What's new
Fantasy Football - Footballguys Forums

Welcome to Our Forums. Once you've registered and logged in, you're primed to talk football, among other topics, with the sharpest and most experienced fantasy players on the internet.

Utility Drafting (1 Viewer)

secretbonus

Footballguy
We've heard all kinds of draft theories over the years, but the one idea that has always stuck with me is a utility based approach. I will broaden this term "utility" to not only mean how you use each pick, waiver wire and potential trades in a season, but how that usage contributes to probability of victory within a game and how that win relates to probability of a championship as well as consolation prizes... An individual player's value is related to the probability that you will use that player towards producing a win. This is a really hard calculation since there are so many different situations you may use a player but we can run simplified calculations to illustrate the importance of depth.

Example problem 1: Given say a 30% chance of QB1 getting injured and 20% chance of QB2 getting injured, what's the chance your QB3 is used? You need BOTH QB1 and QB2 to get injured 0.3*0.2=.06 or 6%. But in order for you use your QB3 he actually needs BOTH QBs to be injured at the same time, or during the other's bye week AND also be better than any waiver wire options. I won't complicate the problem by calculating this here but suffice to say that the odds of you using a QB3 are quite low. Lots of people overestimate the value of a QB3 in injury substitution. There may be some utility in the form of unpredictability of QBs year to year and variance from week to week but that is another subject.

Example problem 2: given say a 40% chance of injury in a given year for your RB1 and 40% chance of injury for RB2, what is the chance you USE your RB3 aside from just bye week?

that means there is a 60% chance RB1 doesn't get hurt. In order to NOT use your RB3 you need BOTH RB1 not getting injured AND RB 2 getting injured*. That means that .6*.6=.36 there is a 36% chance that BOTH RB1&RB2 stay healthy throughout the entire year. This leaves a 64% chance that at least one of the two will get injured. You also will have 2 bye weeks to play (provided RB1&RB2 don't share the same bye week as each other or RB3) You can look at the average number of games given an injury to quantify how many games worth the RB3 is worth, and add in the bye weeks and compare it to alternatives to get an approximate utility value.

 *(More precisely you also need there not to be a better option via your RB4-RBx and no better option on the waiver wire, but we are simplifying)

The more you can add in complexity and model it accurately, the more you can improve the edge off intelligent drafting.

Clustering Your Picks - One of the hidden advantages of "Zero RB" isn't just "antifragility" and injuries as is touted, but it's clustered picks.

Though experiment Questions:

What are the odds that if your WR 3 is Cole Beasley that you'll start him over AJ Green or Jordy Nelson?

What are the odds that if your WR 3 is Demaryius Thomas that you'll start him over AJ Green or Jordy Nelson?

What are the odds that if your WR 3 is Cole Beasley that you'll start him over Kenny Britt or Kenny Stills?

What are the odds that if your WR 3 is Cole Beasley that you'll start him over AJ Green or Dez Bryant? 

The overall point is that you will almost NEVER use an WR 3 (except for injury purposes) if there is a huge difference in talent while you could certainly use a weak WR3 like Cole Beasley if you have very weak WR1 and WR2 and you could certainly use Demaryius Thomas when either AJ Green or Jordy Nelson have difficult matchups or low scoring games where their teams are expected to lead and run often. The last question is a trick question in that you actually get slightly less use of Cole Beasley over Dez and AJ Green due to the bye week, but there's slightly more value since if Dez gets hurt Cole Beasley will  get more volume or if you know Dez is going to be triple covered all game in a high scoring game then Cole may have some value exactly when you need it that may even distance himself from the waiver wire options.

Planning for week to week matchups

If we are being honest,there is a really good chance that the player you draft doesn't score exactly thee points per game predicted and score the exact same every week. This is also why clustering positions together works well. In other words, if a player is predicted to get 150 points in a year, he may score 120 points or he may score 180 points with a very small chance of scoring outside of the 120-180 range given no injury. But given he scores 160 points (10 points per game), those points will still usually vary quite a bit throughout the season and not be super predictable. But the way in which those points vary may be predictable enough to offer some small edge in making lineup changes due to schedules and matchups and game scripts if you build the right team. Perhaps one week you may be really confident he scores between 8 and 18 points and another may be somewhere between 3 and 13 points. The lower 3-13 range will offer games in which you can probably find a substitute in the waiver wire or through your bench.

