"Dissecting QB Value In Fantasy Football -- The Zero QB Theorem"

[stevegrab]

keg,

 

Just ignore him like the rest of us. It only pollutes the thread by you quoting all of his nonsense.

[Axe Elf]

I did, you didn't just guess a number...

 

 

Ok, if it's too much work to find the axual numbers, I will present them here in isolation from any supporting evidence or rationale:

 

18-20

[darin3]

keg,

 

Just ignore him like the rest of us. It only pollutes the thread by you quoting all of his nonsense.

 

 

Actually, he's posed some interesting questions (IMHO) but keeps talking in circles and really can't back up the challenges.

 

Keg's done the research (and then some). The least you could ask for would be some hardline numbers to back up what you say if you're looking for chinks in the armor.

[keggerz]

Actually, he's posed some interesting questions (IMHO) but keeps talking in circles and really can't back up the challenges.

 

Keg's done the research (and then some). The least you could ask for would be some hardline numbers to back up what you say if you're looking for chinks in the armor.

 

I have no problem responding to anyone's questions...what I won't do and won't allow is this to devolve into some schtick for someone to pound their chest with.

[MTSuper7]

I read the whole article. I read a book on the mathematics of poker once, and reading things like CV reminded me of that book. I think you can definitely learn from studying the numbers from seasons past, leagues past, etc. and seeing what trends exist. But in poker, some people are successful because they are really good with the math and others are successful because they trust their gut/instincts/reads. Fantasy football isn't much different - I think you made a pretty good argument in favor of waiting to draft a QB based on what the numbers tell you. And the comments about elite QBs winning you a couple of games by themselves is valid as well, given that the majority of scoring systems will cause a QBs big game to mean more than a RB or WR going nuts.

 

For me, it comes back to one thing - the way to build a successful fantasy football team is to draft guys who outperform their draft spot (I know, earth shattering revelation there). Any time I've ever built a championship level team, it has always been made by the mid to late round gems that end up performing at QB1, RB1, WR1 type levels. Heck, the last championship I won in my main local (sadly, 2007), I started Sage Rosenfels in the playoffs. Yes, Sage Rosenfels. I played QBBC because I focused entirely on building that solid RB-WR corps and took a couple of mid and late round fliers on QBs I thought might break out. I was wrong, but it didn't matter.

 

It's hard to get a guy who outperforms his draft position when you take him in the first few rounds, and since you only have to field 1 starting QB but typically 2-3 RB/WR, it just flat out makes sense to ensure you get elite production at one of the more difficult positions to find good production outside of the first 5 rounds or so. Meanwhile, there is almost always decent value at QB in the mid rounds. I will insert one devil's advocate comment - if I draft an elite QB like Rodgers, I won't bother worrying about a QB2 for a long time. However, if I wait on QB and end up playing the carousel, don't I need to draft my QB2 a lot earlier? Or would this article advocate not even sweating it and just accepting the "Jake Locker" like production at the position which is offset by my strengths elsewhere.

 

I'll also argue that 3 leagues isn't that large of a sample size. Granted, I'm sure you don't have months on end to put together dozens of league comparisons. I liked the article though - it definitely put some metrics behind the idea of waiting on QB. People fall in love with the elite production from QBs, but consistency wins championships. Most of those people in 2007 who rode the Brady-Moss and Romo-Owens trains got burned by one single down week in December. I don't want that much of my team's production to have to rest on one guy personally, and stocking up on quality starting RBs and WRs spreads out my production more effectively.

[darin3]

I have no problem responding to anyone's questions...what I won't do and won't allow is this to devolve into some schtick for someone to pound their chest with.

 

 

Oh I hear ya and completely agree. Unfortunately (despite the one decent "what if"), this guy is 100% schtick and quite frankly, I'm surprised he still has the gall to show his mug around here after what he pulled.

[keggerz]

I read the whole article. I read a book on the mathematics of poker once, and reading things like CV reminded me of that book. I think you can definitely learn from studying the numbers from seasons past, leagues past, etc. and seeing what trends exist. But in poker, some people are successful because they are really good with the math and others are successful because they trust their gut/instincts/reads. Fantasy football isn't much different - I think you made a pretty good argument in favor of waiting to draft a QB based on what the numbers tell you. And the comments about elite QBs winning you a couple of games by themselves is valid as well, given that the majority of scoring systems will cause a QBs big game to mean more than a RB or WR going nuts.

