By John Bush
Best at Best Ball Part 1 Introduction
The Best of Best Ball series is several articles showcasing my landscape approach to Fantasy Football. We will drill down from the top to bottom. The data will drive the questions.
This can be in contrast to other FF pundits’ approach who start bottom up. A simple formula picks a “hot” player and does an article to attract readers. It is like looking at a puddle of water and trying to determine the shape of the ice cube (square, half moon, round etc).
The landscape view is watching the ice cube melt and then using analysis to determine the puddle characteristics. I do not approach writing bottom up as you quickly run out of reasons for why worry about a player. Does it really matter if Brown vs Hopkins is 1 or 2? They are both at the top. As you will see as I lead into the depths we will arrive at the true confusing problems of player selection in Best Ball but only if the data points us to that path.
Algorithmic Approaches to Prediction.
Several years ago, I determined a simple proxy for converting ADPs into projected Fantasy Points per Game. I use this to drill into the current public’s thinking (Crowd Wisdom = ADP). That provides all a target to shoot at. Questions are easily conceived this way!
Figure 1. Overal Characteristics of the ADP based Best Ball FP/G
Below is the data that frames the database I created using current Best Ball ADPs. The overall Average ADPs for QB to WR is shown. As seen elsewhere the value of Best Balls is certainly RBs at 88 followed by WR at 97, then QBs at 112 and finally TEs at 123 ADP averages. This mimics the draft patterns of current ADPs in Best Ball leagues
The FP/G, however, is QB, WR, RB, and TE. The numbers of each position account for this pattern. The % of total FP/G is also been calculated and the QB stand out as the valuable position. Fortunately, late QB drafting is a good solution as many QBs have nice value predicted this year.
Dividing the % total by the position num arrives at a metric that is in line with individual position value. Given the low numbers of QB and TEs ever drafted 2 to 3 each via the typical 6 to 8 RB to WRs drafted it is understandable the individual position values are skewed.
I conclude the projected FP/G data falls in line with common sense reasoning about FF player value and ADP. Thus I continue onward.
Figure 2 A and B. Draft Pick vs FP/G Raw and Scaled to Average Metrics Table and Waterfall Graph.
I have used this draft pick technique for many years to project an “auto-pick” team formulation from draft pick 1 to 12th. Given the current ADPs, what would such a team be worth or in this class produce in FP/G? Please see my discussions of reference class forecasting for planning biases.
(Textbook on Kindle has much to say Kindle Edition https://www.amazon.com/dp/B07DD75F5P )
The raw data suggests the best draft picks are 1st, 4th and 11th while the 7th to 9th draft picks seems to have issues. The waterfall graph is a visual representation of the scaled FP/G averages. It nicely highlights the tops and bottoms of the 12 draft picks.
Why is that? Let’s turn to the next figures.
Figure 3 Draft Picks for Auto Draft Teams and Positional FP/G Averages.
The data from the overall draft pick was interesting but what was the reason? The next level drilling down is the positional FP/G at each position type. For example. Draft pick 1 led to an auto draft team having a QB group at 16.6 FP/G, RB at 13.5, TEs at 12.5 and WRs at 10.4. Note the “Poor Draft Pick from above” #7 had only RBs and WRs of 9.6 and 12.0 respectively. Draft Pick 8, had RB, 10.9, TE, 7.1 and WR at 9.9.
The answer from Figure 2 does not hold up! We, therefore, must drill down more! The analysis from Figure 3 supports that journey.
Figure 4 A, B, C and D. Position Specific Draft Picks for Auto Draft Teams and Positional FP/G Averages.
These data streams focus no only on the position. If the auto draft team did not have a QB then that pick would be eliminated for analysis.
Knowing these data should allow extra homework before a Best Ball (especially Slow Drafts). What plans do you need to change the poorer draft slots? Draft earlier? Stream QBs for sure?
Here is my introduction to Planning for FF
The waterfall graph highlights the QB sweet spots are the 5th and 12th picks while the 1st, 2nd, 6th, and 11th draft picks are the poorer QB draft picks.
4B. Running Backs
The obvious RB hot-spots were 1st and 2nd. The RB poorer draft lots are 3rd, 6th, 7th and 12th.
4C. Tight Ends
Best TEs spots are 1st, 3rd, 4th and 12th. The worst is 6th, 8th, 11th.
4D. Wide Receivers
Hot Spots for WRs are 3rd, 6th, and 11th. Concerns exist at 1st, 5th, and 8th!
These data support a much more robust approach to Best Ball than I have previously seen. I think these selected draft slots issues should invite running mental models of what ifs. I propose using these with my positional landscapes in Best Ball.
