It's that time of year again, the most wonderful and magical time of year where food tastes better, the sting of a hangover is expunged from those 14 shots of bourbon, and dare I say, when women look even more beautiful. Yes my friends, it's the return of the NFL, and with it, fantasy football.
Now, it's probably safe to assume that I have a insalubriously elevated regard for fantasy football. So, many of the more well adjusted among us may find the following to be overwrought or, equally likely, a stupid waste of time. Both are correct. That being said, who cares, fantasy football is awesome!
With that out of the way, I'll share with you how I've been preparing for my fantasy draft: Copious, overly complex models.
As someone who has been playing for quite some time now, I've done a lot of modeling as it relates to fantasy football. However, the typical model I've built deals with projecting results for future games during the season, as opposed to pre-season, draft specific models. So, a lot of what I've been doing while preparing for the season is new terrain so to speak. Luckily, this is an entire website full of excel monkeys who are no strangers to models. If anyone on the internet has a strong and well-informed opinion on this sort of thing, it's the WSO community.
The Model & The Broncos
What I'm particularly interested in going into the draft is how off-season and NFL draft acquisitions will affect the performance of players on a given roster. Since one of my keepers is Peyton Manning (this may seem peculiar but we have very complex keeper rules and I assure you, this is terrific value in my league), and with an already strong (and tempting to draft) receiver corps, the addition of Wes Welker leaves some important questions to be answered:
- To what degree might the addition of Wes Welker help Peyton Manning? I think it's safe to assume that Peyton will not simply add the +100 receptions and +1,000 yards that Welker commonly attains in a given season with Brady.
- Whose receptions and yardage is Welker most likely to cannibalize on an absolute basis? I'm sure you've all heard the case for Welker cannibalizing Eric Decker, but is this the most reasonable? Stokley is now on the Ravens, and the team's tight ends, Tamme and Dreessen also seem like possible candidates, as do the receptions pulled in from others like running backs McGahee and Moreno.
- What will it look like if Welker fails to cannibalize a significant amount of receptions and yardage? In other words, how will Eric Decker, among others, fare if Welker's stats are the ones who suffer?
In order to answer these questions I've built a simple model to see how the addition of Wes Welker could play out. And, as a matter of course, we'll have to make some assumptions. These are the (hopefully reasonable) assumptions I've made:
- Peyton Manning is a fairly consistent quarterback historically, so it stands to reason that this will continue. So, we'll assume three potential scenarios for Manning's receptions/yards, a bear case (380/4400), a base case (400/4600), and a bull case (420/4800).
- It is unlikely that Wes Welker will cannibalize Demaryius Thomas given the differences in the way the two players line up. So, Thomas, like Manning, will have a bear (80/1225), base (95/1450), and bull (105/1600) scenario associated with him.
- Welker will absorb all of the receptions and yardage that Brandon Stokley attained during the previous season, which was roughly 11.50% of Manning's receptions and yards. Additionally, we'll have a bear (80/950), base (100/1150), and bull (120/1400) scenario for Welker.
- The relative distribution of receptions and yards - sans Thomas and Stokley - will remain constant and go unchanged from the previous year (this is something of a stretch, but it's a simple model, let me know if you see a better way to do this without getting too crazy).
The results of the model show a slightly more nuanced picture of the Bronco's passing game for the 2013-2014 NFL season. While it does appear that Eric Decker will take a hit, the results of my model run suggest that the hit won't necessarily be a death knell for his overall production:
The Results
Eric Decker (2012 - 85/1064)
Bull Case: 74/883
Base Case: 67/804
Bear Case: 61/704
Jacob Tamme (2012 - 52/555)
Bull Case: 45/461
Base Case: 41/419
Bear Case: 37/367
Joel Dreessen (2012 - 41/356)
Bull Case: 36/295
Base Case: 32/269
Bear Case: 29/236
Everyone Else (2012 - 83/706)
Bull Case: 72/586
Base Case: 66/533
Bear Case: 59/467
Note: The bull, base, and bear cases are averages across Manning's and Thomas's bull, base, and bear cases, and correspond to the bear, base, and bull (respectively) cases for Wes Welker. I can give you the ranges if you're interested.