 

For me, it comes back to one thing - the way to build a successful fantasy football team is to draft guys who outperform their draft spot (I know, earth shattering revelation there). Any time I've ever built a championship level team, it has always been made by the mid to late round gems that end up performing at QB1, RB1, WR1 type levels. Heck, the last championship I won in my main local (sadly, 2007), I started Sage Rosenfels in the playoffs. Yes, Sage Rosenfels. I played QBBC because I focused entirely on building that solid RB-WR corps and took a couple of mid and late round fliers on QBs I thought might break out. I was wrong, but it didn't matter.

 

It's hard to get a guy who outperforms his draft position when you take him in the first few rounds, and since you only have to field 1 starting QB but typically 2-3 RB/WR, it just flat out makes sense to ensure you get elite production at one of the more difficult positions to find good production outside of the first 5 rounds or so. Meanwhile, there is almost always decent value at QB in the mid rounds. I will insert one devil's advocate comment - if I draft an elite QB like Rodgers, I won't bother worrying about a QB2 for a long time. However, if I wait on QB and end up playing the carousel, don't I need to draft my QB2 a lot earlier? Or would this article advocate not even sweating it and just accepting the "Jake Locker" like production at the position which is offset by my strengths elsewhere.

 

I'll also argue that 3 leagues isn't that large of a sample size. Granted, I'm sure you don't have months on end to put together dozens of league comparisons. I liked the article though - it definitely put some metrics behind the idea of waiting on QB. People fall in love with the elite production from QBs, but consistency wins championships. Most of those people in 2007 who rode the Brady-Moss and Romo-Owens trains got burned by one single down week in December. I don't want that much of my team's production to have to rest on one guy personally, and stocking up on quality starting RBs and WRs spreads out my production more effectively.

 

3 leagues but 1805 games isn't that large of a sample size?...Compared to all the games played in fantasy it isn't but it isn't a small sample..and you know what, I was going to ask people to go to their leagues and do what I did...look at who won, zero out the QBs points and see if they still won or not...if they didn't log the amount the loss was by...send it to me (I'm setting up an email account for this) and I'll add everyone's added research to mine...I'll also ask that a link to the league be included so that I can do some spot checks to ensure the research being turned in is valid. Edited July 18, 2013 by keggerz

[MTSuper7]

3 leagues but 1805 games isn't that large of a sample size?...Compared to all the games played in fantasy it isn't but it isn't a small sample..and you know what, I was going to ask people to go to their leagues and do what I did...look at who won, zero out the QBs points and see if they still won or not...if they didn't log the amount the loss was by...send it to me (I'm setting up an email account for this) and I'll add everyone's added research to mine...I'll also ask that a link to the league be included so that I can do some spot checks to ensure the research being turned in is valid.

 

 

Yeah, it's not so much the number of games but the scoring system range perhaps. And also three leagues assumes most of the same ownership for those games, meaning owner's trends might unfairly affect things. I don't think it's a huge deal, honestly. I think your data supported your argument. Just an observation.

 

And BTW, I feel like I have to proofread my posts now for fear of Darin the Grammarian singling me out.

[darin3]

Yeah, it's not so much the number of games but the scoring system range perhaps. And also three leagues assumes most of the same ownership for those games, meaning owner's trends might unfairly affect things. I don't think it's a huge deal, honestly. I think your data supported your argument. Just an observation.

 

And BTW, I feel like I have to proofread my posts now for fear of Darin the Grammarian singling me out.

 

 

 

It's really only the stuff any third grader should know that I'll call you out on.

[ShadowsFall]

The question I have to axe is how it only says 618 posts under Baby Jane's Profile Picture?

 

Is that a weekly total?

[Axe Elf]

I was going to ask people to go to their leagues and do what I did...look at who won, zero out the QBs points and see if they still won or not...if they didn't log the amount the loss was by...send it to me (I'm setting up an email account for this) and I'll add everyone's added research to mine...I'll also ask that a link to the league be included so that I can do some spot checks to ensure the research being turned in is valid.

 

 

I din't log the amount the losses were by, but in an attempt to kill some time here at work, I did this for my work league last year:

 

http://games.espn.go.com/ffl/leagueoffice?leagueId=676265&teamId=3&seasonId=2012

 

There were 96 games played (16 weeks x 6 games per week), and 2 of them ended in ties, so they were not scored (since there was no "winner" for which to zero the QB). I did include four or five games in which one of the teams had quit fielding a lineup and therefore lost by 70 points or so, which kinda skews the data--but skews it in your favor, so you can have those. I suppose that is balanced out by the two games in which zeroing the QB of the winner would have resulted in a tie, which I had to score as a "miss" (since the winning team was no longer the winning team).

