What Makes a Reach?

[equivocation]

If this graph is correct from PFF, it’s looks pretty damn close even if the NFL draft does generally outperform the big board, especially after pick 80.

https://forums.footballsfuture.com/t...omment=4587181

I posted a link to that graph earlier. A couple regions where the NFL draft outperforms:

First 10-15 picks. Likely explained by teams "reaching" for QBs. Like Mahomes was a "reach".

4th to end of the graph. This indicates that the consensus big board is just missing on some players. This makes sense as NFL teams scout much more thouroughly.

A period from mid 3 to to 4. This indicates the consensus board is misclassifying some 3rd round picks as 4th round picks. Some boards are more thorough than others, so I speculate this is driven by players with a high spread in picks across boards (some boards give plauers a 2nd/3rd grade, others a 4th-6th). In these cases the higher grade is likely correct, but it weighs down their consensus ranking.

The big board does a very good job through mid 3. Based on that, "reaches" should be considered in the range 15-75. So a 5th round grade in round 2, for example. Boards aren't missing on top 60 players.

[equivocation]

2 more notes:

The NFL graph is generally well behaved with few discontinuities. End of 1st, end of 2nd, end of 4th. For the most part each marginal pick is providing less value, which is what you expect. Consensus board is bumpier.

because they're even right around 75, you can cut off the bottom of the graphs. Each marginal pick after 75 provides less value, so the % delta in value per pick is larger than it looks when you look at the whole graph. Almsot do7ble at the end of round 3 and about 33% higher by the end of the draft.

As a whole the NFL outperforms the pundits. This doesn't mean any individual team outperforms the pundits, ir in any given year. LAC is well in the top half of WAR draft from 2017-2021.

[equivocation]

2 more more notes.

1. Scheme fit can impact WAR so teams drafting for fit may also show improved results over big board.

2. 32 NFL teams work independently so the draft is a "true" consensus result. Draft pundits do not work independently.

[dmac_bolt]

Poster called Palmer a reach...among other things. Said nothing possitive about him. We'll see how that plays out in his second season as it is clear he has reached and maybe surpassed Guyton in his first season.I wont use the word fact, but i think Palmer is our #3 from TC day 1.

I’m big on Palmer for 2022 too. He’s just got PRO all about him, idunno. Coaches like guys that play like pros.

[dmac_bolt]

Nobody has ever suggested that big boards were objective, so why are you raising this point as if someone had?

I stated that the procedure that I used to determine significant reaches was objective and not subjective. Under the procedure that I have outlined for assessing a significant reach, which is determining whether or not the consensus big board ranking number of a player minus the draft slot number used to select the player exceeds 32, I am not free to alter the analysis or result based upon how I feel about a particular player. The test is objective even though the consensus big board is not.

You can certainly argue that the consensus big board ranking list is not useful, but you cannot properly argue that the application of the significant reach test outlined above is not an objective test.

There are many fatal flaws in your procedure. it’s quite true you establish an objective, meaning numerically measurable, range to apply to a subjective board. One flaw is that you assume your objectively defined range is an appropriate value for your algorithm. You have no data to demonstrate 32 is the correct value, you just made it up. Is that 3 standard deviations? Is it the 5% probability the null hypothesis should be accepted, ergo it should be rejected? Is your consensus distribution even normally distributed? Do you even look at the consensus distribution? No - its nothing - its just you making up a sound-good number. You make up a lot of things, alot of us notice it frequently. but this isn’t the most significant flaw in your procedure.

You believe that overlaying an objective range value atop a subjective consensus ranking makes the overall process objective. it doesn’t. By your own theory of consensus, the TPB has conducted a consensus assessment in this very thread and the consensus conclusion is that you are wrong.

others already posted but i will repeat - placing an error tolerance around a subjective opinion still leaves you with a subjective opinion.

[Xenos]

2 more more notes.

1. Scheme fit can impact WAR so teams drafting for fit may also show improved results over big board.

2. 32 NFL teams work independently so the draft is a "true" consensus result. Draft pundits do not work independently.

A Ravens fan brought this up:

https://forums.footballsfuture.com/topic/43631-should-gms-just-draft-base-on-the-general-consensus-board/?do=findComment&comment=4587969

I’m sure someone has had to have mentioned this by now but the consensus big board is LARGELY made up from the inputs they get from team sources.

So this is somewhat of a chicken or the egg situation. The consensus big board if left up to the media personalities alone would look AWFULLY different if teams took a 9-5 approach and then didn’t feed them a stream of sourced info to build a consensus board around.

