Jump to content

Welcome Bryce Petty!!


JetBlue

Recommended Posts

There's a difference between "assuming he's different" (which would be asinine) and "being willing to risk a 4th round pick on the chance that he's different" (which is a completely different calculus).

There is no calculus to make. The study is based on pro results. Petty hasn't played a down yet. What is so difficult to understand about this?

Link to comment
Share on other sites

  • Replies 295
  • Created
  • Last Reply

There is no calculus to make. The study is based on pro results. Petty hasn't played a down yet. What is so difficult to understand about this?

What Petty actually does in the NFL is irrelevant.

The calculus is actually pretty simple:

Risk of total failure x ((upside x likelihood of achieving upside) + (likely floor x likelihood of achieving floor)) = expected value of the pick.

Then compare that expected value against the expected value of pick 103 generally, determine your risk tolerance, and make a decision.

You are essentially arguing that the risk of total failure is both defined entirely by the general failure rate of qbs selected in the fourth round and beyond, and so large as to make any other factors irrelevant.

Regardless of what Petty becomes in the NFL, I disagree with both assertions.

First, it makes no more sense to judge fourth round picks by the failure rate in rounds 4-7 than it would to judge third round picks by the failure rate in rounds 3-7. Judged against the fourth round history alone, a qb picked in the 4th round seems to have a 10% chance of panning out (3/31, counting all the "2013 rookies" as busts). That's not great, but it's nowhere near as catastrophic as adding the 1/29 in the fifth round, 3/39 in the sixth, and 3/47 in the seventh makes it appear.

Second, the numbers are being viewed too generally.  What are the commonalities of the successful "4th round QBs," if any?  How about in later rounds? These numbers cry out for more analysis, not a generalized "therefore, never draft a QB round four or down". While it may be true that the total risk of failure for the cohort of QBs drafted round four or later is 134/146 (or thereabouts), it's both facile and untrue to claim that, therefore, the risk of failure for each individual QB taken in that area is 134/146.  By simple application of the law of large numbers, the risk of failure must be higher for some of the QBs in that cohort, and lower for others.  The trick, as always, is identifying which is which.

 

Third, even if the risk of total failure really is 134/146 - that is, there's only about a 7% chance that Petty pans out - and let's say that compares to a 50% chance another player drafted at pick 103 would succeed (and that's very generous, given the past 15 years of picks at 103; the only good players selected there were Davonta Freeman, Frank Alexander, Sam Acho, and Brad Smith) - the question becomes whether a 7% chance of getting Garrard, Orton, or Aaron Brooks is better than a 50% chance of getting Freeman, Alexander, Acho, or Smith.  Or, given your preferences, a 50% chance of a marginal upgrade at OG.  And given this team, and it's context, I don't think it's anywhere near a clear answer.  Good QB play makes this team a legitimate Super Bowl contender - and without good QB play, better OG play doesn't make much difference in outcomes. 

Link to comment
Share on other sites

What Petty actually does in the NFL is irrelevant.

The calculus is actually pretty simple:

Risk of total failure x ((upside x likelihood of achieving upside) + (likely floor x likelihood of achieving floor)) = expected value of the pick.

Then compare that expected value against the expected value of pick 103 generally, determine your risk tolerance, and make a decision.

 

That isn't how regression works.

 

You are essentially arguing that the risk of total failure is both defined entirely by the general failure rate of qbs selected in the fourth round and beyond, and so large as to make any other factors irrelevant.

 

I never said that at all.

 

First, it males no more sense to judge fourth round picks by the failure rate in rounds 4-7 than it would to judge third round picks by the failure rate in rounds 3-7.

 

This isn't a case of whimsical cherry picking. It's a 20 year sample. There is enough data to categorize.

 

Judged against the fourth round history alone, a qb picked in the 4th round seems to have a 10% chance of panning out (3/31, counting all the "2013 rookies" as busts). That's not great, but it's nowhere near as catastrophic as adding the 1/29 in the fifth round, 3/39 in the sixth, and 3/47 in the seventh makes it appear.

 

Standard deviation. It's a thing.

Link to comment
Share on other sites

 

That isn't how regression works.

 

But individual decisions aren't regressions.  They are probabilistic analyses impacted by risk tolerance and perceived upside as much as by the likelihood of any particular outcome occurring. 

 

This is an excellent book on the subject of risk management, btw

 

51Pi4u0LQtL.jpg

 

 

This isn't a case of whimsical cherry picking. It's a 20 year sample. There is enough data to categorize.

 

 

Sure, but your cutoffs are arbitrary.  Why is "fourth through 7th rounds" a thing but "third through 7th" isn't?  The numbers on the 3rd round are 4/24, which means the success rate on QBs drafted in the 3rd round and below over the past 20 years is only 16/170 - or slightly less than the 1/10 hit rate of QBs drafted in round 4 alone over the past 20 years.  So shouldn't your argument extend to round 3 as well? 

