What do you use for your baseline? (1 Viewer)

[Crimedogg32]

I was curious what people use for their VBD baseline. The most popular one seems to be worst starter method but I see quite a few flaws. Does anybody see ADP to see like say top 10 rounds and say 30 RBs are taken 10 QBs 25 WRs etc and use that as baseline? Im just curious what you use and how successful it has been for you.

 

[videoguy505]

Trying something new this year, 12 team league, starts 8, using the past 2 years of draft history with the owners to work out first 97 picks (12*8, +1 to account for the snake pick at the end where the owner could have gone either way, added 1 more kicker to the equation). Anyway, it averaged out to 17 QBs, 32 RBs, 29 WRs, 8 TEs, 3 Ks, 8 DSTs. So that's where I'm starting out once DD comes out, and I'll play with that for a few weeks before the draft and see how it shakes out.

 

[Uruk-Hai]

In a way, this is the elephant in the living room of VBD. Baseline keys all of our VBD values, yet there's no concensus on what the "right" baseline is.

I can use "worst starter", "last drafted", a combo of those two, Joe's secret formula (which is similar the the ADP method mentioned in the previous post) - all popular baselines. Yet they can't all be right.

Hines Ward has a specific numerical value in your league - what is it? Change the baseline & his VBD number changes. How can that be right? Ward's value is what it is; it can't be anything else. So why do we have all of these differences of opinion on what the right baseline is?

VBD values are created from seasonal projections. Yet I think many confuse a "VBD sheet" with ADP & so tweak their baselines until their cheatsheet looks similar to a draft (or how they think a draft should go). By using ADP to arrive at baselines, you're creating a self-fulfilling prophecy where eventually your draft sheet will look eerily similar to ADP. That's not right, and it's self-defeating anyway - might as well just use ADP.

"I've got 3 WRs in my first 12 according to VBD. That's not right - no one's gonna take any in round one. Let me slide the baseline up to get them where they're supposed to be" - I think that's the way that the majority of people set their baselines (if they're not using the default).

I'm not stupid, but I don't know what the right answer is. I've started similar threads to this the last couple of years. We've got some serious analytical & football minds around here. Yet I've seen very few who can defend their choice of baseline over another with much more than "it looks more right".

I believe in VBD as theory (though I don't strictly follow it during drafts - partly because of what I've said above), I just don't believe that anyone really knows if they're getting the correct values.

Two questions:

How do you know what a player's true FF value is?

Does the method of determining that value change from league to league, or is it just the value itself that changes (depending on league set-up)?

 

[AB in DC]

Don't use a baseline at all -- draft dynamically.

 

[MrJimiT]

Uruk-Hai brings up some really good points to the many different ways we can use baseline and adjustments. Albeit, that those same baselines are probably not the most correct way to use them. So I preface my baseline uses as something that I use, I also realize the inherant flaws on my systems. When it comes to drafting baselines, I agree with abrecher and draft dynamically or set baselines per "look-ahead" rounds. When I use actual past statistics, such as 2005 stats, I set my baselines based on last/worst starter, which in my 10 team league is 20QB, 20RB, 30WR, 10K, & 10Def. But, when I use projection numbers/stats, I tend to use my leagues historical drafts and try to approximate the number of positions drafted in the first 90 picks (which in my league would be enough picks for each owner to have drafted a starter at every position). So hence, my "value" system has a tendancy to mirror ADP or at the very least, the ADP of historical draft preferences of my league.

 

[Cajun Cannon]

 

[Rounders]

 

[Anon]

Baselines, I think the problems arise from trying to do too much with the VBD number.

Assuming perfect projections, and using worst starter method. Then the calculated VBD number is perfect. It tells me exactly how much player A is worth compared to Player B, based on the worst starting player at those positions.

Or,

Assuming perfect projections, and using last player drafted/rostered. Then the calculated VBD number is again perfect. It tells me exactly how much player A is worth compared to Player B, while compared to the worst player on a roster.

So which one to use ? actually they are both equally accurate, but both are answering different questions. The first (worst starter method) tells us value for starting players. The second (last drafted) tells us value vs. waiver wire players. Assume, everyone is using perfect projections, and half your league drafts “worst starter”, the other half “last rostered”. The teams drafting worst starter will have optimized their starting team, while sacrificing depth. The teams drafting “last rostered” will have optimized depth, while sacrificing some starter potential.

The other baseline methods you will see are using a combination to optimize for some middle ground. It can be a very personal choice, starters vs. depth. So you will see many variations. None of them are wrong as long as you understand why the baseline is set that way and you understand the intent.

Now, the problem or misconception ... VBD does not tell you when or how to draft! It only tells you value of a player. Certainly it is a good guideline but in an actual draft, there are many times when taking the best value is not the best choice for your team.

 

[inca911]

I'm using a two-line VBD fit, which creates different valuations for premium players (i.e., quality starters) and regular players. The intersection of the two line fit and size of the premium player pools are dynamically set based on league tendencies, roster sizes, positional dropoffs, etc. In effect, there are two different baselines which must be established to account for the difference between premium players and regular ones. Premium players are rated based on the premium by-position baselines, and regular players are rated based on the regular by-position baselines. These baselines are estalished based on actual scoring projections, and are not arbitrarily set.

If you think about it, trying to find only one baseline doesn't make sense. 80-85% of auction dollars are spent on the premium players, with only ~15% of auction dollars used to fill in positional depth and fill in non-premium starters. A generic baseline of "worst starter" or "last drafted" does not account for positional depth, or scoring dropoffs specific to scoring systems, so it isn't suprising that neither works well.

For example, if I have an 8 team, start 1QB league with a scoring system where there are 3 QBs scoring 50 points more than the rest of the QB pack, worst starter will incorrectly set a baseline of the 8th QB, and last drafted might set a baseline at the 13th QB (backups, etc). In actuality, nobody is paying a premium after QB #3, so the premium auction value must use a baseline of QB#4, and depending on the player spread it's likely nobody is paying any extra for QBs as soon as they level off the least bit, which means the regular player baseline could be as shallow as QB#7. QB#8 and beyond are minimum bid players.

I've been working on this for some time, and there are two different types of players in every auction which must be accounted for in any system which trys to predict their true value.

If you're interested, here's some history:

http://forums.footballguys.com/forum/index...howtopic=223643

 

[ookook]

I use two baselines and for each player compute two values:

(a) The worst starter at that positon

(b) The median starter at that position

They tell me different things and I monitor the trade-off between them.

Worst starter tells me something about what will happen if I wait to pick that position until pretty much everyone else has.

Median starter tells me what if I wait until half the other owners have selected.

Of course, I take into account more subjectively what I think other teams will do basedon their current rosters, but the two baseline approach helps inform me which position I can most afford to wait on.

 

[Hobbes]

I use worst starter but I only use VBD in order to create tiers of players...after that the number isn't as interesting to me. I am really looking more at the ADP vs. the players in the tiers and trying to find the bargains. This is pretty similar to the dynamic drafting article.