[equivocation]

A Ravens fan brought this up:

https://forums.footballsfuture.com/topic/43631-should-gms-just-draft-base-on-the-general-consensus-board/?do=findComment&comment=4587969

Yeah, that's true as well. And it is subject to manipulation because of that. The sources have to stay within bounds or lose credibility but there is definitely some lying going into liars month.

Someone tried to push Raimann up the boards. Didn't work. Same with Malik Willis. But if teams find a late round gem they're not making that public, hence the clear separation in WAR by the end.

The NFL teams are the ones doing the actual work, everything we see that is public facing is just us guessing at shadows of that work.

No one here can do real film work on 500+ players. I can do maybe 10, plus the guys we actually draft.

[equivocation]

Of the big boards I think the following three do their own work:

PFF. I have lots of issues with their methodology and conclusions but I do believe they generate their own rankings based on their own system. I don't believe their film study is deep enough or consistent enough.

The Draft Network. They list the games studied, the evaluator, the grades of the 3 or 4 people giving independent grades, and their conclusion. I generally find them thorough and accurate.

Bleacher report. Like TDN but much more crowd sourced which results in some inconsistencies depending on the quality of the evaluator.

[Guest]

There are many fatal flaws in your procedure. it’s quite true you establish an objective, meaning numerically measurable, range to apply to a subjective board. One flaw is that you assume your objectively defined range is an appropriate value for your algorithm. You have no data to demonstrate 32 is the correct value, you just made it up. Is that 3 standard deviations? Is it the 5% probability the null hypothesis should be accepted, ergo it should be rejected? Is your consensus distribution even normally distributed? Do you even look at the consensus distribution? No - its nothing - its just you making up a sound-good number. You make up a lot of things, alot of us notice it frequently. but this isn’t the most significant flaw in your procedure.

You believe that overlaying an objective range value atop a subjective consensus ranking makes the overall process objective. it doesn’t. By your own theory of consensus, the TPB has conducted a consensus assessment in this very thread and the consensus conclusion is that you are wrong.

others already posted but i will repeat - placing an error tolerance around a subjective opinion still leaves you with a subjective opinion.

You are challenging a position that I am not asserting.

I have stated that the procedure I am using is objective. I have never suggested that consensus big boards are objective. In terms of player analyses, however, I think a consensus is more likely to be accurate than an early outlier.

And by consensus, I do not mean that everyone agrees on the exact value of the player. But when the consensus value of a player is #137, that means lots and lots of pundits view the player as worse than #137 overall. And lots of pundits having that view is enough to raise a red flag about the player and the team's early outlier analysis of the player.

People have posted about how the consensus big boards compare to NFL GMs and the overall result is close with NFL GMs possibly having a small advantage. But that is not what we are comparing in the cases of significant reaches versus consensus big board rankings. We are comparing early outlier opinions versus consensus big board rankings. In those cases, while I have seen no data, my intuition tells me that the consensus is right far more often than NFL GMs are. Telesco's track record is a walking monument to that proposition.

I first stated one round as the defining point for a significant reach, but modified that to the more objective 32 draft slots. Maybe it needs to be one round, though. The reasoning behind it was not the rejection of the null hypothesis, but rather that teams were taking players earlier than they likely needed to take them, wasting draft pick value. In other words, if a team has another, later draft pick that the consensus big board suggests could be used to select the player, I see that as problematic and a red flag for potentially not getting good draft value.

An issue that has not been discussed is what happens when the team lacks a pick in the later round or, through multiple trades, has a pick more than 32 slots later. That team really does not have a chance to draft the player within 32 slots and may not have a chance to draft the player at all before the consensus big board ranking point. The theory needs to be further developed to account for these situations as well.

But basically, a team that drafts players significantly earlier than it needs to loses draft pick value. No player is so important that the best draft strategies should be deviated from in order to select a particular violative player.

To me, the concept is generally sound, but there will be plenty of specific exceptions. Honestly, that the Ravens likely drafted better than we did this year when they repeatedly got plus value versus the consensus big board rankings with every pick but one versus our repeated reaches versus the consensus big boards is flagrantly obvious to me.

I keep thinking that we are just a few players away. I have to believe this same thought occurs to others on this forum. And our drafts are why that is the case year after year as we have left an incredible amount of meat on the table in our Telesco era drafts and we miss all kinds of opportunities seemingly every year.

[dmac_bolt]

You are challenging a position that I am not asserting.

