Confusion over Belichick's non-trade with SF on Pick # 33 in draft

[Shelterdog]

It all depends on your objectives. If you just want to slide back because you don't see a premium player then you'll accept the best offer you can find and be done with it.

If you see a guy who clearly deserves selection, then you'll only move if someone really offers enough juce. This example was pretty clear with #33 this past year when BB had some reasonable offers but just went with his man when he couldn't get a 2012 2nd.

As you say, every situation is different.

And the flip side--he did some terrible deals according to the chart in 2007 (trading a 2007 third for a 2008 raiders third and a 2007 7th, and trading the the 28th pick for SF's 2007 4th and the 2008 1th).

The first deal eventually turned into Mario Henderson for Oscar Lua and the Charger's 2009 second (they bundled that with a fourth and a sixth to get Ron Brace.) At best a draw.

The second turned into Joe Staley for Mayo, Crable and Randy Moss.

 

[patchick]

I suggest that you look at it both for your own edification, and because I strongly suspect that Bill Belichick realizes that the chart is useless and does not use it. He undoubtedly uses the fact that other teams use it in his calculations.

Obviously, GMs aren't likely to sit around with the exact chart we're looking at and base their decisions on dumb addition. But they surely do maintain their own valuation systems (presumably adjusted year to year to reflect the draft class, changing draft rules, etc). And leaguewide, the "standard chart" turns out to be a decent approximation of those values. We saw that in this very discussion, since SF paid close to "chart" value for #36.

In other words, it makes sense for us as fans refer to the chart as a reference point -- even though we know it's not a window into the depths of BB's mind.

 

[MaineMan]

It all depends on your objectives. If you just want to slide back because you don't see a premium player then you'll accept the best offer you can find and be done with it.

If you see a guy who clearly deserves selection, then you'll only move if someone really offers enough juce. This example was pretty clear with #33 this past year when BB had some reasonable offers but just went with his man when he couldn't get a 2012 2nd.

As you say, every situation is different.

No question - especially the way BB does things. OTOH, I was actually shocked when I personally ran the numbers to see how many of the trades since 1995 actually came within a couple % of SVC. And, considering the number of trades that involved multiple picks for multiple picks, it was pretty obvious that many teams actually worked the math pretty hard to arrive at as close to an even exchange, SVC-wise, as possible.

 

[Off The Grid]

No question - especially the way BB does things. OTOH, I was actually shocked when I personally ran the numbers to see how many of the trades since 1995 actually came within a couple % of SVC. And, considering the number of trades that involved multiple picks for multiple picks, it was pretty obvious that many teams actually worked the math pretty hard to arrive at as close to an even exchange, SVC-wise, as possible.

Yup.

Obviously, The Chart is only a guideline.

But, just as obviously, it IS a guideline.

Those who dismiss it, altogether, ought themselves be dismissed, altogether.

 

[Fencer]

I bet the biggest thing that differs from team to team is how they assess the time value of picks.

I further bet they decide that every year anew, based on how they feel about the two drafts in question.

 

[plk]

Let me try it this way. Scorecasting is quite an interesting book. You should be able to get it out of your library although you may have to wait a while. Obviously, you do not want to swallow any book whole.

The chapter on the creation of the draft value chart is especially interesting for a football fan. The chart was put together, roughly, based on trades that clubs actually made. It did not attempt to assess the actual value of a particular pick in relation to other picks. It simply documented what clubs did.

In so far as clubs continue to use the chart, their trades will confirm the chart's accuracy. If the clubs use the chart, then the fans will also find the chart useful since it reflects the clubs thinking.

None of this says anything about the actual accuracy of the chart. There seems to be increasing understanding that it is grossly inaccurate for the higher picks. Maybe the chart reflects reality for the lower rounds, maybe it doesn't.

I strongly suspect that Bill Belichick has a highly sophisticated understanding of the ways in which the chart reflects reality and the ways in which it does not. And how this will vary from draft to draft. I strongly suspect that not all GM's have that sophisticated an understanding, and therein lies an opportunity.

 

[patchick]

I strongly suspect that Bill Belichick has a highly sophisticated understanding of the ways in which the chart reflects reality and the ways in which it does not. And how this will vary from draft to draft. I strongly suspect that not all GM's have that sophisticated an understanding, and therein lies an opportunity.