 

Why not evaluate rounds 3 & 4 together?  Then the success rate is 7/55, or about 1/8.  That's not bad at all.

 

Why not group rounds 2, 3, and 4?  Success rate becomes 12/77, or 1/6.5.  Getting better.

 

Bottom line - grouping round 4 with rounds 5-7 makes as much or as little sense as grouping it with rounds 1-3.  So while your data isn't cherry picked, your inclusion and exclusion criteria are seemingly arbitrary.

 

 

Standard deviation. It's a thing.

 

No kidding?  Yes, SD is a thing.  So is risk-reward, risk tolerance, and relevant difference (which can shift what the SD is).

Link to comment
Share on other sites

Sure, but your cutoffs are arbitrary. Why is "fourth through 7th rounds" a thing but "third through 7th" isn't? The numbers on the 3rd round are 4/24, which means the success rate on QBs drafted in the 3rd round and below over the past 20 years is only 16/170 - or slightly less than the 1/10 hit rate of QBs drafted in round 4 alone over the past 20 years. So shouldn't your argument extend to round 3 as well?

There are degrees of selection, sure, but the dropoff begins after 3 and takes a nosedive after 4. I suppose you could group them however you like for argument's sake, I started where the fall begins.
Link to comment
Share on other sites

There are degrees of selection, sure, but the dropoff begins after 3 and takes a nosedive after 4. I suppose you could group them however you like for argument's sake, I started where the fall begins.

I just want you to know that you and Doggin94it are beginning to scare me!!!  Ok... Just had to get that out! :sign0182:

Link to comment
Share on other sites

That would be a sensible argument if someone was drafting Petty based on his college numbers. But that's not what's happening - he's being drafted for his tools and traits. A qb like that is much more scout-analysis dependent than number-analysis dependent. QBASE, DVOA & DYAR are important tools for understanding statistics, but they aren't the only elements that need consideration, and the pendulum can swing too far in the direction of statistical analysis as easily as it can swing too far in the direction of scouting

That's not really true. If a model misses on one guy, this is considered a legitimate reason to throw the analytics right out the window. Nobody advocates stopping sending scouts on the road altogether because one eval gets blown. The risk of stats being given too much weight is never as great as not enough because they're held to a much higher standard.

Link to comment
Share on other sites

That's not really true. If a model misses on one guy, this is considered a legitimate reason to throw the analytics right out the window. Nobody advocates stopping sending scouts on the road altogether because one eval gets blown. The risk of stats being given too much weight is never as great as not enough because they're held to a much higher standard.

If only Mark Twain had SPSS, we wouldn't have to deal with this insolence.

Link to comment
Share on other sites

That's not really true. If a model misses on one guy, this is considered a legitimate reason to throw the analytics right out the window. Nobody advocates stopping sending scouts on the road altogether because one eval gets blown. The risk of stats being given too much weight is never as great as not enough because they're held to a much higher standard.

 

That's silly.  At most, a single miss might be reason to tweak the analytics, not to toss them; more likely, it's an outlier predicted by the law of large numbers.  The concept that a single miss requires tossing the analytics is actually an example of the pendulum swinging too far; it's only when someone (wrongly) expects there to be a foolproof formula for predicting the future that a single miss could be seen as proof that the "formula" is wrong.  It's the inverse of saying "the formula says not X, therefore not X is guaranteed"

Link to comment
Share on other sites

That's silly. At most, a single miss might be reason to tweak the analytics, not to toss them; more likely, it's an outlier predicted by the law of large numbers. The concept that a single miss requires tossing the analytics is actually an example of the pendulum swinging too far; it's only when someone (wrongly) expects there to be a foolproof formula for predicting the future that a single miss could be seen as proof that the "formula" is wrong. It's the inverse of saying "the formula says not X, therefore not X is guaranteed"

Obviously you never heard of Damon Huard.

Link to comment
Share on other sites

All the analytics, statistics etc aside this is how I view it.

 

This year was a lousy QB year.

 

The Jets presently have a VERY suspect young starter, an older Vet who is totally short term and a 3rd stringer that the org has shown zero desire to make into anything more.

 

Thus you take your shot at a QB, can't take the top two they are simply too expensive to go after and are perhaps not even real franchise guys,

 

So, we need a QB.  You take the best of the rest in relative terms, a guy that throws like a Qb has size like a QB, supposidley works like a QB but is not a great overall prospect.

 

Taking this guy in the 4th round means nothing at all about any QB we may take or trade for in the future.  If there is any position on a team that warrants taking players on the off if not ow chance they might amount to anything it is QB.  Teams have made it to and won superbowls with average QBs.

 

The Jets needed another QB badly, they did not over pay or over draft one.  They made a smart move in drafting this guy.  And I'd say this if he is out of the league in two years.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...