2016 Auction Thread (1 Viewer)

[abbottjamesr]

For those trying to land as many studs as possible and then fill out the remainder of the squad with $1 guys.  How are you attacking the auction in terms of nominating players?

Are you trying to land $1 guys really early on?  Or are you trying to lock up your studs first.  

I almost always wait to nominate the $1 guys I want as long as possible.  Often these guys have been built up by the hype machine that is the football media now days.  If you throw them out early some one will bid $2 cause why not its only a $1 right.  

I almost exclusively nominate QB's through the draft.  Its the position I care least about and I want them to be bid up, I am happy to take a nobody for $1.  I love to nominate the #1 QB early as there are always 2-3 guys that want a stud QB in my drafts.  I also like to nominate the top 1 or 2 defenses as they always seem to be get bid up.  

Much is going to depend on how the draft goes though.  It always seems there is some one with too much money at the end that overpays for all the hyped up flyers at the end.  

Worst thing to do is to nominate them early and pay $3 for them.  They are not $1 guys then, and this is especially bad in in $100 auctions.  

 

[matuski]

Sure, but the point of calculating auction prices is to figure out what you should be spending, not what other people are going to spend. 

Disagree.

 

[matuski]

For those trying to land as many studs as possible and then fill out the remainder of the squad with $1 guys.  How are you attacking the auction in terms of nominating players?

Are you trying to land $1 guys really early on?  Or are you trying to lock up your studs first.  

Nominate the positions you are waiting on or have filled.

 

[Hawkeye21]

For those trying to land as many studs as possible and then fill out the remainder of the squad with $1 guys.  How are you attacking the auction in terms of nominating players?

Are you trying to land $1 guys really early on?  Or are you trying to lock up your studs first.  

I like to toss QBs that I don't want out there.  I will also nominate higher players that I don't want like CJ Anderson and Jamaal Charles.  I don't like throwing out my $1 guys too early because someone can easily grab him for $2 and then the value is gone.

 

[Slapdash]

For those trying to land as many studs as possible and then fill out the remainder of the squad with $1 guys.  How are you attacking the auction in terms of nominating players?

Are you trying to land $1 guys really early on?  Or are you trying to lock up your studs first.  

Most nominations in my league approximate ADP.  I grab a couple of studs early...engage over time for QB/TE/K/TD when it makes sense.  But it is a lot of waiting around until the very end where I have my list of sleepers to pick through.

 

[CalBear]

Disagree.

How can you use the information that your league is likely to spend $20 on a player, if you don't know how much the player is worth?

 

[dkp993]

I think the new DD is useless for auctions. You'd be better off with an ESPN AAV cheatsheet.

I don't agree.  I do way more work on my auctions then I need to because I enjoy it, so I've created my own excel sheets to create values, tracked my leagues drafts and created a rolling 3 yr average actual draft position spend and use the DD.    I've always worked to get the DD (classic previously) to sync up with the data I've created so I could use the DD on draft day, and have done this with very good results.  Now while its taken a while and a fair amount of tweaking I've been able to get the new DD app to line up pretty damn close to what my data say.   Enough so that I'm going to use it exclusively at my main league draft the Sunday. 

 

[Hawkeye21]

How can you use the information that your league is likely to spend $20 on a player, if you don't know how much the player is worth?

I feel the way your league drafts every year dictates how much a player is worth.  My league never has spent more than $35 on a player.  When I do my ranking I project what their value is based on how my league drafts.  I believe I have Brown with the highest value of all players with $32.  He went for $30.  If I would use your theory of valuing him at closer to $40 it would not have done me any good at all.  In fact it would have messed up all of my projections.

 

[dkp993]

How can you use the information that your league is likely to spend $20 on a player, if you don't know how much the player is worth?

Agreed.  In fact it's critical to have both pieces of info.  Having only one or the other dramitacly short changes your ability to prep properly.  

 

[CalBear]

I feel the way your league drafts every year dictates how much a player is worth.  My league never has spent more than $35 on a player.  When I do my ranking I project what their value is based on how my league drafts.  I believe I have Brown with the highest value of all players with $32.  He went for $30.  If I would use your theory of valuing him at closer to $40 it would not have done me any good at all.  In fact it would have messed up all of my projections.

No. How many points the player will score dictates what he is worth. The way your league drafts dictates which players can be purchased at value.

 

[Hawkeye21]

No. How many points the player will score dictates what he is worth. The way your league drafts dictates which players can be purchased at value.

I guess it just doesn't matter for me in my league.  I have my system for determining what players are going to go for which is very accurate and it really helps me prepare for my draft. 

 

[matuski]

If your league overpays for the top 20 at each position.. their value is what it takes to get them.

If you don't pay it you are left with a lot of money and nowhere to spend it when you are done cherry picking what's left.  But you nailed it on value, right?

