#1 By: RotoViz, August 3rd, 2017 12:32
Originally published at: http://rotoviz.com/2017/08/how-to-win-the-flex-position-2017-edition/
Perhaps the most complicated position in fantasy football is not an actual position at all, but rather a confluence of multiple skill positions more commonly known as “the flex.” It’s something we’ve written a lot about at RotoViz, because the key to success is to win the flex position. Solving the flex problem usually comes…
#2 By: Sean Stern, August 3rd, 2017 13:30
some days I wish I was smart enough to understand all the nuances of these articles lol. So basically, what you're saying is a league with 3wr, 1rb, 1 te, 1 flex, should most likely grab 4 wr's and perhaps an elite te before a rb? More or less? maybe? perhaps? Send help
#3 By: RotoDoc, August 3rd, 2017 13:52
That's certainly a possibility, and is the strategy I will lean toward this year in PPR redraft leagues (if you want Half-PPR or standard leagues, there's an option to choose those in the app, and that might change your conclusions)
The goal of the article was to make people think about how they want to attack the flex this year. Ideally, we'd do so by looking at recent history (the past few years), as well as forecasting into the future, to make informed decisions.
I'm just walking you through my process and how I came to the conclusions I did. Play around with the app a bit and see what you come up with!
#4 By: Eric Gamble, August 3rd, 2017 14:11
Great read. Lots to think about but really helps solidify my draft strategy this year. Thanks Doc. Be interested to see your individual player equity scores. I appreciate the equity scores Denny Carter does using Rotoviz's simulator but unfortunately he is not doing RBs this year. Thanks again.
#5 By: RotoDoc, August 3rd, 2017 14:14
Yeah, some of us will do our individual rankings/equity scores in the coming days/weeks, myself included.
Thanks for the kind words!
#6 By: DJBees, August 3rd, 2017 14:48
Curious what this looks like in a superflex league
#7 By: RotoDoc, August 3rd, 2017 15:01
Yeah, I would have had to pull superflex ADP, and that's a bit more of a task. I'd love to tackle it if there's time.
#8 By: Dennis, August 3rd, 2017 16:01
Thanks Doc this is awesome...most of my leagues are still 2wr 2 rb 1 flex. curious how that would change your thinking if at all or much...im guessing racing to fill flex in PPR or hppr with high equity WRs is still a priority and you still may consider sprinking in TEs and QBs in the first 5 picks but curious your thought process ....i was really hoping to ask you about your reaction to Ben's piece, this was more than i even hoped for, thx
#9 By: ColdWarriors, August 3rd, 2017 17:50
tell me if I am wrong, but I think this article is making me feel very very good about my strategy this year to grab that one stud ppr RB (targeting Freeman) and hammering out WR for the next 4-5 picks to fill that flex while everyone else plays catch up on their rbs. I have two late keepers in Jimmy Graham and Martavis Bryant.
#10 By: Josh Aronovitch, August 3rd, 2017 21:48
I really enjoy the article and the concept of winning the flex. I (and many) play in leagues with multiple flexes, super flexes, and other unique features, like points per carry. In a current startup draft, for example, we start 1-2 qb, 2-6 rb, 2-6 wr, 1-5 TE, for a total of 10 starters. We get 1 PPR and .25 PPC. Am I wrong in thinking this format, with 4 flex spots (3 if you start 2 QB) the calculus changes? The elite multipurpose backs, and potentially elite get a big boost, because they can average 31-32 PPG if all goes right (2016) but peak WR scoring is only 27PPG, and in 2016 was only 20 PPG. That makes me think in that format, where you can start as few as 2 WR, the play is to load up on young high volume RB and taking the older ones that fall further than they should...
I like a David Johnson, Gurley, Mccaffery, Murray, QB, QB, TE, Decker, Garcon, Brown, lineup
better than an Antonio Brown, Hopkins, Arob, Parker, QB, QB, TE, Decker Powell, Riddick lineup...