 

[John Mamula]

The people that realize the inherent flaw with VBD are onto something. The values you get from static baselines (i.e. baselines that do not change as the draft develops) are fools gold. The reasons for this include:

1. A baseline such as last starter or other methods use projections that are not the same as other peoples. This means that who you consider to be a last starter and someone else may vary. This may lead you to overpay for a guy who is not as valued by other owners and who you could have grabbed later in the draft.

2. As touched on before, people often use VBD as a type of cheatsheet saying that Player X has the best "value" so I will draft them next. Another easily mis-applied use derived from static baselines. People think they are getting value by a number on a sheet when they have tunnel vision not looking at value in the draft at the same position 5 or 6 draft picks down the road. The two positions that people continue to overpay based on "value" is QB and TE.

3. To tie in the previous two points, VBD does not take into consideration ADP. You cost yourself value in the long run by drafting the top VBD value at the expense of seeing what that player's ADP is and waiting. To further this point, based on your draft slot you should know what bucket of players will be available to you in each round based on ADP. Because of this you no longer need to look at VBD and you can look at the draft as a whole and develop the best team.

4. VBD emphasizes individual value at the expense of team value. The goal of fantasy football is to score more points than your opponent each week. As counter-intuitive as it may seem, a team of the best individual VBD value players will likely not be the team that can score the most points week in and week out compared to drafting via other methods.

5. VBD breaks down after a few rounds as uncertainty about mediocre players increases and as starting positions get filled. Since you need to start tweaking these values on the fly because the value of a 4th RB cannot be easily quantified compared to a 3rd WR or 4th WR, the VBD value you get cannot be relied on.

6. Projections are typically not tied to historic average values (AVT) and as such your baselines do not represent an accurate picture of value to begin with. Your dropoffs will all be buggered up giving you false values.

7. VBD does not take into account PPG (HUGE flaw).

Bottom line, VBD looks great when you back test it. In Joe's example of a two team league that has a one round draft where you already know those players values, the theory works great. You don't pick the guy with the highest gross points, but the one with the greatest point differential. Unfortunately a 12 team draft over 18 rounds when dealing with projections that will have inaccuracies in it, combined with the fact that not everyone values the same players the same (creating inefficiencies in the draft when looking at value to ADP) VBD as a drafting tool completely breaks down.

It is a great theory when applied qualitatively knowing that just because a QB scores a lot of points does not mean he is the best player (even though many VBD proponents picked Manning in the first round last year based on a baseline value). However, knowing the theory and being a slave to its values are two very different things. Static baselines in a draft is riding a bike with training wheels. You may not fall on your face but you are not going to go very fast either.

Take off the training wheels, and draft "dynamically". That is the first step. However there are some of the same issues with DVBD that VBD has, just not as many. However, once you get a handle on that it will open up some other strategies that will allow you to crush people who draft off baselines. Especially ones that use common cheatsheets as you know what the herd is going to do.

I put in literally hundreds of hours of projection preparation over the course of the offseason. But that is where the quantitative analysis ends. Going into a draft there are no more calculations. I am not worried about baselines. I already have a draft strategy based on my draft slot, value I see based on inefficiencies in the draft with ADP higher or lower compared to my projections, and a clear understanding of where value lies throughout the draft at the various positions.

Going to drafts in 2006 you should not need to have a laptop in front of you. There is no need (unless there is no draft board and you want to track players taken automatically). All of the work should have been done before the draft and the draft will be the time to reap the rewards from your hard work.

 

[Anon]

The people that realize the inherent flaw with VBD are onto something. The values you get from static baselines (i.e. baselines that do not change as the draft develops) are fools gold. The reasons for this include:

1. A baseline such as last starter or other methods use projections that are not the same as other peoples. This means that who you consider to be a last starter and someone else may vary. This may lead you to overpay for a guy who is not as valued by other owners and who you could have grabbed later in the draft.

2. As touched on before, people often use VBD as a type of cheatsheet saying that Player X has the best "value" so I will draft them next. Another easily mis-applied use derived from static baselines. People think they are getting value by a number on a sheet when they have tunnel vision not looking at value in the draft at the same position 5 or 6 draft picks down the road. The two positions that people continue to overpay based on "value" is QB and TE.

3. To tie in the previous two points, VBD does not take into consideration ADP. You cost yourself value in the long run by drafting the top VBD value at the expense of seeing what that player's ADP is and waiting. To further this point, based on your draft slot you should know what bucket of players will be available to you in each round based on ADP. Because of this you no longer need to look at VBD and you can look at the draft as a whole and develop the best team.

4. VBD emphasizes individual value at the expense of team value. The goal of fantasy football is to score more points than your opponent each week. As counter-intuitive as it may seem, a team of the best individual VBD value players will likely not be the team that can score the most points week in and week out compared to drafting via other methods.

5. VBD breaks down after a few rounds as uncertainty about mediocre players increases and as starting positions get filled. Since you need to start tweaking these values on the fly because the value of a 4th RB cannot be easily quantified compared to a 3rd WR or 4th WR, the VBD value you get cannot be relied on.

6. Projections are typically not tied to historic average values (AVT) and as such your baselines do not represent an accurate picture of value to begin with. Your dropoffs will all be buggered up giving you false values.

7. VBD does not take into account PPG (HUGE flaw).

Bottom line, VBD looks great when you back test it. In Joe's example of a two team league that has a one round draft where you already know those players values, the theory works great. You don't pick the guy with the highest gross points, but the one with the greatest point differential. Unfortunately a 12 team draft over 18 rounds when dealing with projections that will have inaccuracies in it, combined with the fact that not everyone values the same players the same (creating inefficiencies in the draft when looking at value to ADP) VBD as a drafting tool completely breaks down.

It is a great theory when applied qualitatively knowing that just because a QB scores a lot of points does not mean he is the best player (even though many VBD proponents picked Manning in the first round last year based on a baseline value). However, knowing the theory and being a slave to its values are two very different things. Static baselines in a draft is riding a bike with training wheels. You may not fall on your face but you are not going to go very fast either.

Take off the training wheels, and draft "dynamically". That is the first step. However there are some of the same issues with DVBD that VBD has, just not as many. However, once you get a handle on that it will open up some other strategies that will allow you to crush people who draft off baselines. Especially ones that use common cheatsheets as you know what the herd is going to do.

I put in literally hundreds of hours of projection preparation over the course of the offseason. But that is where the quantitative analysis ends. Going into a draft there are no more calculations. I am not worried about baselines. I already have a draft strategy based on my draft slot, value I see based on inefficiencies in the draft with ADP higher or lower compared to my projections, and a clear understanding of where value lies throughout the draft at the various positions.

Going to drafts in 2006 you should not need to have a laptop in front of you. There is no need (unless there is no draft board and you want to track players taken automatically). All of the work should have been done before the draft and the draft will be the time to reap the rewards from your hard work.

Well said.

 

[joffer]

I think VBD is only useful for about the first 8 rounds or so, and even then static baselines have drawbacks as bagger points out above. Assuming I can find some good ADP data for the type of league I'm drafting in (scoring,roster, etc.), I'll use a dynamic baseline that looks ahead 2 rounds.