On the contrary, under the assumption that a slot receiver like Welker has an equal chance of cannibalizing everyone who's not a true wide out (i.e. Demaryius Thomas), Decker appears to fare fairly well despite the appearance of the pro-bowler, Wes Welker. With these results in mind, I'm of the opinion that while Decker will take a hit, it'll be much smaller than what many may assume, making Decker a potential bargain during draft time.
Fellow fantasy football monkeys, what do you think? Who else gets in this deep with fantasy football? What do your models look like?
You seem to have overlooked the most important aspect for FF - TDs. QB completions are irrelevant to FF, and can really make or break a receiver, particularly in standard scoring leagues, and even more so when the 1,000 yards or so is the ceiling, i.e. Welker and Decker. There is no question that Thomas is poised for a top-5 finish.
Also, the modeling is nice and all, but FF is reliant on tremendous amounts of luck. You can run as many simulations as you want, but I believe in standard formats Decker and Welker are risky plays who will simply cannnibalize each others production in years past.
You're correct, but this was intentional. Getting into the issue of TDs is pretty complicated if you're interested in accuracy, at least from the perspective I took. That being said, I think you're underestimating what a receivers receptions/yards can tell you about scoring potential. Unlike with RBs, where the dreaded "vulture" exists, it's unlikely that a WR with few receptions and yards will put up many TDs (outside of, perhaps certain TE sets). So, in my opinion, projecting these stats can be instructive when ranking players pre-draft.
Definitely agree about luck, but running some simulations based on decent assumptions can give you some insight you wouldn't typically have. For instance, I think you're absolutely correct that Decker and Welker are risky plays due to the uncertainty of their production. However, this doesn't mean they're not worth picking up if you can get the right price for them. Certainly you wouldn't pass over either in the 10th round, but where does their intrinsic value actually lie? A good simulation can give you insight here.
Of course, take this all with a grain of salt. It's pretty obvious that I have a slightly slanted opinion regarding the usefulness of models in fantasy football :).
Do you guys do auction drafts? Snake drafts are terrible. How many leagues are we talking about? I did three last year and that was one too many.
Auction drafts are very awesome but come on, snake drafts aren't THAT bad! There's few things more exciting than seeing "1st pick" pop up during random selection!
Yeah, three is too many; I stick with two.
I agree the addition of Wes Welker is going to significantly hit Eric Deckers production this year and I doubt Decker is going to hit double digit TD numbers. At the moment on Yahoo Eric Decker is projected at 55 and Wes Welker is currently 66. I would pass on Decker in the early rounds and if he is there in the later rounds pick him up as a WR3.
Decker at 55 and Welker at 62? That's weird, most ADPs I've seen have the two switched.
http://fantasyfootballcalculator.com/adp.php
Has Welker at 44 and Decker at 65. Welker at 44 seems reasonable, but I think Decker at 65 is something of a steal.
I'm not a fan of snake drafts simply because of the enormous advantage it gives the 1-5 picks. I was the last pick, 14th to be exact, in a draft two years ago and got completely F'ed when my wifi went out. I ended up with two scub WRs. The main problem with snake drafts is that the elite crop of running backs drops off a cliff after the top 4 or 5 guys. The same can be said of QBs. I think everyone should have a shot at grabbing one of the elite position players which is why I prefer auction. If you want a player, you simply have to pay more than anyone else for them.
@mikesswimn I agree Yahoo has some pretty ridiculous ADPs. At 65 Decker would be a steal since all the good RB's will be gone and also the depth of TE and QB is huge this year.
@kingtut I kind of don't mind being last pick this year in a standard 12 team snake draft. You can pick up 2 solid Running backs in the first and second round. Then the late 3rd and early 4th should offer you solid WR's and maybe Peyton Manning.
Snake all the way! And 10-team leagues are where it's at. Any more and everyone's roster will likely have busts every week, where your flex is real shitty. And Welker's ADP of 44 is iffy, but Decker's at 65, now that really is a steal.
