General VBD drafting strategy discussion (1 Viewer)

[Verbal Kint]

I know there are lots of guys hear who understand VBD alot better than I do, so if I'm missing something please enlighten me.

Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline. That way, if you can draft starters at each position with positive values, you know you have an above average team. Whereas with weakest starter, you can still be positive but have a below average team

Projections: When setting projections, do you include an injury risk factor in your projections, are assume a full game season? What about the likelihood of losing touches to another player. As a corollary, I'm trying to find a way to indicate the variability of a player's projections (in other words something to indicate that Rudi Johnson is as consistent as they come, but without the upside of Cedric Benson who is more of a boom-or-bust player)

 

[Brewzers]

(in other words something to indicate that Rudi Johnson is as consistent as they come, but without the upside of Cedric Benson who is more of a boom-or-bust player)

I hear you. Same concept as evaluating a stock, the "Beta" lets you know the volatility of a stock. It would be great to have a projection for Ced & Rudy with a volatility factor, say 1.4 for Ced & 1.1 for Rudy. I don't have anything to offer to that end, but its a good concept.

 

[fishergm31]

Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline. That way, if you can draft starters at each position with positive values, you know you have an above average team. Whereas with weakest starter, you can still be positive but have a below average team

By average starter do you mean NFL starter or fantasy league starter? I'm assuming fantasy starter because in a 1 QB league you could have a "better than average" NFL QB that is the 15th best fantasy QB, and thus not a starter for fantasy purposes. That being said, I've done VBD both ways, average starter and last starter. I've even done total expected drafted (i.e. I expect 45 RBs drafted, how does each RB compare to RB45?). I think no matter how you do it, it still comes down to getting the biggest number possible. With the average starter method, obviously you want all positive players, but what happens when your best choice for WR3 has a negative value? 10 team league that starts 3 WRs only the top 15 players are going to have a positive VBD value.

Projections: When setting projections, do you include an injury risk factor in your projections, are assume a full game season? What about the likelihood of losing touches to another player.

What I've done in the past for injury risk players is guess/estimate/project how many games that player will play and calculate my initial projections based on that. Then add to that the average fantasy pts scored by the average player at that position. So for example, lets say I think McNabb is going to play 12 games and average 25 pts during that time. (25 x 12 = 300), while the average QB in my league is projected to score 15 pts per game. So for the remaining 4 games I added 60 pts (4 x 15). So McNabb's total projection is 360. This drops his avg pts/game from 25 to 22.5 to more accurately represent the injury risk I see McNabb having. The reason I do this is because if McNabb misses an extended period of time, the QB position won't earn me 0 pts, it will earn me whatever I can find on the waiver wire to plug in for him.

As a corollary, I'm trying to find a way to indicate the variability of a player's projections (in other words something to indicate that Rudi Johnson is as consistent as they come, but without the upside of Cedric Benson who is more of a boom-or-bust player)

You could run 3 sets of projections - call them upside, downside, and expected. Then divide upside by downside to calculate a risk/reward ratio. You'd use the expected projections for your VBD calculations, but then have a risk/reward number to help in deciding if you want a consistent player or a boom/bust type player. I haven't done this in football yet, but I have done it in other fantasy sports (baseball, nascar) and it does help if you want to identify lower risk players.

 

[Verbal Kint]

Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline. That way, if you can draft starters at each position with positive values, you know you have an above average team. Whereas with weakest starter, you can still be positive but have a below average team

By average starter do you mean NFL starter or fantasy league starter? I'm assuming fantasy starter because in a 1 QB league you could have a "better than average" NFL QB that is the 15th best fantasy QB, and thus not a starter for fantasy purposes. That being said, I've done VBD both ways, average starter and last starter. I've even done total expected drafted (i.e. I expect 45 RBs drafted, how does each RB compare to RB45?). I think no matter how you do it, it still comes down to getting the biggest number possible.

With the average starter method, obviously you want all positive players, but what happens when your best choice for WR3 has a negative value? 10 team league that starts 3 WRs only the top 15 players are going to have a positive VBD value.