After about the 8th round or so, the ADP data is getting more an more inaccurate in terms of how many of each position is being drafted, but it's still vital for where individual players are taken. At that point I'm filling out depth, so I'm taking the highest ranked player on my board at the last pick I think I can get him.

 

[Keysersoze666]

Well written Bagger ... and joffer as well.

Isn't it amazing how things have evolved in the last 5 years of this hobby. I even find myself minimizing the VBD number by the time I get to round 5ish because my plan and the makeup of my potential starters begins to lead me in a direction that doesn't match the raw numbers. I agree with joffer that by round 8, the VBD rating is essentially meaningless. Good topic.

 

[inca911]

Using VBD to establish auction prices for players is a very different story than drafting using VBD. It's important to carefully differentiate between these two.

Calculating auction value is a much more quantitative exercise since it is much more precise. In draft, the difference between #1 and #5 player is 4 draft spots. In auction, that difference can be $30, or 15% of the total dollars available to the team. Calculating an accurate theoretical auction price is very important. This number tells you when a player has value based on your projections in relation to others' projections. If I project a guy to be $30 and bidding stalls at $21, I know how much value I'm looking at. In a draft situation, ADP may tell me to wait and not pick too early. In auction, the value is immediate.

Establishing a baseline based on your projections is critical, since it is defining your vision and strategy.

VBD does not inherently emphasize individual value over team value, it is a just a tool which can be misused if misunderstood. Buying the top 3 QBs may be a good value, but only if someone will trade you!

VBD in its current one-line form does not account for the shift in player types (premium vs. regular) after the premium players have been auctioned. That's part of why the values for players do break down. In a generic scoring 8 team league using last year's projections, 85% of the auction money was spent on roughly the top 50 players. Of course it breaks down in the later rounds if you're trying to draft with it.

VBD doesn't tell people who to draft, but with some changes, it can be used to define an auction value for a player based on your projections. Buying below that auction value is directly related to your team scoring more fantasy points for each dollar spent. That's the goal of the auction.

 

[Z-Dog]

I'm sure I've posted about this at length before, and I once wrote a few article at XpertSports on this issue. I'll try to keep it brief.

If you use a single baseline for VBD, the most important feature that baseline needs to have is that any player who is at the baseline is identical in value. Otherwise, no comparisons relative to the baseline are valid.

Another important factor is to define what you mean by value. That's where things get tricky, because you have to consider starter value vs. backup valus.

Even if you could perfectly account for the curved dropoff in performance, which is a whole problem in and of itself, you still are left with not knowing how to precisely value backup points.

A starting RB will score fantasy point for you in a game. A backup fantasy RB will score points, but if he's sitting on your bench, those points are of limited value. Still, they aren't worth zero, since strong performances from backups raises their trade value. That's a very difficult thing to measure.

Given how difficult it is, as a mathematical propositiong to properly assign value, we tweak the system with a variety of fixes, which attempt to reflect things we know from experience.

Every VBD setup gives you a different value, but that's ok, because each setup answers a somewhat different question. If you use a last-player-chosen baseline, you are measuring what I consider to be the talent index. You're saying that all rules about starters vs. reserves and starting lineups aside, this is how many points players will score. This VBD setup is most useful for those of us who try to build a lineup that will score more than some set point total. In my dynasty league, I have a sense that I need about 1300 points from startes over the course of the season, so this kind of VBD helps me thikn about my team that way.

Alternatively, a VBD set at last starter gives me a better sense of the relative talent drop-off among the presumed fantasy starters. Other baselines, including baselines by starter and reserve, or by position and starter vs. reserve start taming the dropoff curve issue. Other people use multipliers to achieve the same goal of getting a VBD that mimics the dropoff curve. AVT-VBD is often a feature of these.

These last kinds of VBD are projections rather than talent evaluations. In other words, the attempt here is more to get a sense of what actual auction prices might look like. The names you assign to RB1 or WR7 are the less important part, b/c everyone will have different lists. But as you sit in the draft and track the prices, you get a great idea of where values will appear.

 

[Crimedogg32]

If you use a single baseline for VBD, the most important feature that baseline needs to have is that any player who is at the baseline is identical in value. Otherwise, no comparisons relative to the baseline are valid.

When you say this you mean if your QB 16 scores 100 points and RB 20 scores 100 points WR 10 scores 100 points that those players would be the baseline?Thanks for all help so far guys all the seem great on this topic

 

[Verbal Kint]

The people that realize the inherent flaw with VBD are onto something. The values you get from static baselines (i.e. baselines that do not change as the draft develops) are fools gold. The reasons for this include:

1. A baseline such as last starter or other methods use projections that are not the same as other peoples. This means that who you consider to be a last starter and someone else may vary. This may lead you to overpay for a guy who is not as valued by other owners and who you could have grabbed later in the draft.

2. As touched on before, people often use VBD as a type of cheatsheet saying that Player X has the best "value" so I will draft them next. Another easily mis-applied use derived from static baselines. People think they are getting value by a number on a sheet when they have tunnel vision not looking at value in the draft at the same position 5 or 6 draft picks down the road. The two positions that people continue to overpay based on "value" is QB and TE.

3. To tie in the previous two points, VBD does not take into consideration ADP. You cost yourself value in the long run by drafting the top VBD value at the expense of seeing what that player's ADP is and waiting. To further this point, based on your draft slot you should know what bucket of players will be available to you in each round based on ADP. Because of this you no longer need to look at VBD and you can look at the draft as a whole and develop the best team.

4. VBD emphasizes individual value at the expense of team value. The goal of fantasy football is to score more points than your opponent each week. As counter-intuitive as it may seem, a team of the best individual VBD value players will likely not be the team that can score the most points week in and week out compared to drafting via other methods.

5. VBD breaks down after a few rounds as uncertainty about mediocre players increases and as starting positions get filled. Since you need to start tweaking these values on the fly because the value of a 4th RB cannot be easily quantified compared to a 3rd WR or 4th WR, the VBD value you get cannot be relied on.

6. Projections are typically not tied to historic average values (AVT) and as such your baselines do not represent an accurate picture of value to begin with. Your dropoffs will all be buggered up giving you false values.

7. VBD does not take into account PPG (HUGE flaw).

Bottom line, VBD looks great when you back test it. In Joe's example of a two team league that has a one round draft where you already know those players values, the theory works great. You don't pick the guy with the highest gross points, but the one with the greatest point differential. Unfortunately a 12 team draft over 18 rounds when dealing with projections that will have inaccuracies in it, combined with the fact that not everyone values the same players the same (creating inefficiencies in the draft when looking at value to ADP) VBD as a drafting tool completely breaks down.

It is a great theory when applied qualitatively knowing that just because a QB scores a lot of points does not mean he is the best player (even though many VBD proponents picked Manning in the first round last year based on a baseline value). However, knowing the theory and being a slave to its values are two very different things. Static baselines in a draft is riding a bike with training wheels. You may not fall on your face but you are not going to go very fast either.

Take off the training wheels, and draft "dynamically". That is the first step. However there are some of the same issues with DVBD that VBD has, just not as many. However, once you get a handle on that it will open up some other strategies that will allow you to crush people who draft off baselines. Especially ones that use common cheatsheets as you know what the herd is going to do.