I have stated that the procedure I am using is objective. I have never suggested that consensus big boards are objective. In terms of player analyses, however, I think a consensus is more likely to be accurate than an early outlier.

And by consensus, I do not mean that everyone agrees on the exact value of the player. But when the consensus value of a player is #137, that means lots and lots of pundits view the player as worse than #137 overall. And lots of pundits having that view is enough to raise a red flag about the player and the team's early outlier analysis of the player.

People have posted about how the consensus big boards compare to NFL GMs and the overall result is close with NFL GMs possibly having a small advantage. But that is not what we are comparing in the cases of significant reaches versus consensus big board rankings. We are comparing early outlier opinions versus consensus big board rankings. In those cases, while I have seen no data, my intuition tells me that the consensus is right far more often than NFL GMs are. Telesco's track record is a walking monument to that proposition.

I first stated one round as the defining point for a significant reach, but modified that to the more objective 32 draft slots. Maybe it needs to be one round, though. The reasoning behind it was not the rejection of the null hypothesis, but rather that teams were taking players earlier than they likely needed to take them, wasting draft pick value. In other words, if a team has another, later draft pick that the consensus big board suggests could be used to select the player, I see that as problematic and a red flag for potentially not getting good draft value.

An issue that has not been discussed is what happens when the team lacks a pick in the later round or, through multiple trades, has a pick more than 32 slots later. That team really does not have a chance to draft the player within 32 slots and may not have a chance to draft the player at all before the consensus big board ranking point. The theory needs to be further developed to account for these situations as well.

But basically, a team that drafts players significantly earlier than it needs to loses draft pick value. No player is so important that the best draft strategies should be deviated from in order to select a particular violative player.

To me, the concept is generally sound, but there will be plenty of specific exceptions. Honestly, that the Ravens likely drafted better than we did this year when they repeatedly got plus value versus the consensus big board rankings with every pick but one versus our repeated reaches versus the consensus big boards is flagrantly obvious to me.

I keep thinking that we are just a few players away. I have to believe this same thought occurs to others on this forum. And our drafts are why that is the case year after year as we have left an incredible amount of meat on the table in our Telesco era drafts and we miss all kinds of opportunities seemingly every year.

You are just mistaken. You have an opinion, that is all.

[21&500]

I think Chainy's point is modest and valid, though not necessarily true. There is nothing earth-shattering in his actual claim that justifies this amount of controversy.

Imo it's just triggering to many here because it's likely to show to be true and right now we want to hold on to hope that Woods was more overlooked, rather than a reach.
It would be a very promising sign for the Staley/Telesco duo.

I get it, it's a downer perspective but if he said the same about a raider draft pick, I doubt anyone cares.

[equivocation]

I think Chainy's point is modest and valid, though not necessarily true. There is nothing earth-shattering in his actual claim that justifies this amount of controversy.

Imo it's just triggering to many here because it's likely to show to be true and right now we want to hold on to hope that Woods was more overlooked, rather than a reach.
It would be a very promising sign for the Staley/Telesco duo.

I get it, it's a downer perspective but if he said the same about a raider draft pick, I doubt anyone cares.

He has provided no evidence to support his "intuition". None. Zilch. Zero. Are we to take his word?

Anyway, there's even kinda a rule about this.

https://theramswire.usatoday.com/2020/05/07/nfl-rams-seahawks-bobby-wagner-les-snead-draft/

Bobby Wagner was a 3rd-4th round consensus pick. The Rams traded down assuming he'd still be there. Someone else took him at 47 (35 before big board, meets arbitrary definition of a reach) If at least two teams are reaching for a player, is he still a reach?

https://walterfootball.com/draft2012consensusOLB.php

We only know this story because it's Bobby Wagner and Snead talked about it. How often does this happen? We don't know. It's non-falsifiable except in cases where team management talks.

Anyway, defining a reach as "32 picks above consensus" is also dumb. If anything make the spread on the draft trade value chart or something. An early reach is worse than a late reach. At some point all the players left have a similar grade because NFL draft values have a long tail so after a certain point reaching really, really doesn't matter. What is that point? I have it mid 3rd. Based on the little data we have.



Example: player 8 on board has a 8.6 grade. Player 18 on board has a 7.8 grade. 10 spot reach has a 0.8 point spread.

Player 80 on board has a 6.5 grade. Player 130 on board has a 6.3 grade. 50 spot reach has 0.2 point spread. You're telling me the later is worse tham the former? BS.