Let's Talk Value Based Drafting (1 Viewer)

[Jamesinaz]

Reading this thread I feel like I belong at the little kids table.  Lots of math going right over my head.  So I hope my questions aren't too basic.

Having just purchased the APP and run a few mock drafts I have noticed a few things I don't understand.  

1) it does seem to rank QBs higher than most drafts do.  Now I understand the MATH may say Brees should go in the 2nd round but if no one else in my league is taking him or any QB other than Rogers until the 4th round it makes little sense for me to take him when VBD says I should.  I've practiced on ESPN mock drafts and typically QBs don't really start to go until about the 5th round.   And yeah, I get it that the program is following who is drafted (assuming I can keep it up to date) and reevaluating players based on that.  

2)  The upside discussion earlier which I didn't comprehend.  As an example, if I am in the later rounds VBD might suggest Shane Vereen over Kameer Hunt or Darren McFadden.  And in a purely seasonal production basis as the teams stand now Vereen might be more likely to score more points.  But in the later rounds I am usually drafting guys who I don't expect to start unless there is an injury or bye situation.  And I will "churn" on the waiver wire looking for the next Howard Jordan starter.  And both Kareem and Darren would instantly vault into RB1 status if an injury occurs to the guy they back up.  yet, they are further down.  

I guess I see the value of the app in aiding me in selecting my first 10 players.  But the tail end of the bench it seems like I'm better off taking some flyers on talented backups than solid vets with little upside.  Again, maybe it's just over my head.  

 

[abbottjamesr]

Reading this thread I feel like I belong at the little kids table.  Lots of math going right over my head.  So I hope my questions aren't too basic.

Having just purchased the APP and run a few mock drafts I have noticed a few things I don't understand.  

1) it does seem to rank QBs higher than most drafts do.  Now I understand the MATH may say Brees should go in the 2nd round but if no one else in my league is taking him or any QB other than Rogers until the 4th round it makes little sense for me to take him when VBD says I should.  I've practiced on ESPN mock drafts and typically QBs don't really start to go until about the 5th round.   And yeah, I get it that the program is following who is drafted (assuming I can keep it up to date) and reevaluating players based on that.  

2)  The upside discussion earlier which I didn't comprehend.  As an example, if I am in the later rounds VBD might suggest Shane Vereen over Kameer Hunt or Darren McFadden.  And in a purely seasonal production basis as the teams stand now Vereen might be more likely to score more points.  But in the later rounds I am usually drafting guys who I don't expect to start unless there is an injury or bye situation.  And I will "churn" on the waiver wire looking for the next Howard Jordan starter.  And both Kareem and Darren would instantly vault into RB1 status if an injury occurs to the guy they back up.  yet, they are further down.  

I guess I see the value of the app in aiding me in selecting my first 10 players.  But the tail end of the bench it seems like I'm better off taking some flyers on talented backups than solid vets with little upside.  Again, maybe it's just over my head.  

As Joe said earlier, VBD is there to tell you what the value of the player is, not when to draft them.  If you draft every player at value you will have a mediocre draft.  knowing you can get 2nd round value in the 4th is where you gain advantage.

 

[Joe Bryant]

As Joe said earlier, VBD is there to tell you what the value of the player is, not when to draft them.  If you draft every player at value you will have a mediocre draft.  knowing you can get 2nd round value in the 4th is where you gain advantage.

Absolutely Abbotjames. This is a big one I see people ignore. They'll yell, "I would never draft a player there!". I'm not telling you to draft them there either. I'm saying that's what their value is. Its like buying and selling on Ebay. Ebay "sold for" prices are like ADP. If you're going to buy stuff at garage sales and sell it on Ebay later, you'd better be paying well under the Ebay "sold for" price. Same with drafting. 

J

 

[Joe Bryant]

Not sure if this is the right spot for this but perhaps Joe can help out. Years ago when there still were magazines I read an article that expanded on VBD and converting the fantasy points to auction dollars. Would love to see this article again. It had something to do with figuring how many fantasy points were to be scored by your starters and then dividing that by how much money you allotted for your starters. You would then multiply the fantasy points by the dollars per point. Not sure of the details anymore so would appreciate it if someone could refresh my memory or link the article.

Let me see if we can dig up some info on converting over to Auction dollars for you Kingprawn.

J

 

[Biabreakable]

Obviously that article by JJ was written awhile ago. In 2012, the ADP for QBs had 4 in the first round and a 5th in the 2nd round. Here is a VBD table for the 2011 season which is part of the reason many people were suggesting drafting QBs so high. Rodgers was 3rd, Brees was 5th, Cam was 8th, Brady was 9th and  Stafford was 14th. So ADP for 2012 was actually pushing against the VBD results of 2011. Obviously projections change year to year, but if people were looking at 2011 VBD, they would have been more inclined to take the top QBs earlier than the 2012 ADP suggested. 

QB is a spot where VBD totally misses. In most leagues, the supply/demand of QB is so out of whack that quality starters can be had on the WW as the year goes on. Also, you don't just have to draft 1 QB and play him every week. Below average to bad QBs  take advantage of bad defenses on a weekly basis and as JJ has documented, an owner can use a frankenstein's monster of QBs throughout the year to achieve top 8QB numbers without spending almost any draft capital. With QB, looking at end of the year numbers isn't all that important.

As Joe already pointed out those VBD numbers are for 10 team league that only starts 1 QB and makes the conclusion that 16 QB will be rostered and uses that as the baseline instead of the worst starter baseline. This method increases the VBD for the top QB significantly compared to what it would be for QB 10 and I don't think that is a good application of VBD principles, but that is what was done.

The VBD total column represents the number of players above replacement-level value. There is as much art as science to it. The simplest way to find those numbers in your league is to count how many players were drafted at each position. For example, many teams will draft 2 quarterbacks, but often teams with elite quarterbacks like Aaron Rodgers and Drew Brees will just have one. I estimated that 16 QBs would be owned in this type of league.

Then in the 2nd post the author is talking about the 2011 season having atypical (not usual) results then what those results were in terms of VBD and then posits the number of players that should have been drafted if you had perfect knowledge of how every player scored in 2011.

The real question is whether these changes are anomalous or representative of a shift in relative values. I have to say that given some of the NFL rules changes that have made it difficult for defensive players to hit the quarterback and protect the middle of the field, as well as the increasing prevalence of running back committees, I believe this is more the new rule than the exception. When you go on the clock for your first pick in your 2012 fantasy drafts and you want to take Aaron Rodgers, go ahead. It is not as crazy as it once was

First of all I think projections based off the previous season, instead of something more reasonable like the average of the last 3 seasons leads to a poor conclusion. If you use the last 3 seasons instead of one this gives you an average of 3 seasons performances, which will minimize the impact of a atypical season pretty well, only giving that season 33% of the average.