I put in literally hundreds of hours of projection preparation over the course of the offseason. But that is where the quantitative analysis ends. Going into a draft there are no more calculations. I am not worried about baselines. I already have a draft strategy based on my draft slot, value I see based on inefficiencies in the draft with ADP higher or lower compared to my projections, and a clear understanding of where value lies throughout the draft at the various positions.

Going to drafts in 2006 you should not need to have a laptop in front of you. There is no need (unless there is no draft board and you want to track players taken automatically). All of the work should have been done before the draft and the draft will be the time to reap the rewards from your hard work.

Well said.

Not much more to add except that the 2 great unknowns that affect draft success are 1. accuracy of projections, adjusting for risk

2. confidence level in player ADPs specific to your individual league

If you can manage down those uncertainties any VBD you use will serve you fine.

 

[videoguy505]

If you use a single baseline for VBD, the most important feature that baseline needs to have is that any player who is at the baseline is identical in value. Otherwise, no comparisons relative to the baseline are valid.

When you say this you mean if your QB 16 scores 100 points and RB 20 scores 100 points WR 10 scores 100 points that those players would be the baseline?Thanks for all help so far guys all the seem great on this topic

I don't think that's what that's supposed to mean... value is a little more loosely defined than "X Points". Consider a league with simplified scoring... say 10 rec yds=1 pt., no TD points, nothing else. So your WRs only got credit for yardage. Then imagine in this 10 team league, there was only 1 WR slot per roster. Depending on your projections, WR1-WR5 will all catch for 1290-1320 yards. WR6-10 all catch for 1200-1250 yards. There is little 'value' in WR1 compared to WR5. Depending on scoring for other positions, WR1 might not be that much of a steal compared to WR10--12 points over the course of the season might not matter as much. Therefore, compared to the other positons, the baseline for WRs would be changed to reflect the fact that it doesn't matter which of the top 10 you pick.

 

[Crimedogg32]

Figured id bring this post up. I know have data for previous drafts with pretty much the same people for my $$$ league and I like the idea of using an average of how many RBs went in X number of rounds etc for the baseline. Which round is the best to use for this?

 

[Thoriniii]

Figured id bring this post up. I know have data for previous drafts with pretty much the same people for my $$$ league and I like the idea of using an average of how many RBs went in X number of rounds etc for the baseline. Which round is the best to use for this?

I would enter the data for the entire draft and get more of a "Worst Starter" baseline. I would then look at each round, to see how many of each position are taken to get an idea of which position I need to target at each point in the draft. Many people on the Football Guys Forum will say that VBD breaks down after the 5th or 6th round. If you subscribe to this philosophy, you may want to set your "Worst Starter" for the 5th or 6th round, although you are not technically looking at the Worst Starter in this case. I will usually set my VBD baseline and run a few mock drafts to see if the setting have the right feel for my draft.

 

[ravnzfan]

If you use a single baseline for VBD, the most important feature that baseline needs to have is that any player who is at the baseline is identical in value. Otherwise, no comparisons relative to the baseline are valid.

When you say this you mean if your QB 16 scores 100 points and RB 20 scores 100 points WR 10 scores 100 points that those players would be the baseline?Thanks for all help so far guys all the seem great on this topic

my read on this is in a start 1QB-2RB-3WR-1TE league, baseline players are:QB12

RB24

WR36

TE12

all=0

going up, you have a numerical valuation for each player relative to those from other positions

by tiering, I can "see" where the positions are glutted (say QB7-14), so that if I decide RB25 (therefore below badeline) is more valuable to me because of bye week issues, injury history to starters, etc, I can make the determination to pass on QB7 because of the probability a similar player will be available alittle later in the draft

 

[GregR_2]

One problem with approaching this issue as it was presented (e.g. "What do you use for your baseline") is that the question kind of builds into it meaning that VBD doesn't have.

The high level answer that anyone using it properly as a tool should give is, "I use a baseline appropriate for what I want to see at the moment." If you want to look at value amongst starters you use something along the lines of last starter or average starter. If you want to look at value amongst backups you use last drafted, or last-most-likely-to-see-my-lineup (and you probably should weight for how many games the player may be expected to fill in... if it's a QB it is probably 1-2 weeks on average... if a RB in a start 2 league it is probably 5-6 on average).

There is no right or wrong answer or best answer because it just shows you value of a group of players, it doesn't tell you who to pick. You need to set the baseline to be appropriate for what kind of value comparison you want to see at the moment.

 

[jpwest]

Don't use a baseline at all -- draft dynamically.

Where is part 2?

 

[AB in DC]

Don't use a baseline at all -- draft dynamically.

Where is part 2?

That one never got published, unfortunately. I'll see if I can find a copy and post it in this thread.

 

[Z-Dog]

Has anyone ever tried a multilinear VBD?

Let me explain. One of the big problems with VBD, aside from identifiying the 'correct' baseline is that it doesn't take into account the dropoff from RB1 to RB2, or to RB4 - it just compares RB1 to the baseline, and then RB2 to the baseline, and so forth. The fact that QB1 is better than the baseline by 200 points is only one aspect of the QB1's value - and it's an inflationary one. Last starter baselines often overestimate the true value of QBs b/c of exactly this issue. VBD will tell you that QB1 is great to have, but since the supply of QBs who will come very close to matching QB1 is high, the actual market value of QB1 is much lower.

There must be an appropriate mathematical calculation that can compare RB1 to every other RB as well as the baseline, but it's a little beyond me what the correct calculations are. Has anyone here done anything similar?

 

[GregR_2]

Has anyone ever tried a multilinear VBD?

Let me explain. One of the big problems with VBD, aside from identifiying the 'correct' baseline is that it doesn't take into account the dropoff from RB1 to RB2, or to RB4 - it just compares RB1 to the baseline, and then RB2 to the baseline, and so forth. The fact that QB1 is better than the baseline by 200 points is only one aspect of the QB1's value - and it's an inflationary one. Last starter baselines often overestimate the true value of QBs b/c of exactly this issue. VBD will tell you that QB1 is great to have, but since the supply of QBs who will come very close to matching QB1 is high, the actual market value of QB1 is much lower.

There must be an appropriate mathematical calculation that can compare RB1 to every other RB as well as the baseline, but it's a little beyond me what the correct calculations are. Has anyone here done anything similar?

There are really infinite number of ways you can do it. In its simplest or most common form that would be an average starter baseline though. Basically you're saying, I don't just care how much QB1 is ahead of QB12, I want to know how much ahead of everyone at his position is.So one solution is to subtract everyone's points from him. If you don't want the math, skip down to where it says HERE. If you do, it would be:

QB1-QB2 + QB1-QB3 + QB1-QB4 ... + QB1-QB12.

Which can be expressed as: 11*QB1 - (sum of all QBs other than himself).

Which to simplify if we add in another QB1 on both sides = 12*QB1 - (sum of all QBs including himself).