 

Anyway, of the original 94 game winners, 50 of them would have still won after zeroing their QB--53%.

 

So in this league, you'd give up about half of your wins by going without a QB, whereas in the data you cited, you would only have to give up about 40% of your wins by going without a QB.

 

I'm not particularly fond of either prospect, myself.

[keggerz]

 

 

I din't log the amount the losses were by, but in an attempt to kill some time here at work, I did this for my work league last year:

 

http://games.espn.go.com/ffl/leagueoffice?leagueId=676265&teamId=3&seasonId=2012

 

There were 96 games played (16 weeks x 6 games per week), and 2 of them ended in ties, so they were not scored (since there was no "winner" for which to zero the QB). I did include four or five games in which one of the teams had quit fielding a lineup and therefore lost by 70 points or so, which kinda skews the data--but skews it in your favor, so you can have those. I suppose that is balanced out by the two games in which zeroing the QB of the winner would have resulted in a tie, which I had to score as a "miss" (since the winning team was no longer the winning team).

 

Anyway, of the original 94 game winners, 50 of them would have still won after zeroing their QB--53%.

 

So in this league, you'd give up about half of your wins by going without a QB, whereas in the data you cited, you would only have to give up about 40% of your wins by going without a QB.

 

I'm not particularly fond of either prospect, myself.

 

1. Thank you for proving my theorem is true...53% does equal majority. Unfortunately you only did half he work...logging the amount the losses are by is critical.

 

As for a game ending in a tie, you handled it just as I did, a game that was backed out of the totals...when I zeroed out and it ended in a tie I counted that as a non-win...so essentially a loss

Edited July 19, 2013 by keggerz

[Axe Elf]

1. Thank you for proving my theorem is true...53% does equal majority.

 

 

Allrightythen. I din't realize that your theorem was intended to demonstrate how fantasy football team owners could halve their win total for the season--but I agree, it does that admirably!

[keggerz]

Allrightythen. I din't realize that your theorem was intended to demonstrate how fantasy football team owners could halve their win total for the season--but I agree, it does that admirably!

 

I'm not going to argue with you if you are just looking to argue...if you want to discuss the merits of the article then have at it. If you just don't agree and won't open your mind to it then move along.

 

As for you and anyone else that says a QB wins X number of games per season because they "go off" I give you this:

I looked at 1805 games

Those 1805 games would have all those go off games accounted for in the zeroing out process

If I zeroed out the QBs score...be it 35pts or 7pts if they lost I logged the point differential.

For the QB to have won the game on his own it would have created a loss of the "go off" number of 30 points or more.

Guess how many games in the 1805 there were that fit that criteria?

 

12 or 0.007%

 

The 12 is encompassing of ALL QBs that were started...so that doesn't even mean that those 12 games are attributed to the top QBs. With such a small percentage I have no need to dig any deeper.

 

So to give you an example: Aaron Rodgers scores 31 points and your team wins by 1 point. I zero out Rodgers 31 points and that gets logged as a 30 point loss.

[darin3]

Wow, I'm honestly pretty shocked by that number.

 

I'm starting to re-think my trade for Drew Brees

 

But still, this proves (with a glut of data) that you can wait on QB in redrafts. Not like "wait til the last rounds along with your kicker and defense" wait, but I'm definitely in the camp of "get your starting RBs, WRs and TE before going QB". I may just join a redraft this season to test 'er out.

[Axe Elf]

if you want to discuss the merits of the article then have at it.

 

 

That's what we've been doing. I never argue; I only educate.

 

As for you and anyone else that says a QB wins X number of games per season because they "go off" I give you this:

I looked at 1805 games

Those 1805 games would have all those go off games accounted for in the zeroing out process

If I zeroed out the QBs score...be it 35pts or 7pts if they lost I logged the point differential.

For the QB to have won the game on his own it would have created a loss of the "go off" number of 30 points or more.

Guess how many games in the 1805 there were that fit that criteria?

 

12 or 0.007%

 

 

So out of 1805 games, you only recorded 12 games in which a team won with a QB scoring 30 pts or more--and would not have won if their QB had not been scored. So that means that the vast majority of the time that your QB scores 30 pts or more, you win, and you win in a blowout. Call me crazy, but I think I'd rather have the 30 pt QB.