And as far as receptions/yards predicting TDs, I disagree. Last year, on one end, you have Calvin Johnson setting the NFL record for yards, fuck ton of catches, but...5 TDs? (or whatever it was) On the other end, you have James Jones, decent yards, decent receptions, but led the NFL in TDs at 16 (if I recall.) But anyways, can't wait for drafting to begin. I'll be in 3 leagues, and will be sure to post my teams in this thread when the time comes.
That's a compelling point, but I'd argue that those two are outliers (along with Decker and Hartline). If you look at the typical case of double digit TDs, most everyone has 80-120 catches and 1100-1500 yards (Julio had 79, but that dude is just crazy with the long pass). Perfect? No, but seems close enough for me.
So, yeah, what guides TDs is definitely a bit more complicated - no question there - I just run with receptions/yards because they seem "good enough" and I'm not clever enough to figure out something better, haha.
Auction is definitely the preferred format, especially if you're good because your edge increases immensely.
However, snake is fun for the "traditional" people, and if you allow pick trading, even better!
I read about how one league does auction and hired a professional auctioneer for the draft. Would love to try that.
Am I the only one that thinks the bull case for welker is far too much? He has been averaging about 10.2 yds per reception he doesnt reel off big plays that often. I am expecting a bull case of 100 rec/1150 yds. Base 90/1000, bear 80/850
On another note, what type of keeper do you play OP?
Last year Welker put up 118/1354 with Brady, and 122/1569 the year before. Granted, you're probably still right that it's a bit aggressive, but hey, that's what bull cases are for :).
We do a "N-1" keeper rule which runs for 3 seasons or until you keep someone in the 1st round (whichever is earliest). We can trade picks and, well, anything really so there's some arbitrage opportunities for keeper rounds (also keeping Muscle Hamster in the 3rd - super hype about that). Also, a big point to make is that (prepare to cringe) passing TDs are worth 6 points. So despite the over supply of good passers, QBs are overvalued compared to standard formats in our league.
I know this is an older thread but didnt want to create a new one regarding the same topic. Ive made a couple models. One is a lineup optimizer. It aggregates weekly projections from as many sites as possible and calculates mean and standard deviation. Then I use normal inverse function with a random probablity to get a possible score for each player. Run 1000 trials. Then I find the upper and lower quartiles for my trials. If im playing one of the best teams I run evolver to give me the best line up to max the upper quartile, anyone else I minimize my lower quartile.
Ive got a very similar model for auction drafts. I think its too many players for solver so you may need evolver. Again, maximize projected points while constraining your budget. Havent used the auction model yet as I recently made it.
Anyone have any other cool models or suggestions to improve mine? Also when gauging risk or variability I'm just looking at std dev among projections. I think week to week would be good as well to identify boom or bust guys like Torry Smith has been over the past couple years. The problem is some guys just dont have the sample size or their situation changes too much. Like if I modeled Jordan Reed, I think his last 3 games are the norm but should I include the beginning of the season when they were using fred davis?
how is this model working for you? has it been successful in terms of optimizing your line up? Also, how often does it give you a different line up than just averaging the projections and playing the top guys?
Try modeling a player's snap counts. I've been playing with one that does that and it handles increases/decreases in playing time much better for guys like Jordan Reed and (annoyingly) Kenbrell Thompkins.
I do have a model like the one you've described and what I like to do with it is to use the same method only instead of for actual players, I do it for the teams in my league. Sort of a big "projection" report of sorts. Just for fun mainly, but it makes for a good discussion and an even greater excuse for smack talking.
Dolores aperiam quam ratione assumenda. Rerum ipsam consequatur dolores culpa est. Voluptatem explicabo quisquam ipsa rerum dignissimos et recusandae.
Aliquid qui ullam est est vel assumenda dolorem. Porro esse inventore laborum sapiente. Qui blanditiis quia voluptatem ab nam.
Beatae ducimus inventore et consectetur numquam sit. Velit rem qui quos quis beatae consectetur. Omnis nihil aut dolores iure sint. Dolore maxime nam omnis ut nulla velit laboriosam. Magnam quos dicta necessitatibus aut omnis ut.
Est autem cumque ut. Quidem rerum quod consequatur. Aut eos corporis omnis nam. Unde et aliquam inventore molestiae. Similique qui sed in aut omnis suscipit fugiat.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...