Incog's Trade Value Chart

[incognito_man]

2 minutes ago, squire12 said:

Pre rookie wage scale seems like it would inflate the early 1st round picks and potentially dampen the later picks as there would be less cap space overall to distribute to other players.  

would the rising cap space over time have an impact on things?  More money in later years might give a boost in the recent years as players come off the rookie contract

Numbers are adjusted relative to the cap, so that won't have an impact. 

It does appear that the rookie wage scale somewhat flattens the distribution of career earnings, although if you remove the two outliers in both directions (2008, 2009 low) and (2015, 2016 high) the remainder are all clustered together. 

I'm not convinced there is a causal relationship there yet 

[squire12]

6 minutes ago, incognito_man said:

Numbers are adjusted relative to the cap, so that won't have an impact. 

It does appear that the rookie wage scale somewhat flattens the distribution of career earnings, although if you remove the two outliers in both directions (2008, 2009 low) and (2015, 2016 high) the remainder are all clustered together. 

I'm not convinced there is a causal relationship there yet 

it might be interesting to remove the rookie contract part of this since the rookie wage scale pre-sets the salary (career earnings for that portion of a players career) based on draft slot and not the value of the player's ability/value of the pick 

[incognito_man]

6 minutes ago, squire12 said:

it might be interesting to remove the rookie contract part of this since the rookie wage scale pre-sets the salary (career earnings for that portion of a players career) based on draft slot and not the value of the player's ability/value of the pick 

I don't think it's worth the effort. Subsequent career earnings will dominate and rookie wage scale is still a reflection of perceived future value (just like 2nd and 3rd contracts).

[skibrett15]

15 hours ago, incognito_man said:

(1) Normalize the top "x" number of career earners for each draft class. I wanted to see each draft class individually compared to itself. I assigned a value of 1000 to the top earner. The second highest earner would then get a value proportional to their career earnings compared to the top guy. If he earned 75% as much, he would get 750. I repeated this process for the data I had (roughly 60-80 guys per class). 

So this assumes that pick 2 always takes the 2nd best earner, and so on? As if teams had perfect information about how the careers would play out?

Why did you use this methodology rather than looking at the career earnings of the actual draft slot 1-96?

[incognito_man]

3 minutes ago, skibrett15 said:

So this assumes that pick 2 always takes the 2nd best earner, and so on? As if teams had perfect information about how the careers would play out?

Why did you use this methodology rather than looking at the career earnings of the actual draft slot 1-96?

Goal was to map the value distribution of a given draft class (and then compare across draft classes). It's clear the league naturally forms a logarithmic valuation of players across the top 60-80ish just about every cohort. 

So, it makes the most sense to me to map the resultant valuation (expressed by how the owners spend their real dollars) to the process of talent acquisition as a baseline.

The baseline assumes an efficient market where teams will draft the highest value players first and the draft class talent will be distributed similarly to previous classes.

The standard deviations then handle various talent distributions unique to a given class (using historical actual draft classes).

[skibrett15]

15 minutes ago, incognito_man said:

Goal was to map the value distribution of a given draft class (and then compare across draft classes). It's clear the league naturally forms a logarithmic valuation of players across the top 60-80ish just about every cohort.

https://overthecap.com/draft-trade-value-chart

I don't disagree that earnings is a fair way to measure the value of a pick after all is said and done.

You're measuring the value of a player in a draft class regardless of draft slot by looking at who made the most money, then 2nd most, etc.  Then you order these in descending order, and re-assign them to the draft slot based on that descending order. 

There is variation between classes - some 2nd earners earn almost as much as the top whereas some 2nd earners earn just around half the top earner. The red number is the lowest or the minimum (or the average of the classes lowest?), the green number is the highest or the maximum (or again, the average?) based on percentage of the top career earner in that respective class.

I do believe you are accurately capturing the logarithmic nature of each draft class's respective career earnings.

I'm less convinced that has anything to do with the draft process, let alone the value of the resultant pick slot.  See the link above for a slot by slot evaluation. 

[beekay414]

OTC's TVCs just isn't based on reality. Much better off using the Rich Hill chart when discussing pick swaps. 

https://www.drafttek.com/NFL-Trade-Value-Chart-Rich-Hill.asp

As was stated yesterday, feel like OTC's TVC is just a collection of what those picks have been worth (after the fact) and not necessarily what they actually are worth during the draft process. 