 Also the application of replacement level value (QB 16) isn't the way I would do it for a 10 team league. Baseline QB 5-6 would likely be more appropriate for such a league in my opinion and using that baseline would not cause the VBD numbers to be nearly as high, and therefore suggest that drafting QB later is a better idea, that top QB such as Rogers should be considered as high as round 2 or 3 perhaps, but not round one.

As far as where the players were actually drafted, that is the ADP not VBD and as I said in previous post mostly based on the previous years results, which is short sighted and not a very good way to make projections imho.

Its crazy that ADP on the QBs was that high. I certainly never bought into that and the way I use VBD would never lead me to that conclusion. 

As to you point about projections being wrong in the links thread. Of course they are. But rankings and ADP are also wrong every year. I would say that projection based ranking are more likely to be accurate than a ranking that isn't, and also more accurate than the ADP which is also always wrong. 

Some projections are better than others. That is one way to gain a competitive advantage over the market.

 

[Ilov80s]

@Biabreakable they were using the same method to calculate VBD numbers for all positions. Yes, using their method, it raised the scores for the top QBs, but it also raised the scores for top players at other positions. Here's the difference created by the way PFF did it from using the last starter. You are right that PFFs system aided QB:

QB+31, RB+48, WR+35.

So I looked back at how it wound impact VBD. Staff had a pretty big lead over the next few players so it only dropped him from 14 to 17. So it still was saying to draft 5 QBs in the first two rounds. 

 

[Joe Bryant]

Some projections are better than others. That is one way to gain a competitive advantage over the market.

Absolutely. When we get to the point with advanced players that all understand value and apply VBD and Supply and Demand, the game becomes very simply, "who can most accurately predict what players will do". 

J

 

[Biabreakable]

In regards to the issue of variable performance on a weekly basis, I think that is a good point. I know some people will try to project weekly totals with some consideration of weekly match ups and player totals based on that as a way to address this issue.

I don't do that. To me that is a separate thing. This is the who do I start decision. We may all agree that player X is more valuable than player Y is for the purpose of the entire season, but at the same time player Y may be a better start in week 3 than player X is because of match up or other considerations.

Once the season begins, and a few weeks of data are collected, the landscape of player values changes and you can start reprojecting for them based on how they are performing. Typically by week 4 there is enough data to begin doing this, as far as informing your who to start decisions.

What I do that does partially account for the variable nature of weekly performance of players is to project for upside, downside and a median range for projection for each player. Having this allows me to draft for upside, or consistency in my roster construction decisions. I will typically value consistency over upside in earlier rounds of the draft, then later on in the draft I will use upside projections to swing for the fences. If I find myself in a situation where I have drafted a lot of high upside risk players, then I may favor a player who is more consistent to try to cover that.

VBD can be used to account for these things. It just requires the use of a few different projections to develop what I would call the players projection range. The floor, ceiling and median range projection.

 

[GregR]

Agreed, which is why I always do my VBD chart and then pull in the current ADPs and highlight any spots where there seem to be clear discrepancies to find values and to see what really matches up to know when I have to take players.

Agree, which is why recently I've basically tried to use VDB to target roughly the most efficient starting lineup based on positional availability through the first 6-7 rounds and then look to fill out the rest of the roster with higher upside players.

This is pretty much how I did my draft strategies too.

I'd use VBD to come up with value, and then turn that into where players should be drafted if everyone drafted straight from value.  Then I compare it to ADP and highlight players who are being drafted later than they should.  I'd do something similar with my leagues draft tendencies as well.

Now I have identified players that represent an advantage to me if I can get them where I think they are going.  Now I look over the whole list and where my draft picks fall and try to cobble together the best combined team, hitting as many of those players as I can.  I'd normally go ahead and build the team via mock drafts in DD, sometimes taking over other team's drafts to test out different scenarios.  And I'm comparing the total points or total VBD (either works) for each potential team looking for the best one.

I'm also working in other considerations as I do this. Bye week combinations that are good or bad. Injury risk. Upside. Whether I'd want to handcuff someone in particular and how that affects me.

A final thing I'd take the time to consider is what happens if ADP and my beliefs on my league's draft tendencies end up being wrong. I not only try to identify the best team I think I can get, but I check out likely contingencies. From what happens if an unexpected bargain shows up, to what happens if I miss the last player in a tier I was kind of counting on.

As an example of the latter, let's say I'm hoping I'll get my steal RB in the 4th, but it's a drop off after him.  I also go through a mock draft and ask myself what happens if I miss my player, miss out on that RB tier? How does it affect the rest of my draft?  Do I end up taking my 3rd RB earlier than I'd have done otherwise, expecting to platoon my worsened RB2 I'm going to end up with?  If I take my RB3 early how does that affect the rest of my draft?  Or if I'm going to take another position in the 4th if I miss my RB, how is that going to trickle down, and which position should it be.

I would mock things out, so during the draft I'm not scrambling to figure this out when it happens.

 

[tonato]

I also like to "calculate" VBD on past seasons in dynasty leagues.

In a 16 teams league starting 2 RBs for example, I would take the fantasy points scored in the season by RB #32 as my baseline, him being the "worst starter of the position". I will then take the average of the 4 best RBs (just to average out if one or two players have "unnormal" seasons) to determine the added value of having a "stud" at the position. I will the calculate the "vbd-value" of the 8th best RB, the 12th best RB etc in blocks of 4 to see how the added value goes towards the 0 of the 32th RB.

Doing that for every position, even IDP, gives me a good feeling of the values of positions in that particular league, as each league can be different in terms of starting players AND scoring.

Of course the choice of the baseline is the most important, and choosing the "worst starter" is not optimal, so for 2017 I've thought up a little experiment I will be doing in my main dynasty league: the "worst waiver wire guy". Each week I will write down 2  pre waiver wire available players  and 2 first-come-first-serve available players that I would pick-up to start if I had to, and compile a virtual "waiver wire guy"-baseline. I'm sure the IDP positions will be quite strong, while QB and TE will become relatively more valueable than their 2 big brothers RB and WR in a normal "worst starter"-baseline, due to a lot (at QB virtually ALL) of starters being rostered.

 

[Long Ball Larry]

Not sure if this is the right spot for this but perhaps Joe can help out. Years ago when there still were magazines I read an article that expanded on VBD and converting the fantasy points to auction dollars. Would love to see this article again. It had something to do with figuring how many fantasy points were to be scored by your starters and then dividing that by how much money you allotted for your starters. You would then multiply the fantasy points by the dollars per point. Not sure of the details anymore so would appreciate it if someone could refresh my memory or link the article.

This thread is probably a good place to look: https://forums.footballguys.com/forum/topic/707479-converting-projections-into-auction-values/

Several of the posts basically outline the methodology that I mimicked when I started auctions a few years ago.

