By John Bush
Best of Best Ball Part 2
Best of Best Ball Part 2 continues with my analysis of current Best Ball FF drafting.
I discussed in Part 1 (Best at Best Ball Part 1 https://www.fakepigskin.com/?p=36666 )
the genesis of my investigation. I continue with a top-down view of the positions, player and their ADP based predicted FP/G.
In this article, I am going to look at the position level followed by FF players across the draft rounds and then their associated FP/G.
Questions to consider are;
- Does the player pattern suggest a sensible drafting approach?
- What rounds do we arrive at that position of where the player differences are slight?
- Does the rule of ” take the lowest ADP player in a group of similarly ranked players” make sense?
- Where are the mis-evaluations of the public’s ADPs for that position?
- Does the Value over Replacement Data change your original conclusions?
The Format of Positional Figures
% of the Top FP/G Number
Value vs the Positional Replacement (VOR) Average in a League (QB and TE -12th, RBs 24th, and WRs 36th)
The VOR metric is my focus in approaching the questions about. The differences in VOR become small after Brady is selected. The differences are 3% or so continuing step by step. 10% of VOR difference exist between Cousins and Mariota (7 QBs in this block here).
Note another 10% VOR exist from Mahomes to Tannehill. I conclude I will tend to go late given the relatively small differences and as discussed elsewhere given extra credit for late QBs with rushing ability!
The QB waterfall confirms the VOR pattern and is unsurprisingly skewed seemly to the TOP. Does that suggest over-valued QBs in the slots? Note if one considers the WRs which have the most extreme in waterfall VOR graphs, the QB might be the most efficent position rated by the public.
QB Player FP/G vs Draft Round
This is a visual from high to low FP/G point while going across the draft rounds. If nothing else this presentation should highlight the flow of each position (When, Clusters and Values). Use as a tool along your Best Ball for deeper understanding.
Why draft Watson/Wilson when Brees is very close to FP/G?
From Big Ben (round 9) to Smith (round 11) it’s about 2 FP per game differences. Smith has some rushing game as well. That is about 10 QBs in distance from those 2 points. Late QB!
Streaming with Smith and another QB should be close to getting a weekly average near 14.5 and you only are spending 12 rounds of draft capital less than Rodgers. Based on other data (Not published) I predict the difference is closer to 6 points a week difference from Rodgers to say, Manning. More support for late QB drafting.
It seems unlikely Gurley repeats as a number 1 RB in 2018 (Last RB to do this was Preist Holmes). That conjuncture suggests Gurley’s value in VOR is too high. Note the extremes in overall VOR within RBs. The first 5 RBs are withing 60 VOR units. That is extreme! Note the next 6 RBs are 1/2 of the 125 VOR into the 60% VOR. Thus in the first 14 RB, we have traveled down 100 VPR points. We must drop 28 more RBs to drop another 100 VOR units. That is a 1/7 of drop-in VOR velocity!
Thus a question of value in the top 14 vs then next 28 RBs exists? This has driven the greed for top RBs in previous years. The curve would be plateauing very soon coming out of the top RBs into these mid-RBs.
In WRs (seen later), the top WR is at 50% VOR and it takes 36 WRs in the draft to reach a -50 VOR drop vs only 4 RBs are needed to see a similar drop (125 to 71 VOR). 4RB vs 36WR would imply that public values RBs 8X vs the average RB? Is that real?
The extremes in RBs drafting in Best Balls is based on this and maybe unconscious mental value! You must decide to go contrarian or if an early draft slot grab a key RB. That can work until injury when the ZERO RB bug bites the Best Ball drafter.
The waterfall graph brings home the extreme RB landscape in VOR metrics. The skew has to be dealt with in your plans. I use the draft slot I get to determine my actions. I do not seek ZERO RB but if needed I will do it!