Projections: When setting projections, do you include an injury risk factor in your projections, are assume a full game season? What about the likelihood of losing touches to another player.

What I've done in the past for injury risk players is guess/estimate/project how many games that player will play and calculate my initial projections based on that. Then add to that the average fantasy pts scored by the average player at that position. So for example, lets say I think McNabb is going to play 12 games and average 25 pts during that time. (25 x 12 = 300), while the average QB in my league is projected to score 15 pts per game. So for the remaining 4 games I added 60 pts (4 x 15). So McNabb's total projection is 360. This drops his avg pts/game from 25 to 22.5 to more accurately represent the injury risk I see McNabb having. The reason I do this is because if McNabb misses an extended period of time, the QB position won't earn me 0 pts, it will earn me whatever I can find on the waiver wire to plug in for him.

As a corollary, I'm trying to find a way to indicate the variability of a player's projections (in other words something to indicate that Rudi Johnson is as consistent as they come, but without the upside of Cedric Benson who is more of a boom-or-bust player)

You could run 3 sets of projections - call them upside, downside, and expected. Then divide upside by downside to calculate a risk/reward ratio. You'd use the expected projections for your VBD calculations, but then have a risk/reward number to help in deciding if you want a consistent player or a boom/bust type player. I haven't done this in football yet, but I have done it in other fantasy sports (baseball, nascar) and it does help if you want to identify lower risk players.

Yes, I meant average fantasy starter.

 

[Maurile Tremblay]

Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline.

In a 12-team league that starts two RBs and one PK, using "average starter" will rank the PK5 ahead of the RB13 -- e.g., it will rank Matt Stover ahead of Willis McGahee.Even "worst starter" would rank PK11 ahead of RB25.The 100 pick method is generally better than either "average starter" or "worst starter," IMO.

 

[tuffydawolf]

Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline.

In a 12-team league that starts two RBs and one PK, using "average starter" will rank the PK5 ahead of the RB13 -- e.g., it will rank Matt Stover ahead of Willis McGahee.Even "worst starter" would rank PK11 ahead of RB25.The 100 pick method is generally better than either "average starter" or "worst starter," IMO.

 

[tuffydawolf]

I've got a headache after reading this thread.

 

[gheemony]

Couple of problems to be aware of:

1. If you try to convert to auction values, negative VBD causes problems. You need to make sure every player that will be purchased at auction has a positive VBD value. Note: this, I believe, causes a problem with the FBG conversion to auction values. I wouldn't trust those numbers.

2. I like to use more than one baseline and add up the values to come up with a VBD value. Consider adding VBD using average starter as baseline to VBD using average replacement as baselne (throwing out all negative VBD values before adding). This gives a nice weight to the better players.

 

[Z-Dog]

I know there are lots of guys hear who understand VBD alot better than I do, so if I'm missing something please enlighten me. Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline. That way, if you can draft starters at each position with positive values, you know you have an above average team. Whereas with weakest starter, you can still be positive but have a below average team

The biggest problem with weakest starter is that the last RB starter is usually worth more than the last WR starter or TE starter. The absolute KEY to VBD is that your baseline players must be worth about the same amount, or all of your cross-position comparisons will be biased.The 100th pick method says that in a 12-team league, but the 100th pick, people are selecting their 9th round players. In most leagues, that's just about when people would be drafting their last starter, but not always - many people will take backups at some positions prior to rounding out their D, K, or even TE. 100th pick is usually a better baseline. You can go crazy with this stuff - I once developed a VBD that used a backup baseline and a starter baseline at each position to determine auction values. The different flavors do produce some different results, but at the end of the day, you need to understand what aspects of value your particular flavor of VBD is emphasizing.