 

[GregR]

Not sure if this is the right spot for this but perhaps Joe can help out. Years ago when there still were magazines I read an article that expanded on VBD and converting the fantasy points to auction dollars. Would love to see this article again. It had something to do with figuring how many fantasy points were to be scored by your starters and then dividing that by how much money you allotted for your starters. You would then multiply the fantasy points by the dollars per point. Not sure of the details anymore so would appreciate it if someone could refresh my memory or link the article.

I have a walk through of the basic method at the end of this auction article:  http://subscribers.footballguys.com/2011/11russell_auctions.php

Maurile has a more in depth method (http://subscribers.footballguys.com/2006/06tremblay_auctionpricing.php) which I'd suggest instead. You might want to skim mine to get the gist, then go into his which advances the topic. His takes into account things like how many fantasy games the RB16 actually starts (versus how many he is benched for the team's RB3).  MT's method is built into the draft tools. I'd advise skimming the article and then working with the tool to set the prices.  Also, during the auction then, the tools will constantly adjust prices to reflect the changes in available money and the remaining player pool.

 

[Joe Bryant]

I have a walk through of the basic method at the end of this auction article:  http://subscribers.footballguys.com/2011/11russell_auctions.php

Maurile has a more in depth method (http://subscribers.footballguys.com/2006/06tremblay_auctionpricing.php) which I'd suggest instead. You might want to skim mine to get the gist, then go into his which advances the topic. His takes into account things like how many fantasy games the RB16 actually starts (versus how many he is benched for the team's RB3).  MT's method is built into the draft tools. I'd advise skimming the article and then working with the tool to set the prices.  Also, during the auction then, the tools will constantly adjust prices to reflect the changes in available money and the remaining player pool.

What Greg and Larry said. Thanks Guys.

J

 

[cloppbeast]

The baseline has to be the average of every other rosterable player. Randomly picking a baseline introduces bias.

 

[GregR]

So you can play around with baselines until you get the result you want?

Let me draw a comparison. Seismic data is used by scientists and oil companies to image the layers of rock below ground. They record sound waves after they've bounced around between the different layers in the ground, and then process the result to get an image of below ground. Ultrasound is a similar technology. Something like this image is the result.

While the method is firmly based in science, there are a lot of parameters that can change how the result looks. Experience goes into deciding which ones are right. That doesn't mean it isn't useful. It is hugely useful, and the primary method by which we figure out where to drill for oil and gas.

VBD is similar. Results can turn out differently depending on what baselines you set. While VBD is not as hugely complex, it does take a little experience with fantasy football to be able to vet the results. For beginners, that experience can come from other sources, like Joe. But developing it yourself is helpful. Just the act of moving baselines and seeing how things change can itself be useful and make us question our preconceived notions.

When you boil it down, VBD is doing the same kind of value calculation that you are doing in your head, whether consciously or unconsciously. A major difference is that VBD will be more consistent than your head, and will save you time. You encapsulate as many of your feelings about a player into the projection as you can and use VBD to develop a directly comparable value across positions. After that, every comparison you do using the results includes all of those factors you thought impacted the players. If you're unconsciously doing this in your head, and you're down to 90 seconds to pick, you're not even likely to remember all of those factors you thought would impact each of the players at different positions you're considering taking this round. Not to mention the players you'll take in rounds after that are impacted by this pick.

 

[cloppbeast]

Results can turn out differently depending on what baselines you set. That doesn't mean VBD is not extremely useful. While VBD is not as hugely complex, it does take a little experience with fantasy football to be able to vet the results.

The dissertations on this thread seem to have the goal of convoluting VBD as much as possible. But it is quite simple: value comes from how player compares to the rest of his position.

The unnecessary complexity comes from the discussions about the baseline. As you say, results can turn out differently by using different baselines. Or more accurately: wrong results come from using the wrong baselines.

 

[GregR]

The dissertations on this thread seem to have the goal of makning VBD as complex as possible. But it is quite simple: value comes from how player compares to the rest of his position.

The unnecessary complexity comes from the discussions about the baseline. As you say, results can turn out differently by using different baselines. Or more accurately: wrong results come from using the wrong baselines.

I would say results can be better and worse. A reasonable baseline is generally going to give pretty good results even if it isn't the very most ideal for your league. Enough to be of use to you in making decisions.

It isn't tough to use a reasonable baseline even for an FF novice because people like Joe, and the tools available, lend their knowledge to help in setting one.

 

[abbottjamesr]

In my mind there should be a theoretical perfect strategy for ranking players or assigning auction values if you were given perfect projections and a known scoring system, line up requirements and roster depths.  VBD is the closest we have come to providing a method of calculating this (that has been published).  My issue with VBD in the form as presented in the articles is that it oversimplifies it.  I understand why you need to for an article that aims to help all regardless of league scoring types, roster requirements, etc.  There are so many variable that come into play and  @GregR is right that it is pretty easy to put together a decent ranking using a basic VBD, especially with the resources provided by the site.

But many (some, maybe only me) are looking for a more rigorous way to calculate draft rankings and auction values that seams less arbitrary.  I certainly don't have all the answers to these issues but every year I try to make some progress on my methodology for player rankings and auction values.  

Baselines: Baselines for VBD are a function of only starting requirements and number of teams.  Assuming no one got injured in football and our projections were correct, in a 12 team start 2 RB league only the top 23 RB's have value.  But in the real world players get injured.  I posted the historical games played per season above that I use to calculate my baselines based off of the number of RB games that need to be covered in a season by all teams.  For instance in my 12 team start 2 RB, 1 flex league my baseline is RB41.  This is the number of RB's that are needed to cover all the regular season FF games assuming historical injury rates.  

Lastly, I think part of the problem people have with VBD is that if you draft according to it on the DD app your teams just don't look good, and the rate my team function tends to agree.  If I do the same auction and use my model for value I can get teams that are rated at 99% chance to make the playoffs.  If the VBD model being used was perfect and every team was using it there should be no way to get an above average team assuming the rating function is using the same projections as the VBD model.  I don't know how to reconcile this. 

 

[Long Ball Larry]

In my mind there should be a theoretical perfect strategy for ranking players or assigning auction values if you were given perfect projections and a known scoring system, line up requirements and roster depths.  VBD is the closest we have come to providing a method of calculating this (that has been published).  My issue with VBD in the form as presented in the articles is that it oversimplifies it.  I understand why you need to for an article that aims to help all regardless of league scoring types, roster requirements, etc.  There are so many variable that come into play and  @GregR is right that it is pretty easy to put together a decent ranking using a basic VBD, especially with the resources provided by the site.

But many (some, maybe only me) are looking for a more rigorous way to calculate draft rankings and auction values that seams less arbitrary.  I certainly don't have all the answers to these issues but every year I try to make some progress on my methodology for player rankings and auction values.  

