VBD (1 Viewer)

[Danish Bengal]

I'm working on doing my own value based rankings on my Oracle database and would like some input on a few issues. Any comments appreciated.

Doing your own rankings obviously start by doing projections for all players. Then you calculate each player's projected fantasy points ('FP') which gives a fine ranking within each position. My problem is how to turn those position rankings into a single ranking for all position.

What I have done is comparing each players FP to an average for that position. I feel that's the obvious way to do it, But how many players do you average over? I figure the correct number of players depends on the number of starters for each position in your league. So in a normal 1/2/3/1 12-team league, the average for QB's and TE's would be over the Top12, for RB's the Top24 and for WR's the Top36. So far so good.

My problem is that if I do that and compare my results to the rankings on FBG and other sites (with their own projections obviously), I can easily see that's not how they do it. Actually no matter how I tweek the averages, Aaron Rodgers should be the No.1 in the rankings by quite a margin. And he isn't. So FBG and others obviously doesn't do it the way described above.

How do you feel the correct way to calculate value based ranking is?

Is my approach with averages wrong?

Is fractiles the way to go? (i.e. comparing to the no. 12 ranked QB, not the average of Top 12)

Are combined rankings in reality not based purely on math?

What do you do yourself?

Any other comments?

 

[moleculo]

I'm working on doing my own value based rankings on my Oracle database and would like some input on a few issues. Any comments appreciated.Doing your own rankings obviously start by doing projections for all players. Then you calculate each player's projected fantasy points ('FP') which gives a fine ranking within each position. My problem is how to turn those position rankings into a single ranking for all position. What I have done is comparing each players FP to an average for that position. I feel that's the obvious way to do it, But how many players do you average over? I figure the correct number of players depends on the number of starters for each position in your league. So in a normal 1/2/3/1 12-team league, the average for QB's and TE's would be over the Top12, for RB's the Top24 and for WR's the Top36. So far so good. My problem is that if I do that and compare my results to the rankings on FBG and other sites (with their own projections obviously), I can easily see that's not how they do it. Actually no matter how I tweek the averages, Aaron Rodgers should be the No.1 in the rankings by quite a margin. And he isn't. So FBG and others obviously doesn't do it the way described above.How do you feel the correct way to calculate value based ranking is?Is my approach with averages wrong? Is fractiles the way to go? (i.e. comparing to the no. 12 ranked QB, not the average of Top 12)Are combined rankings in reality not based purely on math?What do you do yourself?Any other comments?

first of all, Aaron Rodgers should not be #1 overall in VBD. If you look at your projections, the difference between rodgers and QB6 is less than the difference between RB1 and RB12 - that is the basis of what BVD is all about.I use an average for establishing a baseline too, but I average the # of starters by 1.5 - that is, in a 12 man league, I use the average of the first 18 QB's. That value typically ends up being around QB9, so my baselines are generally a little higher than last starter.How did I come up with 1.5 * # starters? guess and check, really. I just played around with the numbers until I found a mix that seemed to mimic an ADP fairly closely.how did I

 

[joffer]

Joe's secret formula

 

[GregR]

Joe's secret formula

By which he means, Joe Bryant came up with a formula for setting a baseline which he thought would give a better final cheatsheet.... but taking into account some other factors that don't otherwise get included. I can't tell you what all is or isn't included, but I imagine it included things like the increased likelihood of injury to RBs compared to some other positions, the scarcity of backup RBs compared to backups at other positions, the diminished value of TEs after the top few, etc.So it tends to set a baseline deeper for RBs and shallower for other positions, and how much depends on the position. Now that said, I imagine the Rankings page isn't just a straight up VBD cheatsheet. It probably takes into account even more than Joe's formula does, not to mention a given staffer might not want to go with Joe's formula, they might use their own, or Maurile's method, etc.

 

[Danish Bengal]

Guys, thanks for your input.

I worked quite a bit on this yesterday and ended up using molecule's 1.5 factor. I like the fact that you could actually argue that 1.5 just means that you weight back-up with 50%, so it's not just a number pulled out of somebody's ###.

I don't know what you guys with flex spots, but after some testing I ended up by applying 60% af a flex to RB, 30% to WR and 10% to TE. That means that in a 1/2/2/1 12-team league with 2 flex spots, the baseline is the average of Top-18 for QB's, Top-58 for RB's, Top-47 for WR's and Top-20 for TE's

I like the rankings this gives me quite a lot.

Now on to Dynasty Rankings....