#1 By: RotoViz, January 12th, 2017 13:16
Originally published at: http://rotoviz.com/2017/01/is-it-time-to-scrap-zerorb/
Zero RB appears to be in trouble, and its creator saw it coming. Last August in an article about the 2016 Apex Expert’s League, Shawn Siegele wrote the following about the impending decline of Zero RB as the top fantasy drafting strategy: We’re probably at peak Zero RB. One of the young studs at RB will likely…
#2 By: PaulC, January 12th, 2017 14:45
This was well done, very good insight. But I'm not sure I completely follow how value is completely incidental. Doesn't the distribution become important because WRs score on par (especially with PPR) with RBs? As a hypothetical, let's say the NFL suddenly went back to offensive styles and game plans more like we saw in '90s, only with even less passing, with far fewer pass attempts and far more fantasy points being scored by RBs than WRs. And let's say we were still able to project the top WRs more accurately, same as now (same probability density curves as above). Would it still be optimal to load up on WRs early? The edge of having more accurate projections would be tiny, because each WRs scores such a small amount of points. Maybe I am missing something.
Or is it that the injuries to early-round RBs, the RBs that are so high scoring in my hypothetical, would still happen so much more frequently that you wouldn't be disadvantaged loading up on the lower-scoring (but more accurately projectable) WRs?
#3 By: Genghis Khan, January 12th, 2017 14:47
"Zero RB is a structural strategy". You utilize the years of 2009 through 2016 for the evidence basis. If the strategy was applied in the Priest-LaDainian days, could one not make the same "structural" argument? And wouldn't it be a losing argument (especially in standard leagues)? It doesn't take too many David Johnson's, Zeke, LaDaianian's, Priest's with overwhelming upside to outweigh any advantage?
#4 By: Josh Hermsmeyer, January 12th, 2017 15:56
ZRB is a strategy based on the structure of common fantasy football leagues. This includes roster constraints and the notion of a draft. If you start changing the structure of the game (say it moves from 2-2-flex to all flex and an auction) not every assumption will hold and new analysis will have to be done. That's why I call the strategy structural.
It just so happens that the forecasting advantage for WRs occurs in the early rounds. If it occurred elsewhere the strategy would change. But it is not affected by the scoring settings of the league or the offensive environment. That is value based thinking.
If RBs suddenly became worth three times what they are now, as in an extreme version @NewJerichoMan's example, and the forecast distribution remained the same and the roster constraints remained the same, the Zero RB strategy does not change. The edge remains.
The assumption implicit in both of your comments is that you are missing out on the high scoring players. But this is incorrect, for two reasons.
1) We don't know who the high scoring RBs will be. For evidence, have a quick look at what recenecy bias is doing to people's brains when they tell us to "just draft a top 3 RB".
Here was 2016's top 3 by ADP:
More generally, here's the error distribution for the top 3 over the past 7 years.
2) Injuries allow you to acquire RB at a later time while still reducing your variance relative to your competitors by loading up on WRs. Injuries DGAF about your league settings or the offensive environment. They are a structural feature of football.
The fact that all these pieces fit so well together is the genius of Shawn's creation.
#5 By: Matt, January 13th, 2017 09:04
In general I agree with this and, despite using Zero RB in this year only (with mixed results), I will be continuing to use it as I think your logic is sound.
There is one minor point I would dispute. You wrote
RBs will nearly always have a higher serious injury rate, and an injury means zero fantasy scoring.
You write this as thought it's a bad thing, but I'm not convinced that it always is bad news. A player that you know will have zero fantasy scoring can be replaced in the starting lineup by someone who will score points. I would rather have a player who is injured and misses games than someone who is limited due to injury or who for whatever reason underperforms.
For example, I would rather have drafted Keenan Allen this season than DeAndre Hopkins or Brandon Marshall. At least with Allen I had the certainty that he would be out and knew I had to replace him, whereas with Hopkins or Marshall, I'd guess that owners kept putting them in the starting lineup much of the season and hoping for the best.
I guess what I'm trying to say is that sometimes certainty is better than uncertainty, even when the certainty is that no points will be scored by a given player.
#6 By: Daniel Schultz, January 13th, 2017 13:38
I'm not going to argue that Zero RB should be scrapped, but wouldn't it be appropriate to at least pump the brakes a bit to insure everything is being evaluated? A quick review of the players that won in MFL10s http://rotoviz.com/2017/01/picks-won-mfl10s-2016/?hvid=5EqfMF) this year showed David Johnson, Le'Veon Bell and Elliot to have 26, 15.9 and 9.2 % win percentages among running backs drafted in the first 12. The same numbers for Brown, Beckum and Jones were 13.9, 9.9 and 8.8%. I think if we were all playing to simply be in the top half, the zero RB would still seem to be the best way to go, but when looking at the risk reward, the RBs appear to have a greater upside that must offset at least some of the additional injury risk. I can't say how much, but I know if I am picking in the top 3 in a standard league next year, I will have a hard time passing up Bell, Elliot or Johnson. An additional thing to consider is some of the dominant backs on good running teams have well defined backups that can be drafted near the bottom of the draft. Having a Elliot/Morris or Bell/Williams combo provides some security and the handcuff is usually cheap to obtain.
When looking at injuries, wouldn't the QB injury rate add more risk to a WR value than a RB, and if so, has that ever been accounted for? I don't believe it has, and a quick check of splits shows Antonio Brown at 19.6 PPR with Ben, and 11.2 without him. Le'Veon has a split of 21.8 with Ben, and 20.4 without. A second check showed a less extreme but consistent pattern when looking at splits with Luck, Hilton is 15.1 with and 11.6 without. Gore is 13.6 with, and 11.6 without. I realize my "sample" is insignificant in quantity, but if that trend is valid, then QB injury rates should downgrade the advantage for drafting WRs at least some amount.