Baselines: Baselines for VBD are a function of only starting requirements and number of teams.  Assuming no one got injured in football and our projections were correct, in a 12 team start 2 RB league only the top 23 RB's have value.  But in the real world players get injured.  I posted the historical games played per season above that I use to calculate my baselines based off of the number of RB games that need to be covered in a season by all teams.  For instance in my 12 team start 2 RB, 1 flex league my baseline is RB41.  This is the number of RB's that are needed to cover all the regular season FF games assuming historical injury rates.  

Lastly, I think part of the problem people have with VBD is that if you draft according to it on the DD app your teams just don't look good, and the rate my team function tends to agree.  If I do the same auction and use my model for value I can get teams that are rated at 99% chance to make the playoffs.  If the VBD model being used was perfect and every team was using it there should be no way to get an above average team assuming the rating function is using the same projections as the VBD model.  I don't know how to reconcile this. 

What I think that you are ultimately looking for is the way to anticipate variance and maximize its upside, which I think is a good idea,though inherently extremely difficult.  I like the core concept in your earlier post about tiering the quartiles (roughly) of individual week outcomes, but I need to go back and read and think about it more closely.

 

[GregR]

On the subject of baselines and knowing if you have a good one or if it needs adjusting. It is pretty simple to check, and quick with the tools here. Let's say you have a question like, "This looks like QBs are going earlier than I think they should. Is that right or not?"

Mock drafts in Draft Dominator take seconds to do when you let them auto-run. Set up a mock with all teams drafting ADP except for one. Check the results for that team that used VBD.  If they did not take any of the specific players you are questioning, you might wish to redo it with a different draft slot until you get one who takes the player.

Did they end up with a good team?  In starter points?  In starter+backup points?   Jot down or export the results or take a quick screen shot so you can refer back.

Now change just 1 thing, such as the QB baseline, and redo the mock. Note how the results changed.  Did the team get better or worse from the change?

You can do mocks so fast in Draft Dominator you can just cycle through and generate resulting teams and compare them.  You can mix up how many teams use ADP and how many VBD, etc. It's very simple to answer your own questions and determine what baseline is giving the best team.  Just remember to only change 1 thing each run so you know which parameter is causing your change.

 

[abbottjamesr]

On the subject of baselines and knowing if you have a good one or if it needs adjusting. It is pretty simple to check, and quick with the tools here. Let's say you have a question like, "This looks like QBs are going earlier than I think they should. Is that right or not?"

Mock drafts in Draft Dominator take seconds to do when you let them auto-run. Set up a mock with all teams drafting ADP except for one. Check the results for that team that used VBD.  If they did not take any of the specific players you are questioning, you might wish to redo it with a different draft slot until you get one who takes the player.

Did they end up with a good team?  In starter points?  In starter+backup points?   Jot down or export the results or take a quick screen shot so you can refer back.

Now change just 1 thing, such as the QB baseline, and redo the mock. Note how the results changed.  Did the team get better or worse from the change?

You can do mocks so fast in Draft Dominator you can just cycle through and generate resulting teams and compare them.  You can mix up how many teams use ADP and how many VBD, etc. It's very simple to answer your own questions and determine what baseline is giving the best team.  Just remember to only change 1 thing each run so you know which parameter is causing your change.

My point was if every team is using VBD then no team in theory should be able to pick a better team than any other if the rating scale is the same as that generated the VBD.  But that is not what I see in DD from the rank my team function or just from looking at the results*. This isn't meant to be a knock on DD or the rate my team function.  It is more a general issue with VBD.  I understand the next comment will be "you need to adjust your baseline to get the results you are looking for".  My response will be that a proper tool will have some analytical way of finding the ideal baseline.  

*this is entirely dependent on my using DD correctly which is certainly not a given due to the vast number of parameters involved in the program.

 

[GregR]

My point was if every team is using VBD then no team in theory should be able to pick a better team than any other if the rating scale is the same as that generated the VBD.  But that is not what I see in DD from the rank my team function or just from looking at the results*. This isn't meant to be a knock on DD or the rate my team function.  It is more a general issue with VBD.  I understand the next comment will be "you need to adjust your baseline to get the results you are looking for".  My response will be that a proper tool will have some analytical way of finding the ideal baseline.  

*this is entirely dependent on my using DD correctly which is certainly not a given due to the vast number of parameters involved in the program.

I would almost never expect the bolded to be true.  It would be a fluke it it happened.

There is variation caused by the fact that players drop off in value differently from one position to another. VBD tries to express the value that results.  It does not smooth the values inherent across positions. It just expresses them. There is nothing about VBD to think that everyone drafting from it strictly (which you shouldn't do anyway) would result in every team being equal..  The top few overall players generally are so much more valuable that it is difficult to make up that difference for any team that does not get one.  Put simply, the #1 pick plus the #24 pick is frequently significantly more valuable than the #7 and #18 picks. Because the gap between #1 and #7 is often far bigger than the gap between #18 and #24. 

This trend is why the 3rd round reverse serpentine draft order was created. So the teams who don't get those top few picks pick earliest again in the 3rd to get an extra chance to make up that huge value.

Also, the order in which players are depleted is a major component of value, and it is one that normal VBD is not meant to address.   Simple example, you are picking at 2.10 and the best VBD value is RB17 with WR7 just behind him.  If you could know that all 4 picks before your 3.3 are WRs, then you'd be a fool to take RB17 now because he'll be there next round.  RB17 2nd round and WR12 are worth less than WR7 and RB17. 

That is an intentionally extreme example where the order in which players will be depleted completely dominated what was the right move for your team. But it definitely is an important factor. This is the component of value we go after when we talk about "Dynamic VBD" which basically means looking at how much you expect each position to drop off between your current pick and your next pick.  And then try to take the combinations of players who give the most value.  Which frequently may not be the highest VBD player available at each pick.

Check my post earlier where I gave what my draft prep was. I compared VBD to ADP to note players who are lasting later than they should, then try to work out how to get the combo of players that frequently includes them, and fills a starting lineup's needs.  That might mean passing up a higher VBD player in some round to take another position, because I expect to get a steal later at the position I passed on.

 

[GregR]

If this helps, go back to Joe's example in the original VBD article. It was something like, a 2 team league with 2 roster spots that just starts 1 RB and 1 WR and no other positions.  A simple case meant to illustrate the point.

RB1: 250
RB2: 230

WR1: 200
WR2:  160

VBD should hold as true for this example as any other league setup. You can see pretty clearly, WR1 is a 40 point advantage while RB1 is a 20 point advantage, per VBD. There is no way the team who doesn't get WR1 can end up with an equal team.  The fact it is just 2 teams and just 2 starting slots isn't causing that. You can take real projections or year-end results for 12 teams and normal lineups and run into the same thing.

 

[abbottjamesr]

I would almost never expect the bolded to be true.  It would be a fluke it it happened.