 

Again, I think the stat that we need here is how often a team wins when their QB scores over 30 compared to how often a team loses when their QB scores under 17 points--because that's really the difference you are looking at between a top-5 QB and a mid-10 QB in most weeks.

[darin3]

That's what we've been doing. I never argue; I only educate.

 

 

 

So out of 1805 games, you only recorded 12 games in which a team won with a QB scoring 30 pts or more--and would not have won if their QB had not been scored. So that means that the vast majority of the time that your QB scores 30 pts or more, you win, and you win in a blowout. Call me crazy, but I think I'd rather have the 30 pt QB.

 

Again, I think the stat that we need here is how often a team wins when their QB scores over 30 compared to how often a team loses when their QB scores under 17 points--because that's really the difference you are looking at between a top-5 QB and a mid-10 QB in most weeks.

 

 

I think we need that too. Why don't you go ahead and mine that data and get back to us.

[keggerz]

That's what we've been doing. I never argue; I only educate.

 

 

 

So out of 1805 games, you only recorded 12 games in which a team won with a QB scoring 30 pts or more--and would not have won if their QB had not been scored. So that means that the vast majority of the time that your QB scores 30 pts or more, you win, and you win in a blowout. Call me crazy, but I think I'd rather have the 30 pt QB.

 

Again, I think the stat that we need here is how often a team wins when their QB scores over 30 compared to how often a team loses when their QB scores under 17 points--because that's really the difference you are looking at between a top-5 QB and a mid-10 QB in most weeks.

 

No, because that doesn't match what you stated.

You stated the following:

because someone like Brees or Rodgers has the capability of throwing for 4-6 TDs a couple of times a year, thus winning your game for you all by themselves.

 

So since that is what yous stated that is what I gave you.

 

It doesn't matter if a QB scores 30+ points and they win by 50...in that scenario the QB didn't carry the team to a win all by their lonesome....so that number represents a game where a team would have lost IF their QB didn't score 30 or more points...you got what you asked for.

 

Here is a chart I put together with all the losses that occurred when a winning team had their QB points zeroed

out.

Edited July 19, 2013 by keggerz

[keggerz]

I think we need that too. Why don't you go ahead and mine that data and get back to us.

 

17 points isn't relative because it's a number that was pulled out of the air...the average loss was just a tad over 11 points...you can look at the chart and make your own decision on how many points you need from the QB to be successful...I also included some point breaks under the chart. Edited July 19, 2013 by keggerz

[Axe Elf]

It doesn't matter if a QB scores 30+ points and they win by 50...in that scenario the QB didn't carry the team to a win all by their lonesome....

 

 

If you want to be Captain Literal, the only time we can count a game as being won by the QB alone is when the QB alone scored more points than the entire opposing team. That's not what any of us have meant, either.

 

We're talking about the boost you get in your probability of winning when your QB scores 30 or more points, as opposed to the average score you get from the average QB on the average week. But this really only relates to part of your article--the part that dissects the value of an elite QB vs. an average or even substandard QB.

 

I really think the much larger flaw lies in that which Fleming and I have both pointed out--zeroing out your QB reduces your win total to 61% of the wins you would have had WITH a QB--you don't win 61% of your games OVERALL without a QB.

 

So now it comes down to balancing out the costs and benefits. You cut your win total in half with no QB. That says that a QB is vitally important to the success of a fantasy football team. How many wins do you forfeit going with a #10-#20 QB? That will tell you how important it is to get just any ol' QB. How many wins do you forfeit going with a #6-#10 QB? That will tell you how important it is to get a good, if not elite QB. How many wins are you comfortable forfeiting by downgrading your QB and still feel confident that you can make up those wins by upgrading at other positions by waiting on the QB?

 

For me, the best answer is merely to take a top 5 QB in the 8th round--or at least a top 10 QB in the 13th round (this is why Tony Romo and Carson Palmer are the QBs to have this fantasy season). Then you've accomplished both of your goals.

[keggerz]

If you want to be Captain Literal,

 

Pot meet kettle. You got what you asked for....I don't see any reason at all to account for empty points. Your QB scoring X pts doesn't get your team all riled up so that they go out and dominate.

 

At this point I don't have any reason to engage you because you don't get, don't want to get it and when I disproved what you "thought" you want to deflect and move away from it.

 

If you have a strategy that works for you then I suggest you continue to follow that.