Before we can compare to other positions though we need to deal with the differing number of starters. WR would be really high just because there are more terms (instead of comparisons to 11 QBs giving points there are comparisons to 35 WRs). So we should divide by the number of players (i.e. 12 for QB). So what we end up with is:

HERE:

QB1 - (Sum of all QBs / 12)

That latter term is the average of all QB starters. So we can skip all the mess, find the average starter and subtract it from every starting player. This means half of your players will have negative values, but that is fine. It's the difference between them that matters, not the magnitude of the number. And that number will reflect how the player compares to every other player at his position. A guy at the end of a position who is top-heavy with talent will have a lower value than a guy at the end of a position that is evenly distributed or bottom-heavy.

You can also probably do better than that. For example, there can be so little difference between some positions, like QB8 to QB15, that the depth at the position beyond the last starter makes those last few players have even less value since a guy beyond the "last starter" can do nearly as well. So you may want to come up with a weighting scheme to reflect that.

Though frankly, I don't know that it's worth the effort vs just being cognizant of such a fact and including it mentally in your decision making process.

 

[Winston Wolfe]

Has anyone ever tried a multilinear VBD?

Let me explain. One of the big problems with VBD, aside from identifiying the 'correct' baseline is that it doesn't take into account the dropoff from RB1 to RB2, or to RB4 - it just compares RB1 to the baseline, and then RB2 to the baseline, and so forth. The fact that QB1 is better than the baseline by 200 points is only one aspect of the QB1's value - and it's an inflationary one. Last starter baselines often overestimate the true value of QBs b/c of exactly this issue. VBD will tell you that QB1 is great to have, but since the supply of QBs who will come very close to matching QB1 is high, the actual market value of QB1 is much lower.

There must be an appropriate mathematical calculation that can compare RB1 to every other RB as well as the baseline, but it's a little beyond me what the correct calculations are. Has anyone here done anything similar?

In looking at the value above the base line, you should also be comparing the value against the other value. This is what most people do when determining if it is the right time to take a player. The are looking to see if their is a positional glut. VBD might not specifically spell it out, but a person should be doing it.

 

[Z-Dog]

The answer that I sohuld be doing it b/c VBD doesn't is an acceptable answer to most questions about VBD, but I was asking if there was a way to make VBD do something which it wasn't already doing!

 

[GregR_2]

Has anyone ever tried a multilinear VBD?

Let me explain. One of the big problems with VBD, aside from identifiying the 'correct' baseline is that it doesn't take into account the dropoff from RB1 to RB2, or to RB4 - it just compares RB1 to the baseline, and then RB2 to the baseline, and so forth. The fact that QB1 is better than the baseline by 200 points is only one aspect of the QB1's value - and it's an inflationary one. Last starter baselines often overestimate the true value of QBs b/c of exactly this issue. VBD will tell you that QB1 is great to have, but since the supply of QBs who will come very close to matching QB1 is high, the actual market value of QB1 is much lower.

There must be an appropriate mathematical calculation that can compare RB1 to every other RB as well as the baseline, but it's a little beyond me what the correct calculations are. Has anyone here done anything similar?

In looking at the value above the base line, you should also be comparing the value against the other value. This is what most people do when determining if it is the right time to take a player. The are looking to see if their is a positional glut. VBD might not specifically spell it out, but a person should be doing it.

True that. I mentioned that is the thought behind the average player baseline, and what it is going to show you.And while that is useful to get a feel for how value varies between positions there, it still doesn't tell you what decision to make when you draft. It is input, but when it comes down to it you care most about the difference between the best QB available now, the best QB available at your next pick, the best QB available at your pick after that, etc.

A few years back I ran through some mock drafts with having a team use best player available, drafting straight off a VBD cheat sheet, drafting based solely on biggest expected drop between this pick and next, and one where they drafted the absolute optimal team (which to do you have to know who is available in advance, and then pretty much use a computer to solve it because it isn't a trivial problem).

The absolute optimum team was of course the best, with the biggest expected drop (also called dynamic VBD by many) next best and VBD cheatsheet next best. Though it's worth noting the difference between the latter 2 in particular wasn't much, and neither was all that many FP behind the optimal team. The differences were well down below the uncertainty you would have in predicting what players would be available at your next picks, let alone the uncertainty in your projections.

This is one reason I feel the techniques out there have pretty much already reached the practical end point of the usefulness of quantitative analysis of the draft. I think the more important thing to focus on is what decisions you face in each round and knowing in advance how those decisions will impact your later decisions. The current set of tools, including tiering, VBD and dynamic VBD, already do a good enough job of showing where value lies. There is a bigger gain to be had from focusing more on the depletion of the players than on trying to nail down the value from 95% to 99%.

 

[John Paul Jones]

 

[ookook]

When I said I compute 2 baselines (median starter and worst starter) and which set of values I emphasize depends on the where the draft is I oversimplified a bit.

I really compute 4 from the onset. Median starter, worst starter, median backup, worst backup.

This takes into account much of the "non-linear" function of drop offs that the earlier poster worried about. I have loooked at doing more (e.g., quartile baselines within starters) but there really was not much payoff.

For the first 4 or 5 I look most at median starter values, for rounds 5 to 10 at worst starter, then switch to backup values. Now, really I am also looking at the next lower set of values too, but giving less weight. Sometimes I even do a weighted average.

It all goes into answering "what if I wait and do not take the highest ranking player from that position...what are the consequences".

To sum, I totally agree with GregR's post:

"The high level answer that anyone using it properly as a tool should give is, "I use a baseline appropriate for what I want to see at the moment." If you want to look at value amongst starters you use something along the lines of last starter or average starter. If you want to look at value amongst backups you use last drafted, or last-most-likely-to-see-my-lineup (and you probably should weight for how many games the player may be expected to fill in... if it's a QB it is probably 1-2 weeks on average... if a RB in a start 2 league it is probably 5-6 on average).

There is no right or wrong answer or best answer because it just shows you value of a group of players, it doesn't tell you who to pick. You need to set the baseline to be appropriate for what kind of value comparison you want to see at the moment."

 

[Kleck.]

Not much to add after bagger's and joffer's posts. Those two really get it.

 

[Z-Dog]

As long as you're using VBD for drafting, I understand that you probably don't need much more than to run different projections with a couple of different baselines. Things change when you're using VBD for auctions, and its worth working for additional precision.

 

[inca911]

When using VBD for drafting, one has the luxury of not caring so much about the true value of a player, but only caring that the player in question is the most valuable on the board. Auctions are quite different. In an auction, the value of the player must be as precisely determined as possible. My thoughts are specific to calculating accurate auction values.

To determine an accurate auction value, we must first understand and account for the bimodal distribution caused by premium starter players vs. regular players and then account for that bimodal distribution in our valuation system. In most auctions, most of the auction dollars are spent on the premium starting players (~85%). The remaining 15% are spent to fill out rosters with more regular players who do not have significant drops in value compared to the rest of the league. The valuation system must account for this bimodal distribution. A two line fit should take care of this issue.

Randomly selecting worst starter or average starter as a baseline ignores the actual breakpoints between the players. Using the actual projected scoring for the specific league, one must individually determine where the premium players stop and the regular players begin. This tells us the location of the intersection of our two line fit for the various positions, and most importantly it puts players into the two different pools of auction dollars--premium vs. regular. In many leagues, there may be only one premium kicker and all the rest are regular, or only 3 premium QBs. The key is to use emperical data to set these breakpoints, not random worst starter estimates. To set these breakpoints, one can look at the scoring distributions and see where the difference flattens out.