.

 

[matuski]

I guess it just doesn't matter for me in my league.  I have my system for determining what players are going to go for which is very accurate and it really helps me prepare for my draft. 

If your league is that predictable, good on you, sure makes it easier.

Ours sways widely year to year on what people are chasing/overpaying for.  Some years it is a position, others it is studs regardless of position, some years everyone is cautious..... etc.

 

[Ramblin Wreck]

I look at my league's auction historically over the last 2-3 years and come up with average prices for RB1, RB2, RB3.... and do the same for other positions.  Then I take ADP to rank those players with average values.  Then I scan that list to determine which players I want to target and avoid.  And I know generally what to spend to get them.

So if Eric Decker is WR25 on ADP and WR25 in my auction averages $11, then I know what he should approxiamately be going for.  But if he's WR15 on my personal rankings then that makes him a target for me around $11.  WR15 usually goes for $21 so I may be willing to spend up to $15-$18 to get Decker since I like him a lot if that's what it takes.  But that all depends on when he gets put up for bid too, how many other options I have at that point if I need a WR, and other factors.  But it's absolutely essential to enter the auction knowing how much your opponents typically spend on players.

 

[Hawkeye21]

If your league is that predictable, good on you, sure makes it easier.

Ours sways widely year to year on what people are chasing/overpaying for.  Some years it is a position, others it is studs regardless of position, some years everyone is cautious..... etc.

All you need is data from enough drafts and you can have a very good idea what players are going to go for.

 

[dkp993]

I look at my league's auction historically over the last 2-3 years and come up with average prices for RB1, RB2, RB3.... and do the same for other positions.  Then I take ADP to rank those players with average values.  Then I scan that list to determine which players I want to target and avoid.  And I know generally what to spend to get them.

So if Eric Decker is WR25 on ADP and WR25 in my auction averages $11, then I know what he should approxiamately be going for.  But if he's WR15 on my personal rankings then that makes him a target for me around $11.  WR15 usually goes for $21 so I may be willing to spend up to $15-$18 to get Decker since I like him a lot if that's what it takes.  But that all depends on when he gets put up for bid too, how many other options I have at that point if I need a WR, and other factors.  But it's absolutely essential to enter the auction knowing how much your opponents typically spend on players.

Exactly.   

 

[matuski]

All you need is data from enough drafts and you can have a very good idea what players are going to go for.

I have 5 years of auction data.. I can average it out, but the variation is crazy.  There is no consistency.

For example - last 3 years top player went for $91, $62, and $47. RB, WR, TE.  

The overall spending follows this wild swing as well.. from having to blow your wad to get studs, to overpaying for just one position, to everyone hoarding.  You can have your projected values.. and I do, but you have to be prepared to throw them in the trash 10 minutes in.

 

[Hawkeye21]

I have 5 years of auction data.. I can average it out, but the variation is crazy.  There is no consistency.

For example - last 3 years top player went for $91, $62, and $47. RB, WR, TE.  

The overall spending follows this wild swing as well.. from having to blow your wad to get studs, to overpaying for just one position, to everyone hoarding.  You can have your projected values.. and I do, but you have to be prepared to throw them in the trash 10 minutes in.

Wait.. So one year the most expensive player was $91 and then another year it was $47?  That's a bit ridiculous.

 

[CalBear]

If your league overpays for the top 20 at each position.. their value is what it takes to get them.

If you don't pay it you are left with a lot of money and nowhere to spend it when you are done cherry picking what's left.  But you nailed it on value, right?

I'm pretty sure that fantasy football games are decided by how many points players score, not by how much they cost at the auction.

Value can be measured in dollars per points above the baseline. It's not possible for the league to overspend dollars for points above the baseline; they only have as many dollars as they have. 

 

[abbottjamesr]

Value can be measured in dollars per points above the baseline. It's not possible for the league to overspend dollars for points above the baseline; they only have as many dollars as they have. 

This is exactly right.  Lets do a thought experiment to show it.  Lets say the other 11 guys all spend 85% of their budgets on a top 11 player.  Would you rather spend 86% of your budget for a top 5 guy or be able to get 7 of the lower half of the top 20.  

 

[CalBear]

One of the ways in which people make economically inefficient decisions is through anchoring. Anchoring is a cognitive bias that occurs when people have a expected number in their head. They may realize their number is too high or too low, but when they make a mental adjustment, they don't adjust enough. Ask someone if there are more than 30 countries in Africa, and ask someone else if there are fewer than 75 countries in Africa. Then ask them how many they think there are. The first person is likely to think the number is lower than the second person.

How does this play out in auctions? League tendencies. If League 1 thinks that $30 is a lot of money, players who are worth more than $30 are likely to be undervalued. If League 2 thinks the top players are worth $50, top players are likely to be overvalued.