There are certainly opportunities to be had when you value picks differently from other teams. But I think you're mistaken in suggesting that Belichick is an outlier in intra-draft valuations, throwing the "standard values" out the window. In fact, if you look back over his purely intra-draft moves (like the trades up for Brace & Gronkowski), you'll find a lot of pretty standard value exchanges. The real place BB deviates from the pack is in the valuation of present vs. future picks.

 

[plk]

But I think you're mistaken in suggesting that Belichick is an outlier in intra-draft valuations, throwing the "standard values" out the window.

Understanding to what degree the chart values reflect reality and to what degree they do not is not the same thing as throwing the standard values out the window.

In fact, if you look back over his purely intra-draft moves (like the trades up for Brace & Gronkowski), you'll find a lot of pretty standard value exchanges.

I'm not sure how much information you can gather from that since it is obviously easier to make a trade if the other side believes that the trade is fair.

Draft picks obviously decline in value as the draft progresses. I suspect that almost any reasonable way of starting with a high value and progressing to a lower value for later picks will be reasonably in accord with what clubs do. It would probably be hard to construct a draft value chart beyond the first two rounds that is grossly inaccurate in reflecting what clubs do.

The point is that from the way that the draft value chart was constructed, there is no reason to believe that there is anything especially accurate about it. Unless, that is, you believe that the people who made the trades on which the chart is based had a unerring intuitive sense of the value of draft picks. Maybe that is the case. Maybe not.

 

[patchick]

OK, now I'm really confused. Look at these two statements:

The chart was put together, roughly, based on trades that clubs actually made. It did not attempt to assess the actual value of a particular pick in relation to other picks. It simply documented what clubs did.

The point is that from the way that the draft value chart was constructed, there is no reason to believe that there is anything especially accurate about it. Unless, that is, you believe that the people who made the trades on which the chart is based had a unerring intuitive sense of the value of draft picks.

So basing the valuation on the objective data of what teams actually pay for picks gives you "no reason to believe that there is anything especially accurate about it"? That's how you place a value on EVERYTHING. How much is a particular collectible sports card or Barbie Doll worth? Simple: however much it typically sells for. You use past prices to set current prices.

In so far as clubs continue to use the chart, their trades will confirm the chart's accuracy. If the clubs use the chart, then the fans will also find the chart useful since it reflects the clubs thinking.

OK, so it was based on the reality of what picks are actually traded for, it reflects the reality of what picks are actually traded for, and it shapes the reality of what picks are actually traded for. And yet there's "no reason to believe that there is anything especially accurate about it"? I seriously don't get it.

 

[plk]

Dear patchick,

Since I have already written this, let me post it, and then I will try to reply to your most recent post. This post may actually help to explain what you're asking.

For purposes of comparison with the Official Draft Value Chart, here is a chart that lists the Average Approximate Value, whatever that is, for different draft positions. (The link was originally posted in a thread here.)

2011 NFL Draft Preview - Pro-Football-Reference.com

It is not explicitly comparing the value of draft picks for the purpose of a trade. It is trying to give an idea of the average sort of player that you may get with a particular pick.

Still, it is based on data (I cannot comment on the quality of the data) on the actual performance of players selected with a specific pick. The Official Draft Value Chart is not. I find it interesting to compare them.

 

[plk]

So basing the valuation on the objective data of what teams actually pay for picks gives you "no reason to believe that there is anything especially accurate about it"? That's how you place a value on EVERYTHING. How much is a particular collectible sports card or Barbie Doll worth? Simple: however much it typically sells for. You use past prices to set current prices.

A Barbie Doll has no inherent value. It is completely arbitrary.

A draft pick does have an inherent value particularly in relation to other picks. Looking at the draft value chart, I see that one #17 pick is worth about 10 picks centered around #101. It is reasonable to ask if this is a correct valuation. Maybe a #17 pick is really only worth five #101 picks, maybe it is worth 20 #101 picks. I have no idea, but it is a question that historical data might answer. I just chose the #17 pick because we had one in the past draft. Choose any pick, divide its value by 10, and see how many picks it is equal to. The point is not that the chart is necessarily wrong. It is that we do not know because the chart is not based on the actual performance of players drafted at those spots.

In the link that I supplied above, one #17 pick produces a single player with Average Approximate Value of 34. 10 #101 picks produces 10 players with an AAV of 15. What are the chances of hitting a player, or more than one, with an AAV greater than 34 with one of those 10 picks. I don't know, but it is a question that a statistician should be able to answer.