But more generally, my feeling is that WR is so much deeper than RB this year...with mediocre RBs being overdrafted due to overreaction to 2016, isn't it best to jump on the truly elite early, since everywhere you take RBs will be negative equity?
#11 By: Daniel Schultz, August 4th, 2017 09:31
As much as I love Rotoviz and believe it is an invaluable tool for winning in fantasy, I think it oversells the zero running back, especially for std scoring leagues. At the risk of sticking my neck out and questioning Rotodoc and others who are far smarter then me, I'm going to do it anyway. When looking at the projections (lets use std scoring as an example), I couldn't help feel that something was off because the numbers for the projections looked too low and did not appear to value running backs enough. Ultimately, if projections are off, can we really use those numbers to make informed decisions?
My understanding of what Doc has done, is effectively take preseason positional rank (say RB1) and averaged (LOESS curve fitting to establish a number) the points over X number of years to come up with his projected points for that pick. For the sake of simplicity, lets just examine the top 5 RBs and WRs. Johnson, Bell, Elliot, Gordon, & McCoy are shown on Docs projections as 179, 176, 174, 172, & 169 (std scoring). The WRs are Brown, Beckum, Jones, Evans, & Green at 188, 183, 177, 172, & 167. My problem is these scores are too low, and significantly off more for the running backs. Using the Rotoviz Projecton Wizard, these RBs are projected for 278, 255, 217, 200,& 203. The WRs are projected at 209, 205, 198, 175, & 182. The difference between Doc #s and the wizard projections for the RBs averages 57 points but for the WRs, the difference is only 16. So Doc's projections are penalizing the RBs more the WRs. I know this is due to RBs higher injury rates and possibly less repeatability even if they remain healthy. Still, I know the top five RBs wont score only 179, 176, 174, 172 and 169. Reality is far more likely to be 278, 125, 64, 172 & 169. Obviously, the risk of the RBs is there, two guys have essentially screwed their season. The upside for the guys that hit, appears far greater for the RBs then when you hit on the WRs. Its the difference between swinging for a HR vs swinging for a hit or trying to win a 50/50 vs winning a GPP. I think when articles about the advantage for zero running back are written, the larger upside of the RBs, especially in std league scoring is not completely accounted for. Need to do some work now, but hoping to get some feedback on my thoughts.
#12 By: Blair Andrews, August 4th, 2017 10:44
There are a couple things going on here. First, if I'm understanding correctly, the projections here are based on ADP, not end-of-season ranks. So, for instance, the top RB by ADP in 2016 was Todd Gurley. Adrian Peterson was drafted fourth. Gurley underperformed badly and Peterson was a disaster. These kind of results pull down the average points scored at each draft position. Therefore you would expect these projections to be lower across the board.
The other thing to keep in mind is that part of what you're pointing out is the difference between predictive analytics and forecasting that RotoDoc mentioned. That is to say, we've never really seen two RBs like DJ and Bell who are good enough as receivers alone to be worth drafting as, say, a WR4. The amount of volume they're both getting in the receiving game is kind of unprecedented for a RB--certainly for two RBs to be doing it in the same season to the extent they are. So predictions based on past assumptions are going to be inaccurate for them--basically exactly Doc's point in making the PA/forecasting distinction. You can see that this is reflected in the staff projections, where the top two backs (DJ and Bell, respectively) are kind of in a class of their own, with almost 40 points separating Bell from the next player (Elliott).
And, FWIW, I don't know that anyone's ever said Zero RB is optimal in standard leagues (although it probably was in 2015). Certainly it can work but it's not nearly as dominant a strategy in leagues where you don't get some amount of points for receptions. I think DJ and Bell are great targets in standard leagues--and in PPR--but I also think RBs are being overdrafted, even in standard leagues, so there is some value to taking a WR or TE with your first few picks if you can't land one of either DJ or Bell (or, maybe, Elliott--but I prefer Brown, Beckham, Jones, even in std).