There is variation caused by the fact that players drop off in value differently from one position to another. VBD tries to express the value that results.  It does not eliminate the values inherent across positions. There is nothing about VBD to think that everyone drafting from it strictly (which you shouldn't do anyway) would result in every team being equal..  The top few overall players generally are so much more valuable that it is difficult to make up that difference for any team that does not get one.  Put simply, the #1 pick plus the #24 pick is frequently significantly more valuable than the #7 and #18 picks. Because the gap between #1 and #7 is often far bigger than the gap between #18 and #24. 

This trend is why the 3rd round reverse serpentine draft order was created. So the teams who don't get those top few picks pick earliest again in the 3rd to get an extra chance to make up that huge value.

Also, the order in which players are depleted is a major component of value, and it is one that normal VBD is not meant to address.   Simple example, you are picking at 2.10 and the best VBD value is RB17 with WR7 just behind him.  If you could know that all 4 picks before your 3.3 are WRs, then you'd be a fool to take RB17 now because he'll be there next round.  RB17 2nd round and WR12 are worth less than WR7 and RB17. 

That is an intentionally extreme example where the order in which players will be depleted completely dominated what was the right move for your team. But it definitely is an important factor. This is the component of value we go after when we talk about "Dynamic VBD" which basically means looking at how much you expect each position to drop off between your current pick and your next pick.  And then try to take the combinations of players who give the most value.  Which frequently may not be the highest VBD player available at each pick.

Check my post earlier where I gave what my draft prep was. I compared VBD to ADP to note players who are lasting later than they should, then try to work out how to get the combo of players that frequently includes them, and fills a starting lineup's needs.  That might mean passing up a higher VBD player in some round to take another position, because I expect to get a steal later at the position I passed on.

Thanks the great reply, and I agree with every thing you wrote above.  Your post above should be the exact reason that snake drafts should be replaced with auctions.  The only snake drafts I do these days are the occasional MFL 10.  

Most of my work on this has been done with the goal on auction drafts and it is my fault for not specifying that.  I suppose I thought the same issues would carry over from one style to the other.   I also want to make sure you know, I am not trying to be critical of VBD in general.  Just looking to improve on areas where I feel it doesn't seem to produce consistent results.

 

[GregR]

Yes, auctions are definitely different.

If everyone auctioned from the same set of prices, with the same percent of cap spent on starters, and everyone bid up to exactly that player's price and never went over or under... then I'd expect that you would probably end up with more parity in a result. Though the variation in the pool of players I still wouldn't expect it 100%. 

And I don't think Draft Dominator does the mocks quite like that. It's been awhile, but doesn't it have a file that it uses to make the results a bit more like real world ones?  I seem to recall I'd go open it up in and edit the multipliers it used to make them fit my league better.  Or maybe that was used to compute the VBD values even, I don't recall.

 

[GregR]

I know what I forgot in that last one.  If you're using Maurile's method for calculating auction values, in particular, then it is taking into account things like RB15 generally only starts 8 fantasy each season while RB10 might start 10 times.  Making RB10 more valuable than just his VBD number would express. I don't know what the numbers are for those players exactly, just giving an example.

So after the auction you wouldn't expect total points of the teams to be equal. But each player's points times the fraction of the games he starts, might be close to equal if no other randomness is intentionally introduced.

 

[cloppbeast]

My point was if every team is using VBD then no team in theory should be able to pick a better team than any other if the rating scale is the same as that generated the VBD.  But that is not what I see in DD from the rank my team function or just from looking at the results*. This isn't meant to be a knock on DD or the rate my team function.  It is more a general issue with VBD.  I understand the next comment will be "you need to adjust your baseline to get the results you are looking for".  My response will be that a proper tool will have some analytical way of finding the ideal baseline.  

*this is entirely dependent on my using DD correctly which is certainly not a given due to the vast number of parameters involved in the program.

 

[Spike]

22?

That's a mistake in the calculations somewhere.  Bad cell reference, etc.  It isn't possible to have that many in the top 50 or 60.

I agree - that's why I asked about it.

 

[Spike]

I agree - that's why I asked about it.

Just tried with the version of the VBD app that came out on 7/6/17.  21 QBs in the top 60 players.

Could I get some help here?

 

[Spike]

Joe,

To whom do I send an email for assistance on the app as my findings on the lists are extraordinarily skewed?

 

[LongDuckDong]

Joe/whomever - 

My issue with VBD (and I was an early adopter and disciple) has always been the issue of player pools as a whole and the fact that while DVBD addresses this somewhat, there should be a mathematical way to address the depth of the player pool in the VBD calculation. For example:

Assuming a 4 team league that drafts 1RB and 1WR. If the RB Pool projected points is 50,49,48,10 and the WR pool projected points is 50,12,11,10, then I think WR1 is the best pick, even though his VBD score is same as RB1 (I know this is an overly simplistic example). But somehow the overall percentage of static VBD that each player represents out of the pool should factor into the calculation. And I haven't figured out the best way to do it. 

 

[abbottjamesr]

Joe/whomever - 

My issue with VBD (and I was an early adopter and disciple) has always been the issue of player pools as a whole and the fact that while DVBD addresses this somewhat, there should be a mathematical way to address the depth of the player pool in the VBD calculation. For example:

Assuming a 4 team league that drafts 1RB and 1WR. If the RB Pool projected points is 50,49,48,10 and the WR pool projected points is 50,12,11,10, then I think WR1 is the best pick, even though his VBD score is same as RB1 (I know this is an overly simplistic example). But somehow the overall percentage of static VBD that each player represents out of the pool should factor into the calculation. And I haven't figured out the best way to do it. 

 In this example it doesn't matter if doing a snake draft.  Who ever has pick 1 is going to take one of the 50 point players.  Who ever is second takes the other. Picks end up being.

1 - 50Rb
2 - 50WR
3 - 49RB
4 - 48RB
4 - 12WR
3 - 11WR
2 - 10RB
2 - 10WR

VBD holds up with out exception when you are comparing end of season point totals.  I think what it fails to capture is:

1.  How points are scored on a weekly basis.
     -  would you rather have a player that posted 8, 8, 8, 8, 8 over 5 weeks or one that had 12 three weeks and 2 the other two at random?

2.  Ability to find starter production in any given week from the WW or extremely cheap assets.  
    -  This is the reason QB's are not valuable in standard leagues.  Andy Dalton was QB 18 last season on ppg in my league.  He had 6 weeks of starter production.  Derek Car was QB6 in PPG and had 7 weeks of top 12 production.  Only 8 RB's had at least 6 weeks of RB1 production last year.

 

[LongDuckDong]

VBD isn't used end of season. It's used preseason to try to get an understanding of players values across positions based on preseason projections. 

We will have to disagree that it holds up without exception. 

 

[abbottjamesr]

VBD isn't used end of season. It's used preseason to try to get an understanding of players values across positions based on preseason projections. 

We will have to disagree that it holds up without exception. 

What I meant is that it is commonly used with projections that are for season totals like your example.  That was my bad if I made it seam like something else. 