One aspect that I think is under appreciated is the consistency of scores and this is another area where top RB's out perform top WR's. Using FFTodays consistency scores, the three running backs noted above were 77-82. For the WR's above, the values are 34-38. When playing a weekly head to head game, I will take less variation in scoring assuming the same average. Again, I don't have data to back up that belief, but consistency for a player makes them much easier to simply plug them in on a weekly basis.
Those were just some thoughts and comments, I look forward to reading your responses.
#7 By: Josh Hermsmeyer, January 16th, 2017 12:30
Thanks for the comment. It's an interesting point. I think conceptually there is a way to test it, but I also think it would be difficult. You would need to see how much fantasy scoring is affected by being on the weekly injury report but still playing. And then you would need to see what the distribution of outcomes is. You would basically need to find that the distribution is heavily skewed toward zero fantasy points for this idea to bear fruit. My best guess is that if there is an effect it is small, but that's just informed guessing.
As for this:
For example, I would rather have drafted Keenan Allen this season than
DeAndre Hopkins or Brandon Marshall. At least with Allen I had the
certainty that he would be out and knew I had to replace him, whereas
with Hopkins or Marshall, I'd guess that owners kept putting them in the
starting lineup much of the season and hoping for the best.
I just can't disagree more. Both players got the volume we expected pre-season, and the NFL random number generator that we call efficiency produced an outcome that was on the tail end of the distribution for both players. Give me the shot at a WR1 outcome every single time over an injury. I would just strenuously suggest that taking outcomes and retrofitting them into your previous decision making is maybe not the best process.
@Badger Thanks for taking the time to comment.
MFL10's are structurally different than standard leagues (defined as 2-2-flex with a snake draft,in-season waivers and trading). A large part of the ZRB strat revolves around top RB production being replacable on the waiver wire and acquirable in the later rounds. For instance, here is the probability density curve for top 3 WR and RBs over the past 7 years. The x-axis is pre-season ADP.
You can see that ~ 16% of top 3 RBs were taken with picks greater than or equal to a positional ADP of 25 (ie. outside the 5th round).
The handcuff argument is one that has been explored in excruciating detail, and in fact is a counter-strategy to ZRB's antifragile thesis. Please see Shawn's original article for further detail on why antifragile is a superior strategy. http://rotoviz.com/2013/11/zero-rb-antifragility-and-the-myth-of-value-based-drafting/
I personally have not looked into this, but like all splits I take a very dim view of their value in making forecasting more accurate. Even half-season splits are dreadful at picking up signal due to the high degree of variance in the smaller sample.
This isn't to say that the idea isn't good or that it isn't logical - it's just that when you look at splits what you find is there is nothing actionable in the data from a predictive standpoint. So any moves made based on them is really hunch-driven, not evidence driven.
One aspect that I think is under appreciated is the consistency of
scores and this is another area where top RB's out perform top WR's.
Increasing our consistency relative to our competitors (i.e. lowering our variance) was one of the main arguments, based on evidence, of this article. The fact that top RBs were consistent when you measure them at the end of the season is not surprising. It's tautological. They would not have become top RBs were they inconsistent. The problem is we don't know which RBs will be the top scoring, consistent ones prior to the season.
Thanks for the comments guys!
#8 By: Alvaro, May 13th, 2017 05:04
I'm from Spain, so excuse my english. I love zeroRB and used it with enough succes in RSO and FFPC classic draft format for some years. Now I'm just playing best ball in FFPC and as much as I love the Siegele theory I don't play it. Last year I used a strong RB approach because I thought there were just 6-7 RB with three down upside and dozens of WR in the second half of the draft who were able to light it up on any given sunday.
So now I go something like this (with enough flexibility to snag a WR if It represents a great value): RB, RB, RB, RB, TE, TE.
If I were playing classic draft I would never use this strategy, but best ball format with 28 slots allow me to take at least 12 late WRs with enough volatility to fill every week the two starter WR slots. The RB and TE slots are filled with players from the 1-6 rounds and the flex spots are open for the competition between my other high drafted RB, my second premium TE or my volatility WRs.
I would love to know your opinions. Thank you Rotoviz writers for turning fantasy football into a strategic and philosophical fun game.
#9 By: devinmci, May 14th, 2017 03:20
This is a sensible strategy to take advantage of the best ball format, but zero RB will probably remain competitive with this strategy because RB RB RB RB TE TE is so fragile to injuries. Don't forget that David Johnson and LeVeon Bell both had season-ending injuries . . . they just came late in the year.
Injuries favored the top 3-down backs in 2016, but that probably won't happen again. Keep in mind the injuries that helped shape the seasons of the top RBs:
David Johnson (Chris Johnson injured)
Le'Veon Bell (DeAngelo Williams injured)
Ezekiel Elliott (Lance Dunbar injured, Darren McFadden injured)
LeSean McCoy (Karlos Williams cut right before the season, Mike Gilleslee started year with concussion)
Melvin Gordon (Danny Woodhead, Branden Oliver, Dexter McCluster all injured)
Demarco Murray (Dexter McCluster cut right before the season)
Devonta Freeman (Tevin Coleman missed 3 games)
Theo Riddick (Abdullah, Dwayne Washington injured)
Jordan Howard (Langford injured)
Jay Ajayi (Arian Foster injured)
Legarrette Blount (Dion Lewis injured)