 1/21 (a 4.76% chance) of such an outcome happening just by blind luck.

Next, we have the case of Paul Worrilow, linebacker for the Atlanta Falcons.  Of course, Worrilow wasn't drafted at all, and the Falcons were supposedly the only team to even offer him a chance.  Despite this, he has now accumulated 101 tackles and 1.5 sacks, through his first nine games started.  Unfortunately, the lack of data related to undrafted players complicates things a bit.. Without this data, I decided to examine the issue as if Worrilow had been selected at the end of the seventh round, figuring that if anything the odds for an undrafted player would probably be even lower.  Using the same data from, I examined the situation as if Worrilow had been selected between the 235th pick and the end of the seventh round to get a similar sample size to the one we had with Chris Jones.  In this case, we had 421 players during this 20 year time frame that had been selected in the range, with only 7 managing to start at least 8 games in their rookie season.  That works out to a 1/60.14 probability (a 1.66% chance), of such an event occurring due to dumb luck.

Where this becomes really interesting is when we ask the question "What were the odds of a team selecting 2 such players in the same draft?".  The odds of both events occurring would be 1/1262.94 (a 0.079% chance).  To be fair though, Team Kangaroo had 5 selections from the end of the 6th round to the end of the 7th round, so our actual odds were closer to 1/252.5 (or 0.395% a chance).  Either way, those odds do make you wonder about the likelihood of whether the computer was just lucky, or if taking a more objective view of a player's data might be a reasonable thing to do.  It also makes me have a greater appreciation for late round draft picks.

Will either of these player end up as Hall of Famers?  Probably not, but that would be a rather unreasonable expectation anyway.  Yes, judging a player based on his number of games started as a rookie is obviously questionable.  Still, observing the successes of late round and undrafted players, is much more interesting to me than their counterparts who were selected in the first few rounds.  The hype factor is less of an issue with the late round players, so getting on the field probably does suggest that a player is doing something right, rather than inheriting a starting role because a team has invested a lot in them.  All I'm concerned with is whether I can get anything at all out of a draft pick.  I really don't think that selecting "stars" is nearly as important as avoiding "bums", who can tie up a roster spot for years simply because they were highly drafted.  Why some players succeed, when nobody was really betting on them doing much, probably reveals more about what really matters.

As for the computer's other picks in 2013, well, some will turn out well and some won't.  The biggest question for me is whether a player is given an opportunity to show what he can do, over which I have no control.  As it stands, I think the odds are in our favor that a few more of the computer's picks should turn out quite well...if given a chance.

No comments:

Post a Comment