Utility really can come into play with known and clearly defined injuries and suspensions and also in same team RB pairings (at a cost). For example, you can be really confident that McFadden will have substantially better games when Elliot is out (as of now that may be the first 6games depending on how much the appeal delays the process or perhaps gets him fewer games) and Charles Sims and Jacquizz Rodgers will also probably see a lot more touches the first 3 weeks when Doug Martin is suspended. Additionally if you draft BOTH Jamal Charles AND CJ Anderson... or BOTH Ware and Hunt or BOTH Ingram and Peterson (and keep an eye out for Kamara later on in the season) as say your RB 2 and RB3 or RB3 and RB4 Your team will likely get stronger if you can easily determine when to sub in one of these two players and can find a replacement in the starting lineup such as if:

1)Either player gets hurt(but not both) and the other becomes the workhorse back or at least gets an increase in volume and production for the games in which you will use him.

2)The time share is not split as evenly as predicted and if either player has a substantial workload throughout the year or at least in weeks you use that in a way that is predictable.

3)Both players spend time as clear cut #1 throughout the year with enough advanced knowledge before the game starts for there to be a clear edge in one or the other.

4)There is a clear gamescript type of situation where you can more easily figure out who is likely to be more successful. For instance, let's say there is a big RB that runs between the tackles and a pass catching RB like Isaiah Crowell and Duke Johnson. This may not be the ideal example but it is close enough to illustrate for now. If the Browns are underdogs vs a tough 7 man front Duke Johnson is the back to own. A smaller back can fit between the cracks and has a chance to out finnesse the opponent. If the Browns are clear favorites vs a weaker front 7 they will likely get an early lead and pound the rock and use Isaiah Crowell. The Patriots are much smarter about knowing who to use which usually makes their backfield a more ideal situation to stack but they have an uncertain backfield with lots of backs that could contribute right now. Mike Gilslalee, Rex Burkhead, James White, Dion Lewis,etc.

Stacking has downsides. It's a little more valuable when you have deep drafts. In deep drafts most people aren't going to use their 20th round picks very often unless it is a kicker or defense. The value in drafting stacks is you enhance the probability of using your later picks. This can be a double edged sword. Getting more usage out of late picks is only valuable to the degree by which it increases your points per game and chance of advancing and winning in the playoffs and championship. In other words, if you draft your first 5RBs with the same bye week you'll get more out of your RB 6 and RB7 but there really probably isn't much of an edge in doing that just to use your later pick vs drafting similarly ranked RBs with different bye weeks.

You probably don't usually want multiple stacks because that limits your value elsewhere unless you can do it in a way that increases the number of roster spots elsewhere. Some have suggested 2 different RBs from the same team and then a bye week filler (plus waiver wire). The advantage there is in deep drafts only using 5RBs frees up your roster to stack WRs, TEs and backup QBs...

Utility isn't everything - know your format

In a winner take all type of format, you may sometimes be okay with longshot drafts or even a fragile type of draft where you value based drafting mostly filling your starters and you are just betting virtually everyone stays healthy and produces as predicted and also is available during the playoffs. You definitely are hoping to hit the top QB overall since the years in which QBs hit they can distance themselves despite having a lot of variance year to year and being hard to predict. Collectively this strategy is pretty much a long shot since simultaneously your opponents may have neglected RB but grabbed a decent #1 late and say they also have Adrian Peterson break out at the same time Mark Ingram gets hurt and a single event like that could catypult their team ahead of yours. You need a huge number of events like this to not happen for a decent or below average team to become awesome. Or you need injuries to not last very long or impact everyone but you for this strategy to work. If you want a draft where you simply take the best projected player and fill all of the starters and plan backups for bye weeks and hope for good health you need to understand what you are hoping for and have some way to approximate the probability of things not going wrong. But in 12 teams only 1 can win so if you can win 10% of the time you are beating the average 1/12 and if you can win 20% you are crushing it. Trying to draft on any combination of longshots may in some cases be optimal and possibly forgoing ADP and projected rankings.

In other formats like where half or more than half make the playoffs that has a flat payout to everyone that makes the playoffs you may be better off with a more robust type of draft where you draft an RB with a good backup early like McCoy with plans to grab Jonathan Williams and maybe you end up getting Ty Montgomery who you also handcuff and you get a lot of WR depth and QB late since you can always find at least a decent QB for a given week for most weeks. It isn't too hard to get a total production at QB not much below anyone else if you just play QB week to week and same thing with Kicker and defense. Tight ends are a little tougher because the outlier years by Gronk, Gates, Gonzalaz over the year and the difference between the best few and the worst few make TE hard to keep up with but it's usually easier to give up points to have a more secure team with a higher floor and lots of quality depth at WR (and sometimes RB) that can be a little bit harder to replace. The goal is not sucking.