If we had perfect projections for every player for total points for the season we could figure VBD that would make it impossible for any team to have more points than any other team.  If you deviated from VBD you would be giving up points.  I don't believe this is up for debate.  The issue is that we play a weekly game and VBD would not be as successful in predicting which teams would have won every week just using total points projections.

 

[patsfangr]

Joe,

To whom do I send an email for assistance on the app as my findings on the lists are extraordinarily skewed?

Spike, if you didn't get an answer to this already, I'd suggest sending your question to Bruce Henderson. He's the DD Classic guru; and is usually very prompt in responding. 

 

[Ilov80s]

What I meant is that it is commonly used with projections that are for season totals like your example.  That was my bad if I made it seam like something else. 

If we had perfect projections for every player for total points for the season we could figure VBD that would make it impossible for any team to have more points than any other team.  If you deviated from VBD you would be giving up points.  I don't believe this is up for debate.  The issue is that we play a weekly game and VBD would not be as successful in predicting which teams would have won every week just using total points projections.

Both you and Mr. Duck Dong Long make good points and I think both of you are right on.

 

[patsfangr]

Most of the guys in this discussion have apparently been working with managing your own VBD settings for some time. I, on the other hand, never paid much attention to it, despite using the DD (Classic) for over 15 years. I (lazily) just left it all up to "Joe's Secret Formula", which worked fine for me in most of the leagues I played in ... except for one gnawing problem I had to deal with every season. I explained it in a post I put in the wrong forum a little over a week ago (a mistake for which I got a barrage of flames!). Here's that post:

.......I've been using FBG DD Classic for about 15 years. Throughout those years, I've struggled with the position in the rankings (Player Pool) of the Def/Sts, PKs and IDPs. They are consistently significantly higher than where those positions are drafted in all leagues that I play in. As a result, when I get a few rounds into a live draft, those positions begin to "clog" the screen in the DD Player Pool. (I should note that I try to enter picks into the DD as they happen to keep the Player Pool rankings "current" with my live draft.) 

During this off season, I decided to find a way to resolve this problem. I have been experimenting with the "VBD Baseline" in the Setup window. In the past, I've never used that window. I've just accepted the default, or "Joe's Secret Formula", as the best option because I honestly don't understand how the program uses the VBD to affect the rankings. What I've discovered is that by entering my own numbers in the "User Configuration by Position Rank" section, I can change the way that the positions are weighted for integration into the Player Pool rankings. 

I'm wondering if anybody else has done this; and, if so, how satisfying have the results been for you in getting the rankings to work for you during a draft. I'm also wondering if anybody has any comments on why "Joe's Secret Formula" gets these "out of whack" rankings for these positions. I'd appreciate any comments or discussion on this, or the VBD Baseline tool in general. .........

As you can see, I'm truly a novice with the VBD "User Configuration" options. I could use any "VBD User Configuration for Dummies" advise and instruction that you guys can offer on the best way to determine the numbers to plug in for both the "Position Rank" and "Point Value" windows; as well as your personal opinions on which is the better of the two. 

I play in 2 PPR leagues (MFL host) and 2 non-PPR leagues (Yahoo), BOTH of which are Keeper leagues (with different keeper rules). In brief, all of those leagues have at least the top 10 to 15 RBs and WRs out of the draft pool. I'm also in a couple of FFPC leagues, which are PPR redraft, with 1.5 pts/rec for TEs. So, as you can see, I'm dealing with a lot of variables across my leagues. 

Jump in when ready! 

 

[DropKick]

Joe/whomever - 

My issue with VBD (and I was an early adopter and disciple) has always been the issue of player pools as a whole and the fact that while DVBD addresses this somewhat, there should be a mathematical way to address the depth of the player pool in the VBD calculation. For example:

Assuming a 4 team league that drafts 1RB and 1WR. If the RB Pool projected points is 50,49,48,10 and the WR pool projected points is 50,12,11,10, then I think WR1 is the best pick, even though his VBD score is same as RB1 (I know this is an overly simplistic example). But somehow the overall percentage of static VBD that each player represents out of the pool should factor into the calculation. And I haven't figured out the best way to do it. 

I was also an early adopter and saw VBD as what it was; a means to assign player values given your league's scoring system and starting requirements.  It allowed you to compare players across positions and was customized to your league; a method of quantification without the bias of preconceived notions and group think.

It didn't have the problems many people claimed.... it wouldn't lead to me take 5 QBs in the first 6 rounds,etc.  However, once you understood the mechanics of the math, some issues were obvious.  Special teams and kickers would start to outrank position players once you reached a certain depth in the draft.  You could address that within the calculations, such as by choosing a deeper baseline or simply with your own judgement.  Maybe that high value D/ST is worth reaching for in your scoring system but the kicker, and the randomness of the position, can be deferred until much later in the draft.

VBD was a good tool; I didn't expect it to do everything for me or be the Gospel according to Joe.  It was most valuable early (and later) in the draft but I would often stray based on subjective things like upside; hand-cuffing RBs from run oriented teams and the like.

I don't use VBD anymore since all my leagues are deep keeper or dynasty and I'm drafting with a mix of current and future needs.  I imagine the calculations could have become far more sophisticated since the early days and "real time" calculations within online drafting tools could be invaluable.

In your example, both the RB1 and WR1 have a value of 40 (value above "worst" starter; 50-10=40).    A simple extension of the formula might be to add in value above average starter.  The composite value of the WR1 would jump to 69 (40 + 29) while the RB1 would be 50.75; reflecting the deeper RB pool.  Of course, you could weight the two numbers or use more complex math that considers variance, etc.   I think you can fool yourself into needless precision on top of numbers (projections) that are only guesses to begin with...

 

[abbottjamesr]

I was also an early adopter and saw VBD as what it was; a means to assign player values given your league's scoring system and starting requirements.  It allowed you to compare players across positions and was customized to your league; a method of quantification without the bias of preconceived notions and group think.

It didn't have the problems many people claimed.... it wouldn't lead to me take 5 QBs in the first 6 rounds,etc.  However, once you understood the mechanics of the math, some issues were obvious.  Special teams and kickers would start to outrank position players once you reached a certain depth in the draft.  You could address that within the calculations, such as by choosing a deeper baseline or simply with your own judgement.  Maybe that high value D/ST is worth reaching for in your scoring system but the kicker, and the randomness of the position, can be deferred until much later in the draft.

VBD was a good tool; I didn't expect it to do everything for me or be the Gospel according to Joe.  It was most valuable early (and later) in the draft but I would often stray based on subjective things like upside; hand-cuffing RBs from run oriented teams and the like.

I don't use VBD anymore since all my leagues are deep keeper or dynasty and I'm drafting with a mix of current and future needs.  I imagine the calculations could have become far more sophisticated since the early days and "real time" calculations within online drafting tools could be invaluable.