Projections: When setting projections, do you include an injury risk factor in your projections, are assume a full game season? What about the likelihood of losing touches to another player. As a corollary, I'm trying to find a way to indicate the variability of a player's projections (in other words something to indicate that Rudi Johnson is as consistent as they come, but without the upside of Cedric Benson who is more of a boom-or-bust player)

My take is that the difference in risk between someone we think of as injury-prone versus someone we think is reliable is negligible next to the overall risk that any player will get injured, so I don't factor in an injury risk to an particular player. What I do factor is injury-recover. Guys coming back from injuries aren't so much risks for more injuries as they are risks to not rehab well, and usually they have some risks to their opportunities, as teams hope to insure themselves by bringing in competition. The only injury that I actually factor for is concussions for QBs, since they are most at risk, and since having one concussion puts you at higher risk for more concussions - there's no rehab that can change that. I don't pay much attention to variability either, since I think that the major factor in determining variability is injury, and the next one is opportunity. If Benson hadn't played with Thomas Jones, hadn't held out of camp, and had been healthy throughout, he would have been a more consistent performer. Since I'm already accounting for opportunity, having a volatility factor would be a form of double-counting.

 

[MrJimiT]

I tend to favor the 100th player baseline over the last or average starter baselines when using a static value environment. I only like those baselines (avg - last starter) in already known statistical situations (such as baselining on last years or previous years stats). I play in a couple of different leagues, and the baseline standard that I use by evaluating past years drafting tendencies, is to baseline the number of draft opportunities to draft all starters. For example, in my 10 team league, we require 9 starters each week. So I will place my baseline on the 90th player. Thus, the opportunity for every team to have drafted all starters. This of course never happens, but it does give me a pretty respectable idea on how positions are viewed by my league during a draft. My 14 team league also starts 9 players per week so I would set the baseline for that draft at 126. Is it perfect? No of course not, but I tend to get a pretty decent value feel for the upper rounds of my drafts.

 

[ruelvis]

i'm contemplating a mix of baselines for various positions - not all worst starter or all average starter. Essentially it becomes a modified version of 1st 100 picks. My thought process is this: if my personal preference is to have a deep rb corps or to get an elite wr or to play a qbbc or def bbc i can modify the respective baselines to 'encourage' that. what i mean by that is this: by lowering the baseline player (lower anked player) the top players at that position get a higher VBD score. By raising the baseline player at a particular position, the VBD number for those players goes down. sooooo....

i may chose to raise my qb baseline player if i prefer to go with a qbbc (this causes qb VBDs to go down pushing them lower on the draft board meaning I’ll likely grab a qb later in the draft)

i may choose to lower my baseline rb if i prefer to have a deep quality level rb stable (this causes rb VBDs to go up pushing them up on the draft list meaning I’ll likely grab a few rbs earlier in the draft)

naturally, one should feel free to stray from the VBD as the draft dictates, but my thought is that being flexible with your baseline players can reinforce your personal preferences. In addition, you can tweak your 'need factor' (as mentioned towards the end of the VBD revisited article http://footballguys.com/05vbdrevisited.htm) so it varies by position as well.

Does anyone else do this? If so, how well has it worked for you? Any thoughts/comments?

 

[Rounders]

I know there are lots of guys hear who understand VBD alot better than I do, so if I'm missing something please enlighten me. Baselines: It seems that weakest starter and the 100 pick method seem to be the most accepted baselines. To me, average starter seems to be a better baseline. That way, if you can draft starters at each position with positive values, you know you have an above average team. Whereas with weakest starter, you can still be positive but have a below average teamProjections: When setting projections, do you include an injury risk factor in your projections, are assume a full game season? What about the likelihood of losing touches to another player. As a corollary, I'm trying to find a way to indicate the variability of a player's projections (in other words something to indicate that Rudi Johnson is as consistent as they come, but without the upside of Cedric Benson who is more of a boom-or-bust player)

Believe me when I say this, I'm no "expert" but here's what I do:Baselines - I go back at previous drafts and use the 100 pick method. I actually did this for the first time last year (used worst starter method before) and I can honestly say I was able to use the DD a lot more effective towards my league. It seemed like I was able to use the DD to accuratly see the position runs, just much more league specific.Projections - I start with Dodds projections and tweak them to make them more in align with my opinions.Hope this helps.