I've got a spreadsheet which I've used to do a custom baselinnig of premium vs. regular by position with a two line fit. This then allows me to set the % of total salary cap for premium players and calculate accurate auction values. If anyone is interested, there is a post on improving auction values in the applications forum with more details of the current issue with VBD and examples of how to fix it.

 

[JerseyPaul]

VBD...forgetaboutit.....well, almost. Bagger has it right and here's what I do:

Prerequisites:

1. Projections you can live with. I prefer to use the FBG's Excel because I can add my ADP info easily. Tweak the projections anyway you like and you're ready to go.

2. ADP info you can live with. Note, your final draft prep should use very currrent ADP data as this changes fast. Sources: Antsports, Xperts, FBGs, lots of others.

Now what I do is use excel to grab ADP data and I concatenate the ADP to the player's name in the speadsheet. Now look at the spreadsheet. The inefficiencies will pop right out. You'll see something like:

Able 3.07 175

Baker 3.09 173

Charlie 3.05 173

Dog 5.09 171

Edward 4.03 163

Fox 4.05 162

George 5.11 152

Forget the VBD numbers, you want that Dog on your team. You think this doesn't happen? Wrong, it does.

Now do "mental mocks" using the ADP and projected FPs. For your skill players, do a mock and try different sequences for those difficult issues. Do RB then RB, WR; do RB,RB,WR, etc. Add up the projected FPs for each of the best available players based on ADP and see what combos work best.

Have a backup in mind. If your ADP data gives you a target of Reggie Wayne (ADP 3.7) in the middle of round 3 and he's gone, no problem, you'll take DJax (ADP 3.9).

Sometimes there's a cliff and if a guy is taken, you have a big problem. In the example above, if you wait until round 5 to take Dog, you have no close alternates available if somebody picked him up before you. Think it through, will you take a lesser player at the position or try to get value at another position.

Lastly, obviously if a higher projected player is still available that you did not expect, you smile and take him.

All this takes time but the upsides are great. Most importantly, the stress level at the draft is way down for you. You have a plan. You know what to expect. Surprises will improve your team but you've optimized for the most probable scenarios. You can relax while others stare at their cheetsheets and wonder what to do next.

 

[m_sell31]

i'm not new to fantasy football, but after looking at this thread I am obviously not taking it as seriously as you guys are! You guys seem to have a great conversation going and I don't want to slow it down too much, but could some take a second and explain to me whatsome of the terminology you are using means? Like baseline, worst starter, etc? What is this strategy and what is a baseline and how do i establish one? I know i sould like a ####### but if you could alteast point me towards some good reading that will start me from scratch I would really appreciate it. i tied to just jump in and figure it out from your posts but i'm not getting anywhere. thanks!

 

[AB in DC]

Here's my unpublished Part 2 from 2004.

Honestly, though, bagger and JerseyPaul said it better than I can. Those two posts should be cut out and framed.

---

Drafting Dynamically Part II:

Fine-tuning your "Perfect Draft"

Managing Risk

One risk you take in basing your strategy around a particular value pick is that the player may be taken earlier than you expect. Before settling on this strategy, consider what you could do if that happens.

If you have more than one player that you target as a value pick at that position, around that stage of the draft, you should be in good shape. Chances are that you won't lose out on all of them. But if not, you have three alternatives.

1. Plan on taking your value pick even earlier. If you expect big things from Josh McCown, whose ADP is around the 10th round, you may want to take him as early as the 8th. You may want to do a little math to make sure that McCown is still a value pick if you take him that early.

2. Make a backup plan. Find a few players who you'd be willing to take a couple rounds later, ideally ones who also represent value relative to their ADP.

3. Plan on taking a lesser value earlier in the draft. If waiting for that one player is simply too risky for you, forget about him and plan on filling that position earlier, knowing that you can fall back on option #1 if an unexpected opportunity prevents you from filling that position early.

Choosing the best ADP calculation

A variety of sites calculate an estimate of ADP. If your league is relatively normal in terms of scoring systems, you may want to average the ADPs at a number of sites, such as XpertLeagues and AntSports.

First, though, consider the competition in your league. If your league is filled with inexperienced players, you probably don't want to use ADP from a site like XpertLeagues. Instead, use ESPN, since those drafts are more likely to have data from other inexperienced players. If your league is filled with sharks, ignore ESPN altogether.

I would also recommend avoiding MyFantasyLeague.Com's ADP estimate. Many established leagues have migrated to MFL and use the first few rounds of the draft to pick their pre-determined keepers. This tends to distort MFL's estimates too much to be useful.

Setting up your draft board.

To allow your strategy to be the most flexible during the draft, you need to set up your draft board with enough information to make decisions on the fly.

I recommend using a chart where roster positions are listed across the top with round numbers down the side. In each box, you list players who you expect to be available in that round but no later, along with their ADPs and projected fantasy points. For example, if in a serperntine draft you have pick number 4 (plus 21, 28, 45, etc.), your board may look like this:

QB RB WR TENever none 1 Holmes 319 None None 2 Tomlinson 3151 (#4) none 4 McAllister 294 None 5 Portls 276 4 Green 268 6 Alexander 255 7 James 251 11 Barlow 236 11 Moss 220 9 Lewis 232 10 Taylor 217 13 dillon 201 15 d.davis 189 2 (#21) 22 Culpepper 355 19 S.Davis 204 20 Harrison 217 19 henry 188 19 Holt 207 19 bennett 188 19 faulk 179 24 barber 165 25 owens 178 3 (#28) 26 Manning 330 28 K.JONES 200 28 C.Johnson 184 27 Ward 174 40 McNabb 303 32 MARTIN 196 33 Mason 165 42 Gonzo 124 35 vick 302 33 Westbrook 172 40 s.moss 161 39 Staley 167 38 boldin 152 36 Brown 148 37 horn 147 36 Garner 140A few notes:- Players in [ALL CAPS] are the value picks.

- Player in [all lower case] are ranked lower than a player expect to be available in later rounds. You should avoid them in most circumstances. (Exceptions include cases when the higher ranked player has been taken earlier than expected, or when you expect to take both players)

- This draft board separate players who may be available in each round (ADP very close to your draft spot) and those who are likely to be available (ADP significantly lower). As you look ahead on your draft board, be aware of the risk of relying on players being available very close to their ADP.

By setting up your draft board by ADP, you can see more easily the possibilities you face when you reach a fork in the road. If, for example, both Jones and Martin are taken earlier than you expect, you can look ahead to later rounds to see what caliber of RB will be available in the future. That way, you can reach an informed decision on whether to pull the trigger on a lesser RB now or to stock up on other positions until you reach a point where RBs have value again.

Remember, planning ahead is essential to having a "perfect draft". Your draft board must allow you to look at later rounds and have a sense for what is available then, not just who is available at each round.