In your example, both the RB1 and WR1 have a value of 40 (value above "worst" starter; 50-10=40).    A simple extension of the formula might be to add in value above average starter.  The composite value of the WR1 would jump to 69 (40 + 29) while the RB1 would be 50.75; reflecting the deeper RB pool.  Of course, you could weight the two numbers or use more complex math that considers variance, etc.   I think you can fool yourself into needless precision on top of numbers (projections) that are only guesses to begin with...

I was going to respond with a long post, but DropKick wrote a very good response with regard to VBD usage.

As far as keepers go, they don't really change much.  You can just assume they have all ready been drafted and mark them off you sheet.  The resulting players are where you start your picks.  

 

[Joe Bryant]

Spike, if you didn't get an answer to this already, I'd suggest sending your question to Bruce Henderson. He's the DD Classic guru; and is usually very prompt in responding. 

Hi Pats,

By far the best way to ask that is our help desk here https://sportsguys.zendesk.com/hc/en-us  If you don't see what you're looking for, use the submit a request and we'll get you pointed to the right people. Thanks.

 

[Joe Bryant]

I was also an early adopter and saw VBD as what it was; a means to assign player values given your league's scoring system and starting requirements.  It allowed you to compare players across positions and was customized to your league; a method of quantification without the bias of preconceived notions and group think.

It didn't have the problems many people claimed.... it wouldn't lead to me take 5 QBs in the first 6 rounds,etc.  However, once you understood the mechanics of the math, some issues were obvious.  Special teams and kickers would start to outrank position players once you reached a certain depth in the draft.  You could address that within the calculations, such as by choosing a deeper baseline or simply with your own judgement.  Maybe that high value D/ST is worth reaching for in your scoring system but the kicker, and the randomness of the position, can be deferred until much later in the draft.

VBD was a good tool; I didn't expect it to do everything for me or be the Gospel according to Joe.  It was most valuable early (and later) in the draft but I would often stray based on subjective things like upside; hand-cuffing RBs from run oriented teams and the like.

I don't use VBD anymore since all my leagues are deep keeper or dynasty and I'm drafting with a mix of current and future needs.  I imagine the calculations could have become far more sophisticated since the early days and "real time" calculations within online drafting tools could be invaluable.

In your example, both the RB1 and WR1 have a value of 40 (value above "worst" starter; 50-10=40).    A simple extension of the formula might be to add in value above average starter.  The composite value of the WR1 would jump to 69 (40 + 29) while the RB1 would be 50.75; reflecting the deeper RB pool.  Of course, you could weight the two numbers or use more complex math that considers variance, etc.   I think you can fool yourself into needless precision on top of numbers (projections) that are only guesses to begin with...

Thanks dropkick. Good points. Where it's radically changed from the original days is with the new apps and recalculating the values based on new information after each pick. Something we could never do in the older days. In 1995, in a keeper league, you could you prepare for the draft by taking all the kept players out of the pool of available players. This was a huge advantage over an owner who failed to calculate how keepers affected values. 

Now with the DD App, you make that calculation after every single pick of the draft. It's amazing.

 

[GregR]

Joe/whomever - 

My issue with VBD (and I was an early adopter and disciple) has always been the issue of player pools as a whole and the fact that while DVBD addresses this somewhat, there should be a mathematical way to address the depth of the player pool in the VBD calculation. For example:

Assuming a 4 team league that drafts 1RB and 1WR. If the RB Pool projected points is 50,49,48,10 and the WR pool projected points is 50,12,11,10, then I think WR1 is the best pick, even though his VBD score is same as RB1 (I know this is an overly simplistic example). But somehow the overall percentage of static VBD that each player represents out of the pool should factor into the calculation. And I haven't figured out the best way to do it. 

 In this example it doesn't matter if doing a snake draft.  Who ever has pick 1 is going to take one of the 50 point players.  Who ever is second takes the other. Picks end up being.

1 - 50Rb
2 - 50WR
3 - 49RB
4 - 48RB
4 - 12WR
3 - 11WR
2 - 10RB
2 - 10WR

VBD holds up with out exception when you are comparing end of season point totals.  I think what it fails to capture is:

1.  How points are scored on a weekly basis.
     -  would you rather have a player that posted 8, 8, 8, 8, 8 over 5 weeks or one that had 12 three weeks and 2 the other two at random?

2.  Ability to find starter production in any given week from the WW or extremely cheap assets.  
    -  This is the reason QB's are not valuable in standard leagues.  Andy Dalton was QB 18 last season on ppg in my league.  He had 6 weeks of starter production.  Derek Car was QB6 in PPG and had 7 weeks of top 12 production.  Only 8 RB's had at least 6 weeks of RB1 production last year.

VBD is very good, but it doesn't hold up without exception if we're talking about telling you who to draft. It doesn't take the impact of the order in which players will be taken into account. Something that people use DVBD (Dynamic VBD) to do by comparing the expected drop at each position between now and your next pick.

Those teams were only even because his example drop offs are identical in the two positions, just one is inverted compared to the other.  Consider a slightly changed example:  RB: 50, 45, 40, 10.   WR: 50, 30, 20, 10.  A straight VBD draft then would be:

1 - 50Rb
2 - 50WR
3 - 45RB
4 - 40RB
4 - 30WR
3 - 20WR
2 - 10RB
1 - 10WR

Team 4 ends with 70 points, Team 3 with 65, and Teams 1 and 2 with 60.

If we lock in how the earlier draft played out, Team 3 makes a mistake at his pick by following VBD. If he took the 30WR instead of the 45RB that VBD says, then he would had a minimum of 70 points instead of 65. Where his picks fall, he's assured that he's going to get 1 player out of 45RB or 40RB and 1 player out of 30WR or 20WR.  The difference in WRs he can get is 10 points, the difference in RBs is 5 points. So he should take the best WR first, 30WR.  Even though 30WR has a VBD of 20 while 45RB has a VBD of 35.

1 - 50Rb
2 - 50WR
3 - 45RB 30WR
4 - 40RB 45Rb
4 - 30WR 20WR
3 - 20WR 40RB
2 - 10RB
1 - 10WR

Though the draft shouldn't have played out how it did before Team 3. Because Team 2 made a mistake by following VBD values only, and should not have taken 50WR. He should have taken 45RB. He would end up with 45RB and 20WR for 65 points, instead of 50WR and 10RB for 60.

(Edit to add):  And if Team 2 had drafted right, then Team 3 should then have taken 40RB with his pick (for a combo of 40RB and 30WR for 70 points). Which means the best WR, 50WR, should have been taken by Team 4, after two RBs who both had lower VBD values. 

Which if we go back to the original point by LongDuckDong, the example I gave still fits his hypothetical that the best WR should be the most valuable because he outscores the total of his position more than does the best RB outscore his.  But when we take ALL factors of value into account, including the differences in the positional drop off, we find that the second RB was part of Team 2's best team, while the best WR was not.