Your job as an auction drafter is to exploit these tendencies, not to follow them. In League 1, you should be buying players worth more than $30, which you can probably get for about $30. In League 2, you should be waiting to find good players once people are strapped for cash.

You're not a market researcher. You're a stockbroker.

Warren Buffett said that in the short run the market is a voting machine, but in the long run it's a weighing machine. A fantasy football auction is the same way; in the short run, prices are dictated by what people are willing to pay, but in the long run, value is defined by production.

 

[Slapdash]

I wish FBGs would hire some of you

 

[TripItUp]

One of the ways in which people make economically inefficient decisions is through anchoring. Anchoring is a cognitive bias that occurs when people have a expected number in their head. They may realize their number is too high or too low, but when they make a mental adjustment, they don't adjust enough. Ask someone if there are more than 30 countries in Africa, and ask someone else if there are fewer than 75 countries in Africa. Then ask them how many they think there are. The first person is likely to think the number is lower than the second person.

How does this play out in auctions? League tendencies. If League 1 thinks that $30 is a lot of money, players who are worth more than $30 are likely to be undervalued. If League 2 thinks the top players are worth $50, top players are likely to be overvalued.

Your job as an auction drafter is to exploit these tendencies, not to follow them. In League 1, you should be buying players worth more than $30, which you can probably get for about $30. In League 2, you should be waiting to find good players once people are strapped for cash.

You're not a market researcher. You're a stockbroker.

Warren Buffett said that in the short run the market is a voting machine, but in the long run it's a weighing machine. A fantasy football auction is the same way; in the short run, prices are dictated by what people are willing to pay, but in the long run, value is defined by production.

terrific post, maybe post of the season candidate...nice job

 

[dkp993]

One of the ways in which people make economically inefficient decisions is through anchoring. Anchoring is a cognitive bias that occurs when people have a expected number in their head. They may realize their number is too high or too low, but when they make a mental adjustment, they don't adjust enough. Ask someone if there are more than 30 countries in Africa, and ask someone else if there are fewer than 75 countries in Africa. Then ask them how many they think there are. The first person is likely to think the number is lower than the second person.

How does this play out in auctions? League tendencies. If League 1 thinks that $30 is a lot of money, players who are worth more than $30 are likely to be undervalued. If League 2 thinks the top players are worth $50, top players are likely to be overvalued.

Your job as an auction drafter is to exploit these tendencies, not to follow them. In League 1, you should be buying players worth more than $30, which you can probably get for about $30. In League 2, you should be waiting to find good players once people are strapped for cash.

You're not a market researcher. You're a stockbroker.

Warren Buffett said that in the short run the market is a voting machine, but in the long run it's a weighing machine. A fantasy football auction is the same way; in the short run, prices are dictated by what people are willing to pay, but in the long run, value is defined by production.

Very well said.   

If no one in your league is willing to pay more then 30 but your valuation of say AB is 42 but you too are unwilling to go above 30 (because that's what the league does) the person that reaches 30 first is getting great value and you've lost an opportunity.  

The market may dictate the price but your job is to exploite that and find advantages where you can

 

[IHEARTFF]

Yeah CalBear with a great post.

 

[IHEARTFF]

Yeah CalBear with a great post.

 

[Hawkeye21]

Maybe what I need to ask then is how do I determine what the actual auction value is for the players in my league.  The one spreadsheet that I have that has a formula to determine values had Brown into the $40 range which seemed a bit ridiculous but maybe it was right.  I was more-so trying to determine what players were going to go for in my draft when I did my spreadsheet.

 

[dkp993]

Maybe what I need to ask then is how do I determine what the actual auction value is for the players in my league.  The one spreadsheet that I have that has a formula to determine values had Brown into the $40 range which seemed a bit ridiculous but maybe it was right.  I was more-so trying to determine what players were going to go for in my draft when I did my spreadsheet.

Knowing what players are likely to go for in your draft is great and an important part of the process.  But it's just one part.  I'm not sure what process you used to value AB as 40 but if you trust that process then in turn you need to trust that value.   Now you can exploite the league.  So what if everyone else thinks you overpaid, you shouldn't be concerned with what everyone else thinks is the value if you trust your prep and process.   

 

[Hawkeye21]

Knowing what players are likely to go for in your draft is great and an important part of the process.  But it's just one part.  I'm not sure what process you used to value AB as 40 but if you trust that process then in turn you need to trust that value.   Now you can exploite the league.  So what if everyone else thinks you overpaid, you shouldn't be concerned with what everyone else thinks is the value if you trust your prep and process.   

Could that have a negative affect on the rest of my team when drafting though?  If I get AB and Julio because I know they are worth more wouldn't that hurt my chances more at getting the mid tier player due to lack of funds?  I suppose it doesn't really matter regardless though, does it.