OK, so it was based on the reality of what picks are actually traded for, it reflects the reality of what picks are actually traded for, and it shapes the reality of what picks are actually traded for. And yet there's "no reason to believe that there is anything especially accurate about it"? I seriously don't get it.

If I have not explained what I am saying in the above and in my previous post, tell me and I will try again.

 

[Off The Grid]

OK, now I'm really confused. Look at these two statements:





So basing the valuation on the objective data of what teams actually pay for picks gives you "no reason to believe that there is anything especially accurate about it"? That's how you place a value on EVERYTHING. How much is a particular collectible sports card or Barbie Doll worth? Simple: however much it typically sells for. You use past prices to set current prices.



OK, so it was based on the reality of what picks are actually traded for, it reflects the reality of what picks are actually traded for, and it shapes the reality of what picks are actually traded for. And yet there's "no reason to believe that there is anything especially accurate about it"? I seriously don't get it.

If I go on trial, and you're the prosecuting attorney, I am @#$%ED!!!

 

[unoriginal]

In the link that I supplied above, one #17 pick produces a single player with Average Approximate Value of 34. 10 #101 picks produces 10 players with an AAV of 15. What are the chances of hitting a player, or more than one, with an AAV greater than 34 with one of those 10 picks. I don't know, but it is a question that a statistician should be able to answer.

Well, let's take a stab at it. The chances of hitting on a player of AAV > 34 at draft slot #17 are 21/41 = 51%. The chances of getting a player of AAV > 34 at draft slot #101 are 5/41 = 12%.

To calculate chance of getting a player with AAV > 34 with ten picks at #101:

100 - (% chance of failure ^ number of chances)
(36/41)^10 = 27%
100 - 27 = 73%

Thus you have a 73% chance of ten picks at #101 giving you a player whose AAV > 34, while you only have a 51% chance with one pick of #17.

Conceptual Proof of the math:
You flip a coin four times. What are the odds that at least once you flip heads?

100 - (% chance of tails ^ number of flips)
(1/2) ^ 4 = 6%
100 - 6 = 94%

Thus a 94% chance of flipping heads at least once in four tries. This is the opposite of the chance that you flip tails four out of four times, which is 1/16.

The wrong way to do it is to simply add up all the positive odds. For example, saying ten picks with independent odds of 5/41 means total odds of success of 50/41 is obviously incorrect.

 

[plk]

OK, so it was based on the reality of what picks are actually traded for, it reflects the reality of what picks are actually traded for, and it shapes the reality of what picks are actually traded for. And yet there's "no reason to believe that there is anything especially accurate about it"? I seriously don't get it.

What we may have here is a fundamental philosophical difference about prices.

You seem to believe that prices are inherently "fair." That something is "worth" what someone is willing to pay for it, or has paid for it in the past. (A friend of mine was fond of saying, "You get what you pay for.")

This idea is completely foreign to me. What something costs is what it costs, but I believe that some things are overpriced, some things are fairly priced, and some things are bargains (underpriced.) There is no necessary link between the "price" of an item and its "worth." (I used to reply to my friend, "A fool and his money are soon parted.")

 

[unoriginal]

Fun with numbers, according to the trade value chart:

One #17 is worth approximately 950 happy time fun points.
Two #38's are worth 940 points.
Three #58's are worth 960.
Four #71's are worth 940.
Five #80's are worth 950.
Six #86's are worth 960.
Seven #91's are worth 952.
Eight #95's are worth 960.
Nine #99's are worth 936.
Ten #101's are worth 960.

Odds of getting a player of AAW of 34 and above with those picks:

1x #17: 51%
2x #38: 50%
3x #58: 57%
4x #71: 63%
5x #80: 49%
6x #86: 73%
7x #91: 67%
8x #95: 65%
9x #99: 76%
10x #101: 73%

Just for kicks:
50x #181: 99.96%

 

[MaineMan]

Let me try it this way. Scorecasting is quite an interesting book. You should be able to get it out of your library although you may have to wait a while. Obviously, you do not want to swallow any book whole.

The chapter on the creation of the draft value chart is especially interesting for a football fan. The chart was put together, roughly, based on trades that clubs actually made. It did not attempt to assess the actual value of a particular pick in relation to other picks. It simply documented what clubs did.

