Adjustments

[Average Joe 0]

Let's say I input my scoring system and compile. The overall standings come back with RBs in the top 10 positions then a WR then another 30 RBs. How do I account for this kind of distribution and even recognise when I need to account for it if it isn't quite as obvious as the example I gave? Meaning the WR should go first as he will obviously have good RBs to pick up on the way back.

[Mike MacGregor 20]

Do you mean Joe, how do you tell if the rankings produced by the Compiler are not reasonable for your league, for some reason?

[Average Joe 0]

Here is a highly exaggerated example of what I mean. Lets say four people are going to draft 2 players apiece in serpentine fashion. 1 running back and 1 wide receiver. There are a pool of eight players as follows

POS VBPoints

RB 200

WR 170

WR 140

WR 120

RB 100

RB 90

RB 80

WR 10

 

The first three drafters have a choice whether to take the highest RB or the highest WR. After that the fourth drafter will take the highest of each and the rest will fill in the missing position. This creates eight seperate scenarios.

 

1 200 170 140 120 100 90 80 10

2 200 170 100 140 90 120 80 10

3 200 100 170 140 90 120 80 10

4 200 100 90 170 80 140 120 10

5 170 200 140 120 100 90 10 80

6 170 200 100 140 90 120 10 80

7 170 140 200 120 100 10 90 80

8 170 140 120 200 10 100 90 80

 

If I take the RB first, I will end up with 210 VBpoints whereas if I take the WR first, I end up with 250. If I take the WR then drafter two should then optimize his score and take the next WR to end up with 230 VBPoints instead of 210. Drafter 3 does likewise and this leaves drafter four with the stud RB and the crap WR. I do realise this doesn't take into account someone snatching two of one position and that a real draft is much more complicated than this, but is there a way to recognise if it would be more feasible to take a lower player on the overall cheatsheet in order to maximize my roster?

[HBK22 0]

Well, you're not calculating VBPoints right (assuming you mean thats their VBD). The last RB and the last WR would have VBD's of 0; they would represent the baseline...

 

I do like your question though, it is being discussed extremely thoroughly on some other message boards - is there a mathematically optimal draft order?

 

So far, the answer is yes and no. The yes is that it is a solvable math problem, therefore there is technically an answer. However, the problem is that the answer changes after every pick. Here's why:

 

Assume you're in a 10 team league. You pick first. You assume that say 15 RB's will be off the board before you pick again. So you pick a RB. BUT, assume that all 19 selections after yours are QBs and WRs. In that case, you should have gone with a QB or a WR because of what was available to you with your next pick. Its extreme, but it explains the thought process.

 

I was exactly like you in terms of wanting a mathematical solution, but I've decided to focus my energies on looking at mock drafts to get a feel for what might be available to me in each round, and then really scouting the players. There is way too much variability in a draft to try and do it mathematically. VBD is terrific in that it gives context to player performace (vs a baseline) but it should not be used to determine value past that in my opinion.

 

Mike, your thoughts??

[chtrich 0]

Oooh, I like this topic. I think this is where what they call DVBD (dynamic VBD) comes in. The dynamic part being the predictiveness of what the next pick will be and therefore what you pick should be even if it's not the best VBD option because there's a good chance that option will still be there in the next round.

HBK22- do you have a link to the post in the other forums?

 

Chris

[HBK22 0]

If you go to footballguys.com it's all over the place.

 

DVBD changes the baseline - instead of using the last starter, you use the players you expect to be available when you pick next.

 

I also love this topic but I think it can be overblown.

[Mike MacGregor 20]

My thoughts? From what I've read and thought about on this dynamic VBD thing, it is more theory then practical application.

 

I'm very skeptical of being able to predict what the number of picks will be from point A to B in the draft, which seems to be a key component, especially for the huge variety of leagues out there that all of you guys and gals play in. 1 QB, 2 QB, 3 QB (yes, I've seen it), no flex, monster flex, somewhere in between, TE, no TE, IDP, no IDP, 6 team, 8 team, 10, 12, 14, 20, 24, etc., etc.

 

Even more important though is that even if it is possible to develop it into some kind of practical application, I just don't see the cost-benefit of doing so. I'm thinking it will be very complex, with little in terms or real results in the end.

 

This process works very well for me and I don't see it changing really:

 

1. Run the overall rankings as currently done with adjustments in the weighting of certain positions. Exclude K and DEF as I will always draft them near the very end. Possibly exclude IDP too, as I will tend to draft them later, mostly after the point of using the overall list.

 

(In more complex leagues, I may run and review the overall rankings based on a method with no weighting and all positions included, just to see how the list shakes out.)

 

2. Assess if the overall list at the top end seems reasonable based on my understanding of the league. Are the positions well distributed? Does it seem reasonable to consider these X many RB before my first WR, second WR, first QB, etc. Where are the top 3 TE? If it is off from what I was expecting, then I probably messed up the inputs somewhere, either by typo or mis-judgment.

 

3. Draft Time: Use the overall list for the first 5-6 rounds (12 team), but don't be a slave to the overall list. If I don't want to take a QB yet because there are a handful of lower ranked guys who's upside I really like, but the overall list says take a QB, then don't take a QB. Monitor my Average Draft Position (ADP) data, tiering and other team needs around me. If I have a guy higher up on overall but low relative ADP that I can probably get later, wait on him. If my top overall list picks a bunch of similar ranked guys at the same position (i.e. in the same tier), and quickly assessed drafting needs for other teams is not going to pull all the players out of that tier, then wait there and strengthen somewhere else.

 

4. Once I've had about enough of the overall list, where my team needs are more important, then go to the positional lists. Even more emphasis on tiering now. Continue to follow the ADP and other teams needs. Fill out your roster for probably all your starters except K, DEF, maybe IDP and TE (the most important positions) and key backups who were good drafting value.

 

5. Even later in the draft, still using the positional lists, but will take more chances by picking sleepers and high upside guys even though they are not necessarily ranked as high as others. If they flop, I can almost always replace them via free agency. Flip back to the overall list to see if there is anyone who has really fallen in the draft from where he was ranked, and consider taking them. Get backups for your studs if necessary. And of course, fill out that roster, even taking a K, DEF, etc.

 

Ok, so that is my draft "strategy". I'm not sure how dynamic VBD is going to help me improve on it, or make me see differently that this isn't a solid way to approach things. Certainly I am open to comments and feedback though.

 

[ 07-18-2003: Message edited by: Mike MacGregor ]