 

[ravnzfan]

i'm not new to fantasy football, but after looking at this thread I am obviously not taking it as seriously as you guys are! You guys seem to have a great conversation going and I don't want to slow it down too much, but could some take a second and explain to me whatsome of the terminology you are using means? Like baseline, worst starter, etc? What is this strategy and what is a baseline and how do i establish one? I know i sould like a ####### but if you could alteast point me towards some good reading that will start me from scratch I would really appreciate it. i tied to just jump in and figure it out from your posts but i'm not getting anywhere. thanks!

We'll let Joe and David walk you thru the basics HERE...basic VBD as it was put into use way back in the mid-90's obviously, others have tweaked the system to take into account several factors as they unveil in a draft...I use it to keep the tiers straight w/in a position, so that I can determine if I can wait another round on a player in a particular bucket, or if the dropoff @ the position is too steep...for example

say I sit in the middle of a draft, ~5/6 slot...assume I've drafted Tiki, Chad Johnson, Kevin Jones and Roy Williams, in that order.

VBD assigns a value to each player, based on projections, so that we can compare players from different positions---assume the VBD #'s led me to those selections, which are pretty close to what might happen in a live 12 team draft.

in a start 1QB-2RB-3WR-1TE league, the dilema faced @5.5 is:

do I take my starting QB?

or starting TE?

or a starting WR?

or my b/u RB, because who knows if Tiki plays all year (wrong side of 30) or if Jones tanks?

the VBD value tells you the highest ranked player, as compared to the others in different positions...I also can "see" what the alternatives are if I pass on the highest valued player, by comparing the players left. By counting the spots before my next pick, I can determine if the falloff w/in a position is too great...for example, assume Gates, Shockey and Gonzo are gone in our draft

the VBD numbers might tell you a QB is the highest valued position player left--also a position of need for our squad, so in it's purest form, VBD says take the QB---by examining the tier, however, I "see" 4 other QB's w/in several points of the player, whereas the dropoff from Heap to Crumpler is too steep for me, especially if Crump also gets plucked before the draft gets back to me---I'd probably take Heap in this situation, as the combo of Heap and a QB drafted later will be higher than the QB and whatever might be left @TE in the next round

the decisions become alot clearer when implementing this strategy---we used to call "RB-RB no matter what" something cute(the name escapes me now--I'm getting too old!) , but the reason that strategy was so strong was that the #'s told us those players were the highest rated based on VBD---the reason PPR leagues came about was to level the playing field by increasing the value of WR/TE's to RB's by adding 1pt/reception

a league like this really emphasizes the importance of determining the value of the players to each other , as the draft isn't so cut and dry any more---you don't automatically see 20+ RB's drafted in the first 2 rounds, lending a totally different feel to it as compared to a "basic scoring" draft

read up...and feel free ask away w/any more questions---PM me if you don't want to ask in this thread

 

[BroncoFreak_2K3]

Here's my unpublished Part 2 from 2004.

Honestly, though, bagger and JerseyPaul said it better than I can.  Those two posts should be cut out and framed.

---

Drafting Dynamically Part II:

Fine-tuning your "Perfect Draft"

Managing Risk

One risk you take in basing your strategy around a particular value pick is that the player may be taken earlier than you expect.  Before settling on this strategy, consider what you could do if that happens.

If you have more than one player that you target as a value pick at that position, around that stage of the draft, you should be in good shape.  Chances are that you won't lose out on all of them.  But if not, you have three alternatives.

1. Plan on taking your value pick even earlier.  If you expect big things from Josh McCown, whose ADP is around the 10th round, you may want to take him as early as the 8th.  You may want to do a little math to make sure that McCown is still a value pick if you take him that early.

2. Make a backup plan.  Find a few players who you'd be willing to take a couple rounds later, ideally ones who also represent value relative to their ADP.

3. Plan on taking a lesser value earlier in the draft.  If waiting for that one player is simply too risky for you, forget about him and plan on filling that position earlier, knowing that you can fall back on option #1 if an unexpected opportunity prevents you from filling that position early.

Choosing the best ADP calculation

A variety of sites calculate an estimate of ADP.  If your league is relatively normal in terms of scoring systems, you may want to average the ADPs at a number of sites, such as XpertLeagues and AntSports.

First, though, consider the competition in your league.  If your league is filled with inexperienced players, you probably don't want to use ADP from a site like XpertLeagues.  Instead, use ESPN, since those drafts are more likely to have data from other inexperienced players.  If your league is filled with sharks, ignore ESPN altogether.

I would also recommend avoiding MyFantasyLeague.Com's ADP estimate.  Many established leagues have migrated to MFL and use the first few rounds of the draft to pick their pre-determined keepers.  This tends to distort MFL's estimates too much to be useful.

Setting up your draft board.

To allow your strategy to be the most flexible during the draft, you need to set up your draft board with enough information to make decisions on the fly.

I recommend using a chart where roster positions are listed across the top with round numbers down the side.  In each box, you list players who you expect to be available in that round but no later, along with their ADPs and projected fantasy points.  For example, if in a serperntine draft you have pick number 4 (plus 21, 28, 45, etc.), your board may look like this:

      QB                  RB                WR               TENever  none                1 Holmes 319      None             None                           2 Tomlinson 3151 (#4) none                4 McAllister 294                   None                           5 Portls 276                           4 Green 268                           6 Alexander 255                           7 James 251                          11 Barlow 236      11 Moss 220                           9 Lewis 232                          10 Taylor 217                          13 dillon 201                          15 d.davis 189 2 (#21) 22 Culpepper 355  19 S.Davis 204     20 Harrison 217                          19 henry 188       19 Holt 207                          19 bennett 188                          19 faulk 179                          24 barber 165      25 owens 178 3 (#28) 26 Manning 330    28 K.JONES 200     28 C.Johnson 184                                               27 Ward 174        40 McNabb 303     32 MARTIN 196      33 Mason 165     42 Gonzo 124        35 vick 302       33 Westbrook 172   40 s.moss 161                          39 Staley 167      38 boldin 152                          36 Brown 148       37 horn 147                          36 Garner 140      A few notes:- Players in [ALL CAPS] are the value picks.

- Player in [all lower case] are ranked lower than a player expect to be available in later rounds.  You should avoid them in most circumstances.  (Exceptions include cases when the higher ranked player has been taken earlier than expected, or when you expect to take both players)

- This draft board separate players who may be available in each round (ADP very close to your draft spot) and those who are likely to be available (ADP significantly lower).  As you look ahead on your draft board, be aware of the risk of relying on players being available very close to their ADP.

By setting up your draft board by ADP, you can see more easily the possibilities you face when you reach a fork in the road.  If, for example, both Jones and Martin are taken earlier than you expect, you can look ahead to later rounds to see what caliber of RB will be available in the future.  That way, you can reach an informed decision on whether to pull the trigger on a lesser RB now or to stock up on other positions until you reach a point where RBs have value again.

Remember, planning ahead is essential to having a "perfect draft".  Your draft board must allow you to look at later rounds and have a sense for what is available then, not just who is available at each round.

Is this something you could use in game-planning using the DD? Seems like it would help sort things out better on the fly, once you've mocked a few times with DD. Good info though.

 

[m_sell31]

how does Draft Domminator work, and what is its value? I downloaded it but I havent really figured out how to use it?