Link to see an example of Positional Landscapes
Figure 5 A and B. Tabular and Graphical Analysis of Positional Sectors vs Average FP Per Game.
As I discussed, we must go top down to burrow through the “weeds” to see what is actually of importance and in what context. These parts of Figure 5 continue the digging for FF gold nuggets.
It is important to “know” the worth of player based on ADP values. The proportion of that worth is also displayed in the waterfall graph. Nice visuals.
Another key to understanding this approach of analysis is to “enjoy” the journey. A quick system 1 view of thinking and off you go to some other article! It’s not the end of the road it truly is the journey. The slower and more in-depth your thinking arriving at system 2 thinking the better. (see my textbook on this issue). I also published this article.
Key – Analysis of data suggests that more information does not increase our understanding unless the information increases our coherence of the data set.
Slower thinking and going over the data etc will increase your coherence!
The figure below focuses on the raw positional data and the scaled to average data. That is the clarifier for us! The closer values in QBs is seen vs the extremes in RB and WR top players vs the rest.
QB and TE Top 6 players are ar 4.5 above the average vs the WR and RBs at 8.9. This is a two-fold difference and supports later QB and TE drafting. Those differences grow in the 7th to 12th players of 0.8 (QB/TE) vs 4 and 4.4 (RB/WR). This is a 4 fold difference.
The waterfall graphs hammer home these differences. The extremes between and within the positions are clear. Note the WRs have greater later value in the later 3rd and 4th segments than RBs. RBs are starting to become very similar in those mid segments. You, therefore, can still on average in Best Ball get more WR bargains in segments latter than the RBs hence the early RB hoarding leading to WR collecting.
An alternative view is that in the QBs and TEs the top 6 are the main ones. I agree and I use the “barbell” drafting. I draft QB/TE if a key player drops too low else I skip and go late and use the power of Best Ball to catch multiple good weeks from multiple players.
Figure 6 A and B. Team Level FP/G Averages and Scaled Averages.
This data may be side by side digging but I require an overview of Team’s and their perceived FP/G production from the public.
- Does the public over or under value team level FP/G this year?
- Where are the public overlays and underlays vs value?
These data allow a further depth for Best Ball drafting. Are PIT/LAR/KC/HOU so better than NYG/MIN/ARI/NO? Is WAS/DEN so bad, both have good WRs for PPR! I think MIA/BAL are undervalued as well. I have been collecting WRs on those teams later in Best Ball drafts.
I suggest if you agree or disagree do some homework to confirm. Remember this is based on the current public Best Ball ADPs!
The waterfall graph allows a nice snapshot of team value vs the ADP average! Use to consider the mistakes that may be in here!
Figures 7 A, B, C, and D. Positional Based Team Centric FP/G Averages and Scaled to Average Metrics.
I end this part with a team level view of positions. Where are the bargains? Again these data should add to your journey in the Best Ball world! I will discuss the waterfall graphs for the next data series as I believe these visual presentations are the best for landscape-based understanding.
In the QB waterfall graphs, the clear ADP tagged team QBs are GB, HOU, SEA, NE, and CAR. The question is the bottom teams of MIA, NYG, CLE, CHI, and JAC having poor QBs? There is some issue as NYG has OBJ etc. Also, note the SKEW of the waterfall.
The number of very similar QBs as rated by the BEST BALL ADP is large and skews right. We are either seeing an overvalue of the top QBs of undervaluing of the MID and or Bottom QBs. Investigate that question. I am late drafting 3 QBs include several rushing based QBs as discussed in another article (Late QB article link ___)
The waterfall graph is very skewed! Note the QB discussion. It implies after the first 5 or 6 teams and then everyone else is very close in the midsections. That sector stretches for many teams outward.
Other Parts of this series will make it deeper and the question to consider is;
Does the team have a crowded RBBC or bad RBs or both? You must answer that question in full before BEST BALL Drafting. For example, CLE has good RBs but 3 in the mix. It’s a clear RBBC!
The Top Team for TEs are the obvious ones. NE. KC, PHI, NYG, and GB. Note the SKEW as well. Are the top TEs overvalued or the last TEs undervalued? I use the barbell drafting approach and draft 2 to 3 TE for BEST BALL teams. I am focused on NO, ATL, OAK, and BUF for some late TEs!
In the waterfall, the WR data is SKEWed. We see an across the board SKEW! I propose the ADP are missing key players in the mid and bottom layers. Dig for bargains!
Are the “poorer” teams really that bad? DAL, NYJ, JAC, WAS and MIA are all favorites for my late WRs targets. Be really with further investigates before your real draft.
I end part 1 here. The next article show is coming this same week. It might be one or two more. We will make it to the bottom and finally ask the key questions for investigation. Please come back! Thanks for reading.