EDIT - But I have no desire to rehash that again lol

Edited March 23, 2023 by beekay414

[HokieHigh]

would be cool to check this against some trades that actually happened to see how it performs (calculate estimated from actual value difference), compared to some other charts

[HokieHigh]

also the logarithmic fit makes a lot of sense. money gains at the top are fairly incremental year to year, but the value in the ~20th to 50th percentiles way exceed the bottom of the barrel

[incognito_man]

1 hour ago, skibrett15 said:

I'm less convinced that has anything to do with the draft process, let alone the value of the resultant pick slot

I do like the idea behind their chart as well but don't have the book to know what tweaks they put on it. Comparing them, they place an even higher value on top picks than the method I chose.

The ideology to me is simple: it's mapping the value of the picks to the expected value distribution of the class.  I added the +/- STDEV columns to capture the varying distributions of past classes. Each represents one standard deviation (calculated using the listed data sets) up or down from the median for each value slot.

[incognito_man]

14 minutes ago, HokieHigh said:

would be cool to check this against some trades that actually happened to see how it performs (calculate estimated from actual value difference), compared to some other charts

Yeah I'm curious to look at actual trades as well and grade them against this chart.

[wgbeethree]

Looking forward to seeing Justis post this on Twitter. 

[ThatJerkDave]

3 hours ago, wgbeethree said:

Looking forward to seeing Justis post this on Twitter. 

[skibrett15]

17 hours ago, incognito_man said:

The ideology to me is simple: it's mapping the value of the picks to the expected value distribution of the class.  I added the +/- STDEV columns to capture the varying distributions of past classes. Each represents one standard deviation (calculated using the listed data sets) up or down from the median for each value slot.

ok that makes sense.  that's what the table said it was, I just didn't get the same interpretation from reading through your examples.

 

For me, one of the biggest gaps is that teams do not realize how terrible they are at picking the correct player.  Usually they do catch on quick once the player is on the team though.  So one of the biggest flaws in the draft process is that teams do not account for the high level of uncertainty in their pick - they assume they are getting a good player with every selection they make. And why shouldn't they? I mean they are the ones whose job it is to scout these players.  Maybe they do realize the process is very imprecise and with a lot of error, but they certainly don't seem to behave like it.

 

So I like the charts which emphasize past production of the draft slot more than a method which assumes they will get the next best player with every subsequent selection.  I think it does a good job of emphasizing the inherent variance of the outcomes and would emphasize the errors teams are making by trading up and wasting picks.

 

To @beekay414 's point, there is also an established standard "market rate" for picks.  It's important to know that market rate for the picks in question.  Whare are the picks worth to you, vs what are they worth in the market?  Is there a delta you can find for a gain?

Edited March 24, 2023 by skibrett15

[incognito_man]

13 minutes ago, skibrett15 said:

ok that makes sense.  that's what the table said it was, I just didn't get the same interpretation from reading through your examples.

 

For me, one of the biggest gaps is that teams do not realize how terrible they are at picking the correct player.  Usually they do catch on quick once the player is on the team though.  So one of the biggest flaws in the draft process is that teams do not account for the high level of uncertainty in their pick - they assume they are getting a good player with every selection they make. And why shouldn't they? I mean they are the ones whose job it is to scout these players.  Maybe they do realize the process is very imprecise and with a lot of error, but they certainly don't seem to behave like it.

 

So I like the charts which emphasize past production of the draft slot more than a method which assumes they will get the next best player with every subsequent selection.  I think it does a good job of emphasizing the inherent variance of the outcomes and would emphasize the errors teams are making by trading up and wasting picks.

 

To @beekay414 's point, there is also an established standard "market rate" for picks.  It's important to know that market rate for the picks in question.  Whare are the picks worth to you, vs what are they worth in the market?  Is there a delta you can find for a gain?

I think charts that include actual picks selected then inherently include "bad" data. There is value to that data, absolutely, but I think it becomes less useful when it's buried within the other data. 

I wanted to separate that out and come up with a pure value baseline that mimics the actual value distribution. I felt that distribution would be relatively static (turns out it's not), but the trend (logarithmic decay) is consistent.

I'm not sure I agree that including the human behavior (draft misses) within a TVC is helpful. Consideration of the human element should occur AFTER consulting the true value chart (i.e. Bad Team offers a trade package that yields surplus value on our teams' adjusted chart - pull trigger on their offer).