Some leagues pay a certain amount for most points in a given week. Depending on how much they leave for the playoffs and how they distribute it to #1 on the year you may be better off grabbing the QB on the same team as your WR or top TE. Potentially in these leagues depending on how they break it down you can be hit and miss and even miss the playoffs if you have Gronk and Brady and still get paid enough with say 4 highest score weeks and profit or at least win enough of your buy in back 90% of all years where you don't need to win it all or make the playoffs as often and so reaching slightly and stacking combos like this and betting on health of the two may even be optimal in some cases.

Know what you are betting on and how it connects with your goals and payouts of the league -

if you get a QB early you are betting heavily on the health of that QB where as a late QB is betting against the health of the top QB OR against the top QB doing as projected OR betting that the RB/WR you select there is a superstar that distances himself so much from the competition it doesn't matter.  If you get a single TE early it is really hard to replace him if he gets hurt unless there is a handcuff with a lot of potential (maybe Reed/Vernon Davis or Gronk/Allen) that at least has a non zero chance of breaking out given an injury. You can't really burn an early pick on 2 TEs unless you have a flex spot plus 1.5X PPR for the TE because the top RBs and WRs are usually too important so you are basically betting not only on the TE being a major difference maker that separates himself in PPG dramatically from competition but also that he stays healthy for enough games which is sort of a little bit hard for me to bet on with Gronk and Reed.

Putting it together

Although it is definitely more rare for a top WR to go down AND be replaced by some no name WR that fills the void and does almost as well or for a very late WR to have a major resurgence or breakout year or several breakout games right as you predict them and need them it does happen. A utility based draft needn't be a zero RB approach. It should take into account your ability to handicap injury probabilities and injury duration given injuries and how well the players filling the void do as well as account for a large degree of uncertainty and model an entire season that accounts all of this... and you should make the pick that maximizes your expected payout considering all probable and possible picks your opponents could make after your pick. That is a really complicated formula with so many variables and lots of room for error in modeling those variables.

Zero RB happens to be a decent strategy at extracting the value admidst the chaos and uncertainty with a strategy that is strongly weighted towards the later games since season ending injuries accumulate over the course of the year. It gives you a better chance at having several of those variables align but it isn't complete enough and limits other options that will be more optimal. I believe if everyone was doing zero RB there would be a huge advantage in maybe grabbing only one top WR, sniping several of the top players at every other position, using quality RB depth and RB flex and then getting several WRs late and trying to hit on them, (or also doing zero RB but only if you are the first to get each and every position). I can't imagine a team not having a really good chance at winning with at least one of the top top several WR like say Mike Evans, two top RB like Bell or Johnson ANd maybe you miss getting both but you get LeSean McCoy or Melvin Gordon or Freeman or Murray as RB2 and you also get Gronk, Rodgers and maybe a couple other young RBs like Dalvin Cook and Christian McCaffrey for depth/flex and maybe no real WR 2 but you make up for it with high volume of late upside guys that you can sub in week to week.

One other form of utility draft may be for those really good at projecting schedule difficulty by position and drafting to get off to a good start say the first 3-6 games, as well as create some sort of potential playoff edge and relying on trades (if you use them) and ignoring a huge chunk of the middle of the season in the draft since so many things will change once more data becomes available and injuries occur. Grabbing Ezekiel Elliot, Doug Martin and Martavius Bryant all on the same team is a big bet that these players will be healthy and available and not resting during your fantasy playoffs (such as if any of these 3 teams clinch first round bye 12 weeks in and decide to rest their stars) but is also a big bet that you can start off with a good enough start without them. If you can pair all of them together with players that you can project to have a major edge early and coordinate your draft well enough, you can draft players a round or two ahead of ADP, totally ignore the middle of the season aside from the benefit of having key players return and still come out way ahead, particularly if you can sell high and buy low after the first say 5 weeks, with contingency option of selling one or more of your players when they come back if you get off to a bad start. Ezekiel Elliot will slowly increase in value closer to the Leveon Bell and David Johnson range each s

For example, A draft in which you take 3 or 4RBs early (and maybe handcuff one of them late as opposed to worrying about getting a lot of RBs on your roster) may still allow you to get a lot of points at RB and mix in plenty of usage from those RB but your success will be heavily contingent upon you being able to either hit at WR or blend in WRs at the right times to make up for the points you lost (and other positions). In such a draft you need top WRs to get TE hurt to hurt your opponents or to bust and score out of the top 15. You don't necessarily need your RBs to hit but you need to have 2RBs in a given week that you can get points out of and you need your WRs drafted later than your opponents to do well enough. 