I agree with LongDuckDong's reasoning that conceptually, you'd think the best WR should be more valuable (at least when we ignore the DVBD aspect of value) in both our examples.  And I've spent awhile thinking about this a few times and never felt like I fully proved to myself that was actually right, nor came up with a method I was happy with in expressing it. 

And even if we do, you still need to work in the DVBD-part of value to actually get at your team's optimal draft.

 

[LongDuckDong]

Greg thank you for articulating it better than me. And you're right, it's a DVBD process, but even so, there has to be a mathematical way to articulate it - some sort of evaluation of remaining player pool. Now, that being said, using 4X2 examples is a bit of a strawman, it's only for illustration purposes, but I do truly believe that player pool considerations should be part of VBD, and maybe it's just individual secret sauce to do it.

 

[SeniorVBDStudent]

VBD is a sham.  @SeniorVBDStudent has been enrolled for 14 years and still hasn't graduated!

Its a nostalgia thing.  I done gageated.

 

[DropKick]

VBD is very good, but it doesn't hold up without exception if we're talking about telling you who to draft. It doesn't take the impact of the order in which players will be taken into account. Something that people use DVBD (Dynamic VBD) to do by comparing the expected drop at each position between now and your next pick.

Those teams were only even because his example drop offs are identical in the two positions, just one is inverted compared to the other.  Consider a slightly changed example:  RB: 50, 45, 40, 10.   WR: 50, 30, 20, 10.  A straight VBD draft then would be:

1 - 50Rb
2 - 50WR
3 - 45RB
4 - 40RB
4 - 30WR
3 - 20WR
2 - 10RB
1 - 10WR

Team 4 ends with 70 points, Team 3 with 65, and Teams 1 and 2 with 60.

If we lock in how the earlier draft played out, Team 3 makes a mistake at his pick by following VBD. If he took the 30WR instead of the 45RB that VBD says, then he would had a minimum of 70 points instead of 65. Where his picks fall, he's assured that he's going to get 1 player out of 45RB or 40RB and 1 player out of 30WR or 20WR.  The difference in WRs he can get is 10 points, the difference in RBs is 5 points. So he should take the best WR first, 30WR.  Even though 30WR has a VBD of 20 while 45RB has a VBD of 35.

1 - 50Rb
2 - 50WR
3 - 45RB 30WR
4 - 40RB 45Rb
4 - 30WR 20WR
3 - 20WR 40RB
2 - 10RB
1 - 10WR

Though the draft shouldn't have played out how it did before Team 3. Because Team 2 made a mistake by following VBD values only, and should not have taken 50WR. He should have taken 45RB. He would end up with 45RB and 20WR for 65 points, instead of 50WR and 10RB for 60.

(Edit to add):  And if Team 2 had drafted right, then Team 3 should then have taken 40RB with his pick (for a combo of 40RB and 30WR for 70 points). Which means the best WR, 50WR, should have been taken by Team 4, after two RBs who both had lower VBD values. 

Which if we go back to the original point by LongDuckDong, the example I gave still fits his hypothetical that the best WR should be the most valuable because he outscores the total of his position more than does the best RB outscore his.  But when we take ALL factors of value into account, including the differences in the positional drop off, we find that the second RB was part of Team 2's best team, while the best WR was not.

I agree with LongDuckDong's reasoning that conceptually, you'd think the best WR should be more valuable (at least when we ignore the DVBD aspect of value) in both our examples.  And I've spent awhile thinking about this a few times and never felt like I fully proved to myself that was actually right, nor came up with a method I was happy with in expressing it. 

And even if we do, you still need to work in the DVBD-part of value to actually get at your team's optimal draft.

We're talking about a greatly simplified scenario where you're all but certain of the picks.  After all, each team needs just one RB and WR.  No one will load up at RB or go zero RB or anything else unpredictable.  During the dynamics of the season, things will prove to be quite different than we guessed at draft time..

I understand you're trying to demonstrate that the best "value" at each pick isn't guaranteed to produce the best team.   Gotcha.  But how do you make that useful?  It would be tough for me to pass on a "stud" early in the draft because I was worried that I would be too weak at another position later.  There are plenty of in-season fixes such as trades, waiver, sleepers, streaming options, etc.

 

[drunken slob]

dot, cause I've moved to Mac

 

[GregR]

I understand you're trying to demonstrate that the best "value" at each pick isn't guaranteed to produce the best team.   Gotcha.  But how do you make that useful?  It would be tough for me to pass on a "stud" early in the draft because I was worried that I would be too weak at another position later.  There are plenty of in-season fixes such as trades, waiver, sleepers, streaming options, etc.

Short version:  I'd use it to identify players who I believe will go later than just VBD value says they should... because your ideal draft in a real scenario probably includes a number of players like that. Then I use mock drafts to understand the options likely to be available to me where my picks fall, and figure out my best possible drafted teams, as well as prepare myself for other contingencies.

Long version:

Agreed, which is why I always do my VBD chart and then pull in the current ADPs and highlight any spots where there seem to be clear discrepancies to find values and to see what really matches up to know when I have to take players.

Agree, which is why recently I've basically tried to use VDB to target roughly the most efficient starting lineup based on positional availability through the first 6-7 rounds and then look to fill out the rest of the roster with higher upside players.

This is pretty much how I did my draft strategies too.

I'd use VBD to come up with value, and then turn that into where players should be drafted if everyone drafted straight from value.  Then I compare it to ADP and highlight players who are being drafted later than they should.  I'd do something similar with my leagues draft tendencies as well.

Now I have identified players that represent an advantage to me if I can get them where I think they are going.  Now I look over the whole list and where my draft picks fall and try to cobble together the best combined team, hitting as many of those players as I can.  I'd normally go ahead and build the team via mock drafts in DD, sometimes taking over other team's drafts to test out different scenarios.  And I'm comparing the total points or total VBD (either works) for each potential team looking for the best one.

I'm also working in other considerations as I do this. Bye week combinations that are good or bad. Injury risk. Upside. Whether I'd want to handcuff someone in particular and how that affects me.

A final thing I'd take the time to consider is what happens if ADP and my beliefs on my league's draft tendencies end up being wrong. I not only try to identify the best team I think I can get, but I check out likely contingencies. From what happens if an unexpected bargain shows up, to what happens if I miss the last player in a tier I was kind of counting on.

As an example of the latter, let's say I'm hoping I'll get my steal RB in the 4th, but it's a drop off after him.  I also go through a mock draft and ask myself what happens if I miss my player, miss out on that RB tier? How does it affect the rest of my draft?  Do I end up taking my 3rd RB earlier than I'd have done otherwise, expecting to platoon my worsened RB2 I'm going to end up with?  If I take my RB3 early how does that affect the rest of my draft?  Or if I'm going to take another position in the 4th if I miss my RB, how is that going to trickle down, and which position should it be.

I would mock things out, so during the draft I'm not scrambling to figure this out when it happens.