 

[Ramblin Wreck]

Yeah I would be curious in how Cal, or anyone else, determines what a player value should be.  My process of doing an auction I described earlier allows me to know what I think each player will go for based on historical results and ADP so I can target players accordingly and know how I'm going to build my team.  It's worked so far but curious how to take that one step further to get real value on a player based on projections.

 

[Hawkeye21]

Yeah I would be curious in how Cal, or anyone else, determines what a player value should be.  My process of doing an auction I described earlier allows me to know what I think each player will go for based on historical results and ADP so I can target players accordingly and know how I'm going to build my team.  It's worked so far but curious how to take that one step further to get real value on a player based on projections.

I do the same as you and I just don't see how going any further is going to help me any more.  I know that AB and Julio were worth more but I just couldn't bring myself to spend that much at the time.  I could have spent $31 on AB and still been $1 under what I had him valued at.

I didn't just project the player's value at what I thought they were going to get taken for but for what I thought they should go for based on how our league drafts.  If that makes any sense at all.

 

[dkp993]

Could that have a negative affect on the rest of my team when drafting though?  If I get AB and Julio because I know they are worth more wouldn't that hurt my chances more at getting the mid tier player due to lack of funds?  I suppose it doesn't really matter regardless though, does it.

That's exactly what you want.  2 elite players then value ( which there always is) mid draft.  Win win.  

 

[matuski]

Wait.. So one year the most expensive player was $91 and then another year it was $47?  That's a bit ridiculous.

Exactly.

 

[Hawkeye21]

Exactly.

Do you have the same people drafting every year?  That seems absolutely crazy.

 

[matuski]

I'm pretty sure that fantasy football games are decided by how many points players score, not by how much they cost at the auction.

Value can be measured in dollars per points above the baseline. It's not possible for the league to overspend dollars for points above the baseline; they only have as many dollars as they have. 

A league can overspend versus your baseline in segments.. if the league "overspends" on all players with ADPs in the first 3 rounds, it will "underspend" on the rest.

In the end, you are correct - the dollars are the dollars.  But if you didn't overspend you didn't get any of the top players and you are left bragging about the "values" you got on your roster chock full of 4th and 5th rounders..

 

[Mr. Irrelevant]

A league can overspend versus your baseline in segments.. if the league "overspends" on all players with ADPs in the first 3 rounds, it will "underspend" on the rest.

In the end, you are correct - the dollars are the dollars.  But if you didn't overspend you didn't get any of the top players and you are left bragging about the "values" you got on your roster chock full of 4th and 5th rounders..

CalBear's follow-up response was a fantastic post, but your above reply to his first post is spot-on as well. They're not mutually exclusive.

Here's the thing: The VBD difference between the #1 and #10 options at each position is far greater than that between the #11 and #20 options at the position. In fact, it's usually greater than the difference between the #11 option and a bench-level replacement. Think about it: in a sport beset by randomness ... with 15, 17, sometimes 20 rounds of drafts ... not to mention starting lineup requirements, waiver wires, etc., the #1 draft slot is still, statistically significantly, far more likely to win his/her league than the guy drafting at #12. That's how much of a built-in advantage the top of the draft board offers.

Auctions attempt to correct for this, but people's pre-set perceptions generally don't shift far enough towards the elite guys in terms of value. If the 10th-rated RB sells for $35, the #1 guy should probably go for $70. But he usually doesn't - because people will opt for the "second-round talent" at $35 over the #1 guy at $60, even though on a dollars per VBD basis the #1 guy is still a better value in that situation.

I've run and participated in a huge number of auctions, and it's vanishingly rare that owners' perception of top-end value adjusts far enough to where the studs are no longer offer value. But it does happen. And in those rare cases, yeah, to some extent, you have to go with the flow and try to get yours. It's almost impossible to win a typical fantasy league with the equivalent of eight 4th-round draft picks and no studs to speak of in your starting lineup ... no matter how much theoretical value you've captured in the process.

 

[CalBear]

Yeah I would be curious in how Cal, or anyone else, determines what a player value should be.  My process of doing an auction I described earlier allows me to know what I think each player will go for based on historical results and ADP so I can target players accordingly and know how I'm going to build my team.  It's worked so far but curious how to take that one step further to get real value on a player based on projections.

I don't think this is a solved problem; earlier in this thread I talked about some ideas for a statistical model which would account for subtleties in rankings. Right now we boil everything down to a single number (projected fantasy points) and then assign values based on that number, and that process has a number of issues. But given that caveat, here's how I do it.

The goal of the auction is to come out with the greatest return on your investment, in terms of fantasy points in your starting lineup per dollar. Leaving aside the inherent uncertainties in that calculation, the way we capture the potential is by the VBD number, which represents the number of fantasy points the player is expected to score relative to a baseline level of production.