In so far as clubs continue to use the chart, their trades will confirm the chart's accuracy. If the clubs use the chart, then the fans will also find the chart useful since it reflects the clubs thinking.

None of this says anything about the actual accuracy of the chart. There seems to be increasing understanding that it is grossly inaccurate for the higher picks. Maybe the chart reflects reality for the lower rounds, maybe it doesn't.

I strongly suspect that Bill Belichick has a highly sophisticated understanding of the ways in which the chart reflects reality and the ways in which it does not. And how this will vary from draft to draft. I strongly suspect that not all GM's have that sophisticated an understanding, and therein lies an opportunity.

If by "higher picks" you mean #1-#5, yes, those are dominated by large variations from SVC. However, after that and until about pick #25, variations aren't nearly so large and those that are beyond -7% to +7% comprise significantly less than a majority of the trades. After another "bump" in the late-1st/early-2nd, things settle way down to the point that trades varying from SVC by more than about 5% (either way) are exceptional by the later half of round two. If you consider that "later rounds".

This is not a guess, this is what actually happened from 1995 to 2010.

 

[MaineMan]

Dear patchick,

Since I have already written this, let me post it, and then I will try to reply to your most recent post. This post may actually help to explain what you're asking.

For purposes of comparison with the Official Draft Value Chart, here is a chart that lists the Average Approximate Value, whatever that is, for different draft positions. (The link was originally posted in a thread here.)

2011 NFL Draft Preview - Pro-Football-Reference.com

It is not explicitly comparing the value of draft picks for the purpose of a trade. It is trying to give an idea of the average sort of player that you may get with a particular pick.

Still, it is based on data (I cannot comment on the quality of the data) on the actual performance of players selected with a specific pick. The Official Draft Value Chart is not. I find it interesting to compare them.

Now THERE is a horse of a completely different color!

First of all, there's nothing "OFFICIAL" about the Standard Value Chart. There are no "rules" enforcing it on anyone, it's merely a standardized reference GUIDELINE.

Second of all, it has nothing to do whatsoever with how individual teams in their unique circumstance in each draft interacting with the positional strength of each unique draft class may (or may not) value "the average sort of player tat they may get with a particular pick. It's merely a generalized comparative evaluation of OPPORTUNITIES to take players.

You're making it out to be much, much more than it really is and then criticizing it for not living up to that in practice. This is essentially a strawman argument.

 

[plk]

First of all, there's nothing "OFFICIAL" about the Standard Value Chart. There are no "rules" enforcing it on anyone, its merely a standardized reference GUIDELINE.

Calling it "official" was a joke reflecting the way many people treat it.

You're making it out to be much, much more than it really is and then criticizing it for not living up to that in practice. This is essentially a strawman argument.

Umm, I thought I was pointing out that a lot of people take the value chart to be more than it probably really is.

What I'm really doing is encouraging you to read a chapter in a book because you might learn something if you did.

 

[plk]

Thus you have a 73% chance of ten picks at #101 giving you a player whose AAV > 34, while you only have a 51% chance with one pick of #17.

After I figured out where you are getting your raw data, I repeated the calculation. I used draft picks 99-103 for a larger sample size. I used the years 2008 to 1970 since no one drafted in 2009 or 2010 has yet achieved a CareerAV >34. They simply have not been around long enough.

I get an 80% chance that 10 draft picks around #101 will produce at least one player with a CareerAV > 34 which is the AverageAV for draft position #17.

Amusing, isn't it. Thank you for your work. I would not have found the data by myself.

 

[patchick]

What we may have here is a fundamental philosophical difference about prices.

You seem to believe that prices are inherently "fair." That something is "worth" what someone is willing to pay for it, or has paid for it in the past. (A friend of mine was fond of saying, "You get what you pay for.")

This idea is completely foreign to me. What something costs is what it costs, but I believe that some things are overpriced, some things are fairly priced, and some things are bargains (underpriced.) There is no necessary link between the "price" of an item and its "worth." (I used to reply to my friend, "A fool and his money are soon parted.")

OK, I think I finally see what's going on! I was talking about draft picks in themselves as objects of value in a marketplace, and you were talking about draft picks as representing the ultimate benefit of players contributing to a team.

So when you said "there's no reason to believe the trade value chart is 'accurate,'" you meant in predicting the long-term outcome in terms of the players' contributions. And when I said there is reason to believe it "accurate" I meant in describing their value in the draft-day marketplace. So we could both be right, happy day!