Also, I say you should definitely be questioning the things you read here. Best way for us all to learn.
#13 By: Ftw Wrestling, August 4th, 2017 10:45
Most articles on here, and The Godfather himself @FF_Contrarian, don't advocate zero rb for standard scoring leagues. Mostly effective in ppr with a flex where you can start 4 wr's
#14 By: Mike, August 4th, 2017 11:21
@coldwarriors Yeah, I'm thinking of using a similar strategy this year. Grab that stud RB early (Mccoy or Freeman if I miss out on the top guys), then load up on WR's and TE's.
#15 By: RotoDoc, August 4th, 2017 12:58
The problem isn't these points are too low, it's that you're not factoring in guys like Adrian Peterson and Jamaal Charles who scored 10 and 11 fantasy points respectively in standard scoring. And they went as RB4 and RB8.
Injuries happen, and that's what makes the values as they are.
#16 By: RotoDoc, August 4th, 2017 13:00
Exactly, I was using PPR as an example in this article.
Standard scoring is different.
#17 By: RotoDoc, August 4th, 2017 13:03
The projections aren't off
they are factoring in injury rates etc. The ADP RB1 can be expected to score 216 fantasy points, because things like injuries happen, and they happen at a far higher rate to RBs than WRs.
The end of season RB1 will undoubtedly score much more. But we don't know who that will be.
#18 By: Daniel Schultz, August 4th, 2017 14:03
Let me start by saying that calling the projections "off" was poor wording on my part (no offense was meant), so I should have stated the flex numbers are different than the numbers from the projection wizard. I am in a std scoring league with a flex, so my interest is in only trying to get a better understanding of the article and the numbers. What I'm still trying to understand, if accounting for injury rates gives the best analysis for the flex, would the best analysis for the projection wizard also take injuries into account in the same manner? I have your NASCAR subscription and for your projections there, you assume the driver always finishes a race which appears to be similar to the projections in the projection wizard. I would guess the difference has to due with frequency of not getting to the finish as to whether or not adding the DNF (or injury) data is important. I just want to stress, I'm not trying to be a PITA or confrontational, I know I'm out gunned, but I am hoping to understand the situation better.
#19 By: RotoDoc, August 4th, 2017 14:33
The main difference is, it's easier to project injury rates than DNF rates, which is why I don't include DNFs in my NASCAR projections.
But @friscojosh has done great work around positional injury rates, and there are statistically significant differences that are meaninful, whereas in NASCAR I can't come up with anything useful on predicting DNF rates, even with super sophisticated machine learning models.
The flex projections are of course different from the projection wizard. The flex projections strictly use past positional ADP to create projections, whereas the projection wizard is our rotoviz writers' projections conglomerated into one number (with a standard deviation).
In other words, the projection wizard, or my machine learning model last year, or josh's air yards model use more than just ADP.
The exercise here wasn't to say these are my projections for the flex ...because they aren't. The app is just there to give you an idea of positional equity. Think macro scale around positional trends, not micro scale around individual players.
#20 By: Josh Aronovitch, August 4th, 2017 15:17
These thoughts are still percolating but here goes:
What this App shows quite well is the "expected value" of drafting the RB1, or the WR1 based on historical data. That is a useful statistic. But it masks a huge amount of variation by collapsing a huge range of out comes into a single figure. I think the conclusions drawn based on this analysis ignores an important assymetry. Because we are all bad at predicting, the only thing we know for sure is that we will have hits and misses. What really matters is what we get if we hit and how much it costs us to take the shot. Whether it comes to stockpiling multiple first round rookie picks or picking as many high upside PPR plus rush volume RB as we can, there is a logic to taking multiple shots at the assets with the highest return if they hit. Zero RB in some contexts seems to suggest a playing not to lose mindset. A great example of this would be deciding between Gurley and a WR with similar ADP. We talk about things like "league winning upside" but I'm not sure we have a good way to quantify it yet.
next page →