For drafts, usually we just choose a baseline like "worst starter" (the number of points the #12QB, #24RB, #36WR are expected to score), which has a number of issues, but probably works fine for drafts because they're really not very complicated. Auctions have way more going on. Among other things, at least half of the players bought at auction will not be starters, and most of those will still cost more than $1, some significantly so. So some amount of your budget has to be allocated to players whose production may not be in the starting lineup most of the time. So the baseline has to be lower for auctions than for drafts.

I'm personally using a point-based baseline at each position, which for my league's scoring is 260 points for QBs, 75 for WRs, and 50 for RBs. So a QB projected to score 270 points, a WR projected to score 85, and a RB projected to score 60 would all have VBD scores of 10. A Rb projected to score 100 would have 50 VBD, and so on.

With FBG projections, in my scoring this adds up to 6958 VBD points for the entire league. There are $1200 available, and you'd rather not spend any on players with negative VBD numbers, so a VBD point is worth about ($1200/6958)=$0.17, all else being equal. In a simple linear model, someone projected with 50 VBD is worth $8.50, and someone with 100 VBD is worth $17.

But, a simple linear model doesn't account for the fact that 100 VBD is more than twice as good as 50 VBD. A player producing 100 VBD only takes one slot in your lineup, is more likely to produce at a high level each week, and is more likely to stay in your lineup all year. ~100 VBD this year is Odell Beckham; and Beckham is definitely worth more than the equivalent amount of money spent on Randall Cobb and Brandin Cooks. Tomlinson in his prime, who was probably over 200 VBD, was probably worth over 50% of your auction budget. So, the linear model needs to be curved to give more weight to elite players. The difference between 90 and 100 is bigger than the difference between 0 and 10.

There's probably a calculus solution to this, but I just tweaked my fudge factor until the curve looked reasonable. I'm using:

1.0135^VBD-1

This creates another entity I call VBDC, which is basically a curved VBD. 100 VBD=2.823 VBDC; 50 VBD=0.963. 0 VBD=0. The total for the league is 152. So, $1200/152 = $7.89/VBDC point. So the 100 VBD player is worth ~$22, and the 50 VBD player ~$7.50. The change in the curve moved money from the lower-VBD players to the higher-VBD players, which seems like the right thing to do.

This gets recalculated dynamically as the auction goes on. The value of a VBDC point is always ( $$ remaining / VBDC remaining ).

1.0135 is an arbitrary number, which is why I think it would be valuable to build a real statistical model. I started with a higher number but tweaked it downwards based on eyeballing the player curve. What I'm looking for in that process is something of a fit to league tendencies, but also at how well the curve describes scoring tiers. QB value falls off the cliff after Eli Manning (actually, looking at it right now I may change Eli's projections downwards), with Palmer, Wilson, Bortles, and Rivers all at 68-75 VBD points. A tier like that should all be valued similarly, and players in the tier above should be significantly more expensive. Bortles (73 VBD) is $13 and Brees (102 VBD) is $23, which sounds about right. Or you could grab Joe Flacco (26 VBD) for $3.

 

[Ramblin Wreck]

Advice I would give without even considering dollar amounts would be.  Normal lineups with no strange scoring stuff.

1.  Get two of the top 15 or so studs wherever your dropoff point is.  Gurley, Peterson, L. Bell, David Johnson, Freeman, Miller, Charles, Elliott at RB and Brown, Julio, Beckham, Hopkins, Green, Robinson, and Dez at WR.  

2.  Go cheap at QB.  QB9 through QB 22 will have very little different average points per game.  QB9 is going to cost a lot more than QB15/QB22.  Get two QB in that $2-$4 range amongst Winston, Cousins, Eli, Rivers, Carr, Stafford, Tyrod, Dalton, Romo, Ryan, Mariota, Fitzpatrick. A. Smith, and Tannehill.  Ideally, I think I would end up with one of the vets here and one of the younger upside guys.

3.  Go cheap at TE.  Gronk or Reed would be nice and if you decide to pay for them I wouldn't criticize.  But after those two, ignore Olsen and Kelce and then find the value amongst Walker, Fleener, Ertz, Thomas, Gates, Barnidge, Witten (PPR), and Eifert.  If you play it right you can get two of these guys and play matchups or see if one becomes an easy start decision.  Or get one and grab Allen, Graham, Bennett, Ebron, Green (if healthy) as a backup or take a flyer on Seferian Jenkins, Miller, Clay, Virgil Green, or Walford.  You can get a 2nd TE for $1.

4.  There will be insane WR value.  You can get two stud RB and still be able to get three or more WR from Landry, Cobb, Decker, Maclin, Tate, Edelman, Kelvin, Fitzgerald, Baldwin, Moncrief tier.  Even after those guys Floyd, Lockett, Parker, Hurns, Crabtree, Sanders and others will go for $4-$6.  Those are very good back end starters if you're pairing with 2-3 studs.