 

[Shark Attack]

Don't use a baseline at all -- draft dynamically.

excellent piece and really what this hobby is all about drafting wise. I read this when it first was published and it has been how I look at all important drafts I am in. I have drafted this way for quite some time now but this read just put it all into words that make sense..lol but so true for me. Great stuff for anyone that can take drafting that far to use those stradegys.

 

[Shark Attack]

The people that realize the inherent flaw with VBD are onto something. The values you get from static baselines (i.e. baselines that do not change as the draft develops) are fools gold. The reasons for this include:

1. A baseline such as last starter or other methods use projections that are not the same as other peoples. This means that who you consider to be a last starter and someone else may vary. This may lead you to overpay for a guy who is not as valued by other owners and who you could have grabbed later in the draft.

2. As touched on before, people often use VBD as a type of cheatsheet saying that Player X has the best "value" so I will draft them next. Another easily mis-applied use derived from static baselines. People think they are getting value by a number on a sheet when they have tunnel vision not looking at value in the draft at the same position 5 or 6 draft picks down the road. The two positions that people continue to overpay based on "value" is QB and TE.

3. To tie in the previous two points, VBD does not take into consideration ADP. You cost yourself value in the long run by drafting the top VBD value at the expense of seeing what that player's ADP is and waiting. To further this point, based on your draft slot you should know what bucket of players will be available to you in each round based on ADP. Because of this you no longer need to look at VBD and you can look at the draft as a whole and develop the best team.

4. VBD emphasizes individual value at the expense of team value. The goal of fantasy football is to score more points than your opponent each week. As counter-intuitive as it may seem, a team of the best individual VBD value players will likely not be the team that can score the most points week in and week out compared to drafting via other methods.

5. VBD breaks down after a few rounds as uncertainty about mediocre players increases and as starting positions get filled. Since you need to start tweaking these values on the fly because the value of a 4th RB cannot be easily quantified compared to a 3rd WR or 4th WR, the VBD value you get cannot be relied on.

6. Projections are typically not tied to historic average values (AVT) and as such your baselines do not represent an accurate picture of value to begin with. Your dropoffs will all be buggered up giving you false values.

7. VBD does not take into account PPG (HUGE flaw).

Bottom line, VBD looks great when you back test it. In Joe's example of a two team league that has a one round draft where you already know those players values, the theory works great. You don't pick the guy with the highest gross points, but the one with the greatest point differential. Unfortunately a 12 team draft over 18 rounds when dealing with projections that will have inaccuracies in it, combined with the fact that not everyone values the same players the same (creating inefficiencies in the draft when looking at value to ADP) VBD as a drafting tool completely breaks down.

It is a great theory when applied qualitatively knowing that just because a QB scores a lot of points does not mean he is the best player (even though many VBD proponents picked Manning in the first round last year based on a baseline value). However, knowing the theory and being a slave to its values are two very different things. Static baselines in a draft is riding a bike with training wheels. You may not fall on your face but you are not going to go very fast either.

Take off the training wheels, and draft "dynamically". That is the first step. However there are some of the same issues with DVBD that VBD has, just not as many. However, once you get a handle on that it will open up some other strategies that will allow you to crush people who draft off baselines. Especially ones that use common cheatsheets as you know what the herd is going to do.

I put in literally hundreds of hours of projection preparation over the course of the offseason. But that is where the quantitative analysis ends. Going into a draft there are no more calculations. I am not worried about baselines. I already have a draft strategy based on my draft slot, value I see based on inefficiencies in the draft with ADP higher or lower compared to my projections, and a clear understanding of where value lies throughout the draft at the various positions.

Going to drafts in 2006 you should not need to have a laptop in front of you. There is no need (unless there is no draft board and you want to track players taken automatically). All of the work should have been done before the draft and the draft will be the time to reap the rewards from your hard work.

Absolutly excellent and as far as I am concerned right on the money. This is one of the best threads i have ever read here. Just teriffic for anyone trying to really get deep into our hobby here. I do a lot of the same work bagger does and I do somtimes feel i can walk into my drafts come September with a pen and one piece of paper to right down my team on and still come out with the best team. I am that fimiliar with the players and any draft slot I may have.

Anyone looking to really get better at this hobby needs to take in and understand this thread. But it does come down to doing hundreds and hundreds of hours of what I call my homework to to be in the positon I want to be in heading into any key draft. I know almost exactly where every player will or should be taken in a draft. That kind of knowledge can be very dangerous in any FF draft as long as its used correctly..

 

[GregR_2]

i'm not new to fantasy football, but after looking at this thread I am obviously not taking it as seriously as you guys are! You guys seem to have a great conversation going and I don't want to slow it down too much, but could some take a second and explain to me whatsome of the terminology you are using means? Like baseline, worst starter, etc? What is this strategy and what is a baseline and how do i establish one? I know i sould like a ####### but if you could alteast point me towards some good reading that will start me from scratch I would really appreciate it. i tied to just jump in and figure it out from your posts but i'm not getting anywhere. thanks!

ravnzfan posted a link to the article on it that will explain the basics. Let me add a few things, really just restating what we have said here, but hopefully geared for someone new to it.* VBD is a great tool for seeing where value lies in a position. It doesn't tell you the best place where a player should be taken, however... which is a mistake many people do by drafting from the combined rankings it can produce.

* There are several reasons for that last statement. At the highest level, you could say VBD shows you the value of a player based on the supply of players in the league and the part of the demand based on the starting lineup requirements of your league. However, it does not account for the part of value that comes from how the positions are going to be depleted during the draft. That means, if everyone is taking WRs between now and your next pick, even if a RB is your best VBD value, he isn't the best value to your team at this pick since you can get him at your next pick. You should instead (probably) take a WR. Incorporating that aspect of value is what people mean when they talk about "dynamic VBD" or using ADP (average draft position).

* The baseline you set will have a lot to do with what you see. Choose one appropriate for what you are looking for.

* You will frequently hear things like, "I don't agree with VBD. Stud RB is better." If you hear that, it means the person doesn't understand VBD. The analogy I like to use is that is like saying, "I don't agree with Newton's laws of gravitation. Things falling down is better." They aren't mutually exclusive. Newton's laws explain why things fall down. Similarly, VBD is the way to see numerically WHY Stud RB works in a normal league. It will help tell you how much of an advantage it brings. And it also will help tell you for other league setups when it is no longer as advantageous.

 

[AB in DC]

By setting up your draft board by ADP, you can see more easily the possibilities you face when you reach a fork in the road. If, for example, both Jones and Martin are taken earlier than you expect, you can look ahead to later rounds to see what caliber of RB will be available in the future. That way, you can reach an informed decision on whether to pull the trigger on a lesser RB now or to stock up on other positions until you reach a point where RBs have value again.

Is this something you could use in game-planning using the DD? Seems like it would help sort things out better on the fly, once you've mocked a few times with DD. Good info though.

I haven't worked with DD very much, so I don't know if it's compatible with what I wrote. I'll give it a shot sometime this week, but anyone who wants to try this in the meantime, please post here and let us know how it worked. Thanks.