Edited July 19, 2013 by keggerz

[keggerz]

By zeroing out only the wins, the data is badly skewed. You're not saying zero QB teams will win 61% of their games, you are saying they still win 61% of the 50% of the games they originally won. If a team plays 13 games and wins 8, and you zero out their QB and get a 61% success rate, that's just 61% of 8, or ~4.8. That's not a 7.9 win season. So you are no longer measuring the value of an "elite" QB vs. the 16th QB, you are measuring the overall strength of the winning team at any moment and demonstrating that, if the rest of your team is good enough you can get by with any donkey. If you want to truly understand the Top 3 impact, do a data set where you zero out only their wins (not all games) and see what % of their games were still won once zeroed.

 

Curious about methodology - when you say Top 3 QB are you going with literally the first three drafted, or are you going with the top 3 as they finished the season statistically?

 

If would be more on-point to zero out the RB1, the WR1, and the TE against each data set and see what the Remaining Win % is. Do teams only 61% with a zero QB theorum but only 49% of a Zero RB1 theorum?

 

=====

I looked at winning teams and winning teams only...the number of games still stays the same...there were 1805 games which means there are 1805 winners (I threw out games that were tied, only like 3 or 4 total max). The reason I looked at winners was because I had a feeling that the QB spot wasn't as important as many thought it was. My gut wasn't good enough to just use the QB position which is why I went thru the metrics to see if QB should be used...the math supported it.

 

To further clarify, I am saying that if a team won their game that they would have still won 61% of those games with a zero at the QB spot (the QB position is the highest scoring position so if you used RB, WR or TE the numbers would likely be greater, but the value calculations didn't support using those positions).

 

As for your 8 wins being equal to ~4.8 you are correct...but you changed the equation in doing that...the Win % didn't change...if 61% of teams repeat a win with a zero at QB it isn't a stretch to surmise that a team could win 61% of their 13 game schedule with a zero at QB.

 

Methodology: I thought it was clear in the article that it was based on the Top-3 by ADP, if not sorry about that. The thought process of using ADP is because those 3 QBs were drafted with the expectation that they would be the top 3 at their position and in turn would give their owner an advantage. Using the EOS top-3 wouldn't be a fair representation of the value for taking a QB early.

 

As for the Zero out all positions part...like I've said, the QB position is the highest scoring position so I would assume that doing that exercise for the other positions would return a greater percent than the QB percent of 61...but again, when you look at the value metric and replaceability metric it isn't a prudent thing to do...another thing that I didn't hit on but is also a part of it is the number of games that non-QBs miss compared to QBs...Be it due to injury or being benched due to matchup...by part of it I mean why I felt that miss-rate and replaceability was important.

 

I like the CV analysis. This is close to how I do mine, and really illustrates the supply/demand imbalance between QB and RB/WR. I always thought it was crazy to start 1 QB and 2 RB when really each team in the NFL only had one "primary" running back. This is really the heart of the issue and why QB's get devalued. If the raw data is easily accessible, run it against a 2 QB starting model if you really wanna blow your own mind.

 

 

The CV really levels the playing field so that value can be compared among positions...it is a big part of the new value model (rV, real value) that I am creating to replace VBD.

 

I actually do plan to have someone help me pull some numbers on some 2QB leagues.

 

 

 

Miss rate is interesting. Any consideration for severity of miss? StDev of the miss, or comparing miss value to PPG differential?

 

 

Not yet, but yes it is something worth looking at...but getting this to where it is now has taken around 7 months worth of work.

 

 

If you have raw data you'd like to share I'd love a look at it.

 

Better yet would be a true Replacement Value analysis - how often was the (Top 3 QB score - replacement QB score) the key to a win/loss? In my homer league the Top 3 averaged 26.9 and the 13-15 QB's averaged 21.4. How many games were won by 5 or less points?

 

 

I have posted a couple of the charts on my blog (5-yr positional scoring averages and loss log)

As for my data, I have it across 5 different workbooks and around 35 different sheets

 

As for how many were won by 5 points or less you can refer to the loss log I just published on my blog...I listed what the win differential was for the 704 losses when a team that won had their QB points zeroed out...I did not record what the point total was if a team won with their QB points zeroed out.

 

I hope that helps and answers your questions...if not just let me know.

Edited July 19, 2013 by keggerz

[Chief Dick]

Statistics, Schmatistics. FF is 94% luck and 6% luck. Just like I've always said.

[bpwallace49]

World of Warcraft just got pwned by keggers

[flemingd]

Statistics, Schmatistics. FF is 94% luck and 6% luck. Just like I've always said.

 

 

What happened to 95% and 5%? Or 91% and 9%. R U really Baby Jane in disguise?????