A draft in which you neglect a single position like WR2 and try to stack lots of WRs 2,3,4,5 late hoping to either hit on a breakout player or piece together enough weeks to get WR2 production out of the combination of WRs is another way to do it.

A "zero QB, zero TE, Zero K, Zero D" type of approach where you use a late QB to get you week one production plus waivers to piece together QB/TE/K/D or all of the above by committee is usually a good idea.

 

Bronco Billy

Footballguy
Predict injuries accurately and draft accordingly.

Actually I don't know, that's as far as I got.  I appreciate the effort though.


It was thoughtful, but I'm wondering what kind of actuarial expertise you need to have to even remotely be able to credibly predict the probability of injury for any given player.  And if you can't get there, it kind of undermines his whole thesis.

 

Chaka

Footballguy
It was thoughtful, but I'm wondering what kind of actuarial expertise you need to have to even remotely be able to credibly predict the probability of injury for any given player.  And if you can't get there, it kind of undermines his whole thesis.
It may have been about something else but I haven't gotten that far yet. Maybe later.

 

Biabreakable

Footballguy
Hi secretbonus,

Outstanding first post and you definitely have some stream of consciousness going here. I tried to follow what you are thinking about here, and I think the subject is very broad. You describe a lot of variable situations and perhaps a method of using odds to determine a players value from that.

It might be helpful to the reader if you could break some of these scenarios down with illustrated examples of the point you are making in a shorter post. There is a lot you are describing here and looking at each one as a smaller piece may be better for others to understand and provide feedback.

I say this as someone rather long winded myself. What you have presented are several different but related topics that could be fleshed out individually before making those connections. As is it's a bit overwhelming and busy, everything at once.

Specific league rules, scoring, starting requirements, roster space are all factors that can change the utility of players. There are times when a QB 3 may be a better play than your QB 1 or QB 2 for example. Sometimes even predictably based on match ups or specific situations. Often times not predictably and you would not have started the player over a reasonably better option, yet the QB 3 outscores the other two that week just based on random chance.

I get the sense that through your use of odds and deductive reasoning that the QB 3 would be deemed worth next to nothing. While in some situations, they actually are worth more than you think. Have a career year. Lots of different things can happen.

For starters, I would like to discuss this aspect of utility then perhaps branch to the other topics related to this introduced as well. 

 

mbuehner

Footballguy
there's useful stuff here, but it's missing the importance of upside. nobody drafts a WR3 with the expectation of only replacement value. You generally draft because you expect the player to outperform what you're paying for them. The core of value based drafting if identifying relative value and exploiting it.

 

secretbonus

Footballguy
there's useful stuff here, but it's missing the importance of upside. nobody drafts a WR3 with the expectation of only replacement value. You generally draft because you expect the player to outperform what you're paying for them. The core of value based drafting if identifying relative value and exploiting it.
That's true. I suppose I didn't communicate that as well. Ultimately the probability you use a player and points you get if you use him also should include the probability that the player breaks out and fulfills an upside.

The idea is anyone you are drafting is not a number but a range of probable numbers with higher upside players usually containing a higher but less certain range (boom/bust) or possibility to play a large role is what allows you to account for uncertainty and manage it effectively.

 

Chaka

Footballguy
Sadly I can't get my leagues to change but I am guessing the vast majority of leagues are 1QB.
It's a shame really. I have never played in a 1 QB league and likely never will, I just can't get behind a system that values the most important player on the field at the same level as kickers, defenses & #4/5 WRs/RBs.  Bizarre IMO.

 

Hawkeye21

Footballguy
It's a shame really. I have never played in a 1 QB league and likely never will, I just can't get behind a system that values the most important player on the field at the same level as kickers, defenses & #4/5 WRs/RBs.  Bizarre IMO.
I enjoy the one super flex league I'm in but I still prefer 1 QB leagues.  Just like I prefer PPR and a 2RB, 3WR, 1Flex roster.