5.  You probably want one of the first tier RB even though it will cost you.  I'm not a fan of going after the next tier of guys like Martin, Lacy, Rawls, and Hyde.  I would much prefer an elite guy paired with Gio Bernard, Matt Jones, Frank Gore, Jeremy Hill or some cheaper RB.

6.  Never pay more $1 for a kicker.

7.  I never pay more than $1 for a defense.  I usually get two of them.  One is a top 10-12 defense that I think can be solid and consistent.  The other is one with an easy schedule to start the year since I'll probably end up cutting them anyway.

 

[abbottjamesr]

I don't think this is a solved problem; earlier in this thread I talked about some ideas for a statistical model which would account for subtleties in rankings. Right now we boil everything down to a single number (projected fantasy points) and then assign values based on that number, and that process has a number of issues. But given that caveat, here's how I do it.

The goal of the auction is to come out with the greatest return on your investment, in terms of fantasy points in your starting lineup per dollar. Leaving aside the inherent uncertainties in that calculation, the way we capture the potential is by the VBD number, which represents the number of fantasy points the player is expected to score relative to a baseline level of production.

For drafts, usually we just choose a baseline like "worst starter" (the number of points the #12QB, #24RB, #36WR are expected to score), which has a number of issues, but probably works fine for drafts because they're really not very complicated. Auctions have way more going on. Among other things, at least half of the players bought at auction will not be starters, and most of those will still cost more than $1, some significantly so. So some amount of your budget has to be allocated to players whose production may not be in the starting lineup most of the time. So the baseline has to be lower for auctions than for drafts.

I'm personally using a point-based baseline at each position, which for my league's scoring is 260 points for QBs, 75 for WRs, and 50 for RBs. So a QB projected to score 270 points, a WR projected to score 85, and a RB projected to score 60 would all have VBD scores of 10. A Rb projected to score 100 would have 50 VBD, and so on.

With FBG projections, in my scoring this adds up to 6958 VBD points for the entire league. There are $1200 available, and you'd rather not spend any on players with negative VBD numbers, so a VBD point is worth about ($1200/6958)=$0.17, all else being equal. In a simple linear model, someone projected with 50 VBD is worth $8.50, and someone with 100 VBD is worth $17.

But, a simple linear model doesn't account for the fact that 100 VBD is more than twice as good as 50 VBD. A player producing 100 VBD only takes one slot in your lineup, is more likely to produce at a high level each week, and is more likely to stay in your lineup all year. ~100 VBD this year is Odell Beckham; and Beckham is definitely worth more than the equivalent amount of money spent on Randall Cobb and Brandin Cooks. Tomlinson in his prime, who was probably over 200 VBD, was probably worth over 50% of your auction budget. So, the linear model needs to be curved to give more weight to elite players. The difference between 90 and 100 is bigger than the difference between 0 and 10.

There's probably a calculus solution to this, but I just tweaked my fudge factor until the curve looked reasonable. I'm using:

1.0135^VBD-1

This creates another entity I call VBDC, which is basically a curved VBD. 100 VBD=2.823 VBDC; 50 VBD=0.963. 0 VBD=0. The total for the league is 152. So, $1200/152 = $7.89/VBDC point. So the 100 VBD player is worth ~$22, and the 50 VBD player ~$7.50. The change in the curve moved money from the lower-VBD players to the higher-VBD players, which seems like the right thing to do.

This gets recalculated dynamically as the auction goes on. The value of a VBDC point is always ( $$ remaining / VBDC remaining ).

1.0135 is an arbitrary number, which is why I think it would be valuable to build a real statistical model. I started with a higher number but tweaked it downwards based on eyeballing the player curve. What I'm looking for in that process is something of a fit to league tendencies, but also at how well the curve describes scoring tiers. QB value falls off the cliff after Eli Manning (actually, looking at it right now I may change Eli's projections downwards), with Palmer, Wilson, Bortles, and Rivers all at 68-75 VBD points. A tier like that should all be valued similarly, and players in the tier above should be significantly more expensive. Bortles (73 VBD) is $13 and Brees (102 VBD) is $23, which sounds about right. Or you could grab Joe Flacco (26 VBD) for $3.

Have you tried working backwards from your values to see if your system is correct?  You know you are valuing players correctly when no matter which guys you select in the budget you get the same total of VBD.  In a fairly valued system there should be no arbitrage.

I also believe every ones values will be different even if we all used the same projections.  How you value depth on your team changes how you do your baselines.  If you only value starters as worth more than $1 your values will be different from the guy who values one depth spot or even two depth spots.  

 

[Hawkeye21]

Advice I would give without even considering dollar amounts would be.  Normal lineups with no strange scoring stuff.