RB Player FP/G vs Draft Round
The first round has the highest cluster of RB of 9 RBs. The 2nd Round has only 4 RB drafted. Continue that exercise and map the RB clusters. The RB ordinal series looks like this pattern: 9, 4, 5, 3, 4, 5, 8, 3, 1, 5, 3, 1, 6, 3, and 4
In my unpublished data with the 13 to 24th RBs. The top 13 to 36 differ by .25 points/game/draft pick. I use that a finer ruler to consider RBs in this range. Is there a big difference from Guice at 12 to Duke at 10 FP/game? That’s about 17% difference. Very small considering the draft pick differences.
Again in the middle area where there are a lot of same guys to draft. How do you separate them? Predraft plans! Knowledge of Team dynamics etc goes a long way to figure out your selections. Note the ruler number would say 2.5 points per game difference which is 21% difference.
In most leagues of Best Ball, player gathers 2 to 3 players to ensure a weekly production from them. The draft strategy tends to be as varied as the QB drafting. I use the Barbell approach if the opportunity arrived. If not then I tend to draft late TEs for my Best Ball teams.
The top VOR is Gronk at 48% that is in contrast to 125 VOR in RBs and 47 VOR in QB. The reason the TE and QB are seemingly linked are their numbers on Best Ball teams are the same and the VORs are identical. I employ the barbell approach in QB as well.
The differences proceeding down the TEs are 13.4, 7.8, 5.3, 3.5, 2.9, 2.6 etc. Note that plateau effect in VOR drops. The change from Gronk to Reed is 40 VOR units which are 5 VOR unit per player! The plateau effect invites a late TE approach!
The waterfall graph visualizes the VOR drops and in TE we see an extreme skew late. This is very dramatic. One can interpret this as early TEs are overvalued.
TEs Player FP/G vs Draft Round
We have the TE drafting rounds vs FP/G. Gronk is at 18.3 FP/G and Ebron is at 8.3. Ten points per game. That is 17 TEs to get to 8 FP/G. Thus 10 FP/G difference can be divided by 17 and that implies at 0.58 FP/G difference per player. I have unpublished data that its overall closer to 0.44 and thus these differences projected by the public are too high. Overvalued by ADPs. Late TEs drafting being supported.
Antonio Brown is at 57 VOR and to get to -50 VOR units requires a drop of 25 WRs to get to 5.7. That is a drop of 2 VOR per player. In the draft, the first 25 VOR units are gone by AJ Green 7 WRs drafted. The next 25 VOR units then are gone by 18 more WR. The plateau is clear in here. This has driven early WR drafting given the extreme drop in VOR.
The extreme waterfall of WRs is like the TEs in that the late WRs are almost interchangeable. That suggests an overall tendency of early WRs being as a group over-valued? The skew pattern was not predictable from any of my previous data investigations and I have never seen it discussed.
WRs Player FP/G vs Draft Round
As with the previous positions, I present the FP/G across the draft rounds. The loss of 10 FP/G goes from Brown to Baldwin (the first 12). That’s an average of 0.833 FP/G per player drop. My own data suggest this is an overvalue and it’s closer to 0.54 FP/G/Player. I believe the public is in error even in the top 12 WRs. Thus my data suggest Baldwin is at 18.8 FP/G on average vs 15.4 FP/G. the public has undervalued him by 20%. Beware of the ADP based estimates.
The ordinal for WRs is; 4, 6, 5, 8, 4, 3, 6, 4, 4, 4, 2, 6, 5, 2. 4 and 6.
The ordinal for RBs is; 9, 4, 5, 3, 4, 5, 8, 3, 1, 5, 3, 1, 6, 3, and 4
I added red highlight to emphasize the patterns
The clear differences are RBs in round one and WRs in round 4. The draft patterns are RB then WR/RB, then WR/RB and WR in round four. The need for a balanced team supports this pattern. Extremes of RB RB RB RB and WR WR WR and WR are definitely contrarian. Can we may a case for 3WR and 1RB vs 3 RB and 1WR? Predraft plans!