 

Deamon

Footballguy
It's a shame really. I have never played in a 1 QB league and likely never will, I just can't get behind a system that values the most important player on the field at the same level as kickers, defenses & #4/5 WRs/RBs.  Bizarre IMO.
It's called fantasy football... it's not supposed to translate exactly to real life football, or you'd give points for touchdowns, winning a coach's challenge, QB winning the football game, etc.  2 QB leagues are for rookies.

 

Hot Sauce Guy

Footballguy
The chances of me using my WR3 are very strong since my league starts 3, has an O-Flex, and there are BYE weeks to account for. 

I read about that far and my eyeball went all wonky - but I applaud the effort. It's way deeper than perhaps I'm capable. 

 

BoltBacker

Footballguy
I enjoy the one super flex league I'm in but I still prefer 1 QB leagues.  Just like I prefer PPR and a 2RB, 3WR, 1Flex roster.
One day when you are feeling really wild you should try french vanilla ice cream. Sure it's more scary than just plain old vanilla, but no risk no reward brah.

 

BoltBacker

Footballguy
It's called fantasy football... it's not supposed to translate exactly to real life football, or you'd give points for touchdowns, winning a coach's challenge, QB winning the football game, etc.  2 QB leagues are for rookies.
True. My third level orc usually kills my opponent's QB by the second dungeon anyway. Leagues where your starting QB makes a difference are so lame.

 

Chaka

Footballguy
It's called fantasy football... it's not supposed to translate exactly to real life football, or you'd give points for touchdowns, winning a coach's challenge, QB winning the football game, etc.  2 QB leagues are for rookies.
:lmao:  <-- for the rookie comment. (I am presuming you are being funny and not snarky with that one)

It's great that you like your type of league, I mean that sincerely. You're right about it being fantasy football which is why I always find it surprising that people cling to the notion that you should only have one starting QB. 

However, if you were serious about the rookie comment then I disagree with the premise that adding viable starting positions makes the game easier somehow. By definition it adds complexity, just like PPR, starting 3WRs, dedicated TEs, but to an even greater degree than those examples. It forces you to work harder to be better than your opponents.  If Aaron Rodgers goes down and you can just pick up Eli Manning or Matthew Stafford off the WW, that's easy but when your options are Paxton Lynch and Cody Kessler then things become much more complicated for you. You typically can't pick up players of Manning's or Stafford's quality at other positions, so why have that for QBs? It entirely changes your approach to drafting, waiver wire moves and trading.

I love it, you don't it's all good.

But my way is clearly superior and that's all good too. ;)

 

Chaka

Footballguy
One day when you are feeling really wild you should try french vanilla ice cream. Sure it's more scary than just plain old vanilla, but no risk no reward brah.
:lmao:

Easy killer, funny but let's be respectful. Over the years I have found that this is a hot button issue that can trigger truly passionate responses. I personally don't want this to turn into a snark fest when there are sincere arguments to be made from both perspectives.

Obviously our perspective is superior but that doesn't mean we should dismiss other perspectives entirely.  That is, unless you are a strict two QB guy and not a super-flex guy because I hate two QB guys and never listen to anything they say.

:D

 

shadrap

Footballguy
appreciate the post, secret------but a lot of us just like pro football & do not have PHD's in physics, engineering, or fortran.  still appreciate the effort it took to post.

 

Snorkelson

Footballguy
:lmao:

Easy killer, funny but let's be respectful. Over the years I have found that this is a hot button issue that can trigger truly passionate responses. I personally don't want this to turn into a snark fest when there are sincere arguments to be made from both perspectives.

Obviously our perspective is superior but that doesn't mean we should dismiss other perspectives entirely.  That is, unless you are a strict two QB guy and not a super-flex guy because I hate two QB guys and never listen to anything they say.

:D
Disrespected by the ice cream comment? Dont post in the politics subforum....

 

Hawkeye21

Footballguy
One day when you are feeling really wild you should try french vanilla ice cream. Sure it's more scary than just plain old vanilla, but no risk no reward brah.
I already play in multiple leagues with different formats so I don't really need to try anything.  I can't help that I prefer one over the other.  I have one league that has a Super Flex, basically two QB.  A dynasty league.  A league that has IDP, by far my least favorite.  Some are standard scoring, some full PPR, some half PPR, different roster sizes, snake drafts, auction drafts and more.  Just like food I like trying it all but just like food I still have my favorites.

PS: Chocolate is my flavor of ice cream.

 

Users who are viewing this thread

Top