1.  Get two of the top 15 or so studs wherever your dropoff point is.  Gurley, Peterson, L. Bell, David Johnson, Freeman, Miller, Charles, Elliott at RB and Brown, Julio, Beckham, Hopkins, Green, Robinson, and Dez at WR.  

2.  Go cheap at QB.  QB9 through QB 22 will have very little different average points per game.  QB9 is going to cost a lot more than QB15/QB22.  Get two QB in that $2-$4 range amongst Winston, Cousins, Eli, Rivers, Carr, Stafford, Tyrod, Dalton, Romo, Ryan, Mariota, Fitzpatrick. A. Smith, and Tannehill.  Ideally, I think I would end up with one of the vets here and one of the younger upside guys.

3.  Go cheap at TE.  Gronk or Reed would be nice and if you decide to pay for them I wouldn't criticize.  But after those two, ignore Olsen and Kelce and then find the value amongst Walker, Fleener, Ertz, Thomas, Gates, Barnidge, Witten (PPR), and Eifert.  If you play it right you can get two of these guys and play matchups or see if one becomes an easy start decision.  Or get one and grab Allen, Graham, Bennett, Ebron, Green (if healthy) as a backup or take a flyer on Seferian Jenkins, Miller, Clay, Virgil Green, or Walford.  You can get a 2nd TE for $1.

4.  There will be insane WR value.  You can get two stud RB and still be able to get three or more WR from Landry, Cobb, Decker, Maclin, Tate, Edelman, Kelvin, Fitzgerald, Baldwin, Moncrief tier.  Even after those guys Floyd, Lockett, Parker, Hurns, Crabtree, Sanders and others will go for $4-$6.  Those are very good back end starters if you're pairing with 2-3 studs.

5.  You probably want one of the first tier RB even though it will cost you.  I'm not a fan of going after the next tier of guys like Martin, Lacy, Rawls, and Hyde.  I would much prefer an elite guy paired with Gio Bernard, Matt Jones, Frank Gore, Jeremy Hill or some cheaper RB.

6.  Never pay more $1 for a kicker.

7.  I never pay more than $1 for a defense.  I usually get two of them.  One is a top 10-12 defense that I think can be solid and consistent.  The other is one with an easy schedule to start the year since I'll probably end up cutting them anyway.

I agree with most of what you said but I need to ask what type of a bench do you have.  I hope it's a big bench because according to your strategy you are carrying a second QB, TE and defense.  To me that's an incredible waste of roster spots unless you have plenty of room.

 

[abbottjamesr]

I agree with most of what you said but I need to ask what type of a bench do you have.  I hope it's a big bench because according to your strategy you are carrying a second QB, TE and defense.  To me that's an incredible waste of roster spots unless you have plenty of room.

I'll second this.  I like all of the advice except I will use the WW for my second QB, TE and will stream Defenses.

 

[Ramblin Wreck]

I agree with most of what you said but I need to ask what type of a bench do you have.  I hope it's a big bench because according to your strategy you are carrying a second QB, TE and defense.  To me that's an incredible waste of roster spots unless you have plenty of room.

Start 9, 7 bench, limited waiver moves for season. Have to plan in this league. Can't flip kickers and defense whenever you want. 

 

[Hawkeye21]

Start 9, 7 bench, limited waiver moves for season. Have to plan in this league. Can't flip kickers and defense whenever you want. 

That's a big difference then.  I have no problems with that.

 

[CalBear]

Have you tried working backwards from your values to see if your system is correct?  You know you are valuing players correctly when no matter which guys you select in the budget you get the same total of VBD.  In a fairly valued system there should be no arbitrage.

I also believe every ones values will be different even if we all used the same projections.  How you value depth on your team changes how you do your baselines.  If you only value starters as worth more than $1 your values will be different from the guy who values one depth spot or even two depth spots.  

I haven't gone back and verified the results, but it's a good idea.

In thinking about it, there's a piece that's not being captured, which is the roster size. My system would output the same costs for selecting 18 players (as we do) or 10 (as Hawkeye does). Now, you can partially address that by adjusting the baselines, but there's also at least a small effect that your cheapest players aren't $0, they're $1, so drafting in an 18-slot league is different than in a 22-slot league. There's less money available in leagues with larger rosters. Maybe I should just subtract out the total number of rosters spots from the money available, and then add $1 in to the calculated prices. [tries that out...] No, I don't really like it, because that is including $1 as the value for a lot of zero-value (well below baseline) players. I mean, someone could theoretically spend $1 on Brian Hoyer but I don't want to take that potential dollar away from the top players. 

More TBD.

Baselines are crucial in these valuations. So is the fudge factor, which is probably what's getting adjusted when you mess with the "Studs and Duds" slider in the new DD. The lack of real control of these factors is why the new DD is useless for auctions.

 

[Hawkeye21]

Now we are really getting into some good information for auctions.  This is the stuff I really like seeing discussed.

Is there anyway to add an attachment of a spreadsheet on here?

 

[abbottjamesr]

I haven't gone back and verified the results, but it's a good idea.

In thinking about it, there's a piece that's not being captured, which is the roster size. My system would output the same costs for selecting 18 players (as we do) or 10 (as Hawkeye does). Now, you can partially address that by adjusting the baselines, but there's also at least a small effect that your cheapest players aren't $0, they're $1, so drafting in an 18-slot league is different than in a 22-slot league. There's less money available in leagues with larger rosters. Maybe I should just subtract out the total number of rosters spots from the money available, and then add $1 in to the calculated prices. [tries that out...] No, I don't really like it, because that is including $1 as the value for a lot of zero-value (well below baseline) players. I mean, someone could theoretically spend $1 on Brian Hoyer but I don't want to take that potential dollar away from the top players. 

More TBD.

Baselines are crucial in these valuations. So is the fudge factor, which is probably what's getting adjusted when you mess with the "Studs and Duds" slider in the new DD. The lack of real control of these factors is why the new DD is useless for auctions.

Roster size is captured in figuring auction values.  Your total money available to spread across the total VBD, If you have 18 roster slots and 12 teams and $200 budget is:

12 teams X $200/Team = $2400

$2400 - (18 roster spots X 12 teams X $1) = $2184

Only the top 216 players are worth a minimum of $1 every one else is worth $0.  So your calculation still works of adding the dollar back in just it only applies to the top 216 players.

I adjust my baselines to shift my values more toward studs and duds rather the fudge factor that you use.  In theory studs and duds is just saying I don't value depth as much so I only want to include the top 24 RB's and WR's in my calculations. 

 

[CalBear]

Roster size is captured in figuring auction values.  Your total money available to spread across the total VBD, If you have 18 roster slots and 12 teams and $200 budget is:

12 teams X $200/Team = $2400

$2400 - (18 roster spots X 12 teams X $1) = $2184

Only the top 216 players are worth a minimum of $1 every one else is worth $0.  So your calculation still works of adding the dollar back in just it only applies to the top 216 players.

I adjust my baselines to shift my values more toward studs and duds rather the fudge factor that you use.  In theory studs and duds is just saying I don't value depth as much so I only want to include the top 24 RB's and WR's in my calculations. 

The problem is, the 216th player isn't worth $1, he just costs $1. Assigning him (and N players above him with value < $1) a value of $1 will take money away from more valuable players.

I don't think adjusting baselines is sufficient to get good auction values, for two reasons. One is that many bench players have non-zero value, some of those significant. Your top bench WR and RB are probably worth upwards of $5, maybe even $10 depending on your roster ($100 cap). So your baseline has to be low enough to capture value in more than just the starters. 

The other is that a linear equation can't account for a non-linear value curve. Stick pre-2007 Tomlinson, projected for 50+ VBD points more than the #2 RB into a linear model and it's going to severely undervalue him. This year the top of the draft doesn't have much separation, so it'll probably work OK.

 

[abbottjamesr]

The problem is, the 216th player isn't worth $1, he just costs $1. Assigning him (and N players above him with value < $1) a value of $1 will take money away from more valuable players.

I don't think adjusting baselines is sufficient to get good auction values, for two reasons. One is that many bench players have non-zero value, some of those significant. Your top bench WR and RB are probably worth upwards of $5, maybe even $10 depending on your roster ($100 cap). So your baseline has to be low enough to capture value in more than just the starters. 

The other is that a linear equation can't account for a non-linear value curve. Stick pre-2007 Tomlinson, projected for 50+ VBD points more than the #2 RB into a linear model and it's going to severely undervalue him. This year the top of the draft doesn't have much separation, so it'll probably work OK.

It doesn't matter if they are worth $1 or not because it must be spent per the rules.  That $1 is not available to be spent on any one else therefore you can not figure it into the value of another player.

 

[dkp993]

It doesn't matter if they are worth $1 or not because it must be spent per the rules.  That $1 is not available to be spent on any one else therefore you can not figure it into the value of another player.

Agreed.  This is what I do too

 

[CalBear]

So, I redid my sheet with this change: The calculation for the top 216 players (sorted by VBD and then projected fantasy points) is:

( ( Total $$ - # roster slots ) / ( total VBDC * fudge^VBD-1 ) )+ 1

And for the rest it's the same without the +1.

After making this change, I felt like I needed to tweak the fudge factor, which is now 1.015.

Overall I think it's OK, although I'm a little concerned at what feels like overinflation of second-tier players. Carson Palmer is $16. But it definitely addresses the issue of only auctioning 10 players vs. 18. Top players are worth about $2 more in the 10-player version, which seems reasonable with an extra $8 per team spread out over 10 slots.