General QB Drafting Strategy (Response to Doug Drinen) (1 Viewer)

[fornov]

(I posted this on the Assistant Coach board, but I was told it fits here better. So I'm going to give it a try.)

I am writing this in response to the "Waiting on QB's" article by Doug Drinen, posted on 8/7. This is something I've been thinking about for a long time, but have never been able to come to a conclusion on. I know people usually draft QB1 "too early." But why is it too early? I understand the VBD principles and what you lose by passing on a RB. But there is a lot of value to having Peyton Manning on your team and not having to make a weekly decision on who to start at QB.

Here is what I mean by this: if I have Manning on my team, I'm starting him every week except one (his bye). I know that he'll be healthy all season (always has), and I don't have to worry about what his matchup looks like. If I have 2 middle of the road QB's, it presents a lot of problems.

For example, in one of my leagues last year, I had Aaron Brooks and Jake Plummer as my 2 QB's. They were the QB10 and QB 13 off the board (Palmer was QB11, BTW). The chart in Drinen's article tells me I should have expected around 312 points from this combo. I don't know what they would have given me if I would have followed his rules for starting one over the other, but I'm pretty sure that this case would have been less. What's worse is that I didn't follow this starting pattern - I played the matchups. And the matchups were almost always wrong.

I don't still have the data, but if I remember correctly, only twice did I play the more productive of the two QB's when they were both actually on the field (take out byes and the end of the season when Brooks was benched). This was extremely frustrating. After about 8 weeks, I felt like no matter who I picked to start, the other guy would do better, and we all know what it feels like to have a bench player go off for a huge game.

So my real question, which is probably unanswerable, is how do you quantify the value of a known, proven commodity? Within the limits of practicality, you know that Manning is going to be a stud this year, hands down. Conversely, there is a roughly 50% chance your RB1 is going to be a bust. It also is much more likely that a valuable RB will emerge from the undrafted scrap heap (Mike Anderson, Rudi Johnson, Ruben Droughns, Nick Goings, and Samkon Gado to name a few from various recent seasons) than a QB. If the adage "you can't win your league in the first round of the draft, but you can lose it" is true, it seems much more likely that you will lose it by drafting a RB than a guy like Manning.

Is this article refuting this statement? I have never had Manning on any of my teams, and I have had 20 teams over the past 7 years. I never draft him because I have always adhered to the VBD principles, and my average regular season finish is just below 3rd place (3.3), so it seems to be working out OK so far. But I'm very interested in alternative ways at looking at a problem, and I'd be curious to see what others think about this.

If this article tells me that I need to stick with what's been working, then I'll listen to that. Maybe what I need to do is continue to wait on my QB's, and then follow the starting pattern described in the article. That would lessen my frustration of starting the wrong guy. What does everyone think about this issue?

 

[bjabrad]

Your problem wasn't strategy as much as player evaluation at the QB position. Brooks and Plummer were both on my do not draft list last year. There were at least 10 other QB's that I had rated higher. I don't need Manning to win but I do need somebody that gets me a consistent performance and you can't say that about either Brooks or Plummer. Brady doesn't finish at the very top but he's incredibly consistent even against good defenses and that's why I want him on my fantasy team.

 

[(HULK)]

I, for one, like your line of thinking and would like to suscribe to your newsletter.

 

[JayMan]

Hi fornov

The only problem with VBD (and don't get me wrong - it is the best best to assign value to players) is that it looks at the overall fantasy points each player is suppose to score - for the entire season... while we all know, since most leagues are H2H, that is it the good consistent players that helps you win your league.

You need a WR that will get his usual 6/7 receptions for 80 yards and the occasional score... not the guys that gets a 2/30/0 on the first weekend and 8/143/2 the next - especially if they don't follow the "predetermined" matchups!... both guys might score the same amount of points at the end of the season (thus VBD is equal for both)... but the first one is far more valuable...

I think this is what is reflected in your posting - you have selected an extreme case (with P. Manning verus J. Plummer and A. Brooks) because Manning is the ultimate consistently good player (never misses a beat - usually apdat very well to matchups) - while Plummer and Brooks are known for their inconsistent play...

The excerpt below is from a post in a thread where someone was asking about Standard Deviation to evaluate consistent players... I think it corresponds directly to your issue...

This is why I think the future of FF analysis relies on the consistency rankings I mentioned below to match with VBD to really determine who would be on your roster to win your league - This is just my humble opinion, don't take it for granted, far from it!

--------------------

... and the other issue with using Standard Deviation (StDev) to find if a player is consistent or not is an incorrect way to look at this.

As you might know, StDev shows the dispersion to the mean from a dataset... but the influant variable here is the "mean"... For example, one player can score 2 points every week of the year - thus is StDev will be 0 - he's the most consistent FF player there is - but he's constantly poor!

On the other hand - it is a well known fact that we are looking for consistent players - i.e. guys that can score a good amount of point week-in week-out to help our teams win H2H games... and it is also proven that a group of "consistent good players" will beat a group of "inconsistent good players" over 70% of the time... (but as radballs mentioned - it is very difficult to predict the consistent players just by looking at historical data)...

What should be used instead to figure out if a player is consistent or not - is the percentage of games that he scores at least the number of points the "last starter" from his position scored on every week...

You can even cut into those percentages and state that if a guy was in the best 25% of the starter at his position for a given week - he had an "elite week"

(i.e. if you start 12QBs... then the best 3 QB performances on week1 gives an "elite" week - while the top12 gets a "starter" week)... gathering that information, by position, for every week of the year...

Depending on your scoring system - that % can vary obvisouly, but for example in '05 - the most consistent RBs were:

name----- elite --- starterAlexander - 12 ------ 14Johnson --- 10 ------ 12James ------ 8 ------ 12Barber ----- 9 ------ 11Tomlinson -- 8 ------ 11With that in mind - it can help you identity guys that can consistently score enough points to warrant a starting spot on your FF roster...My 2 cents

 

[fornov]

bjabrad,

I do like Brady, but he went as QB7 last year, 2 rounds before I took Brooks, and 5 rounds before I took Plummer. In this particular league, there is some weird anti-Jake Plummer vibe, so he is usually a good value.

On the other thread, another poster made the comment that the choices are not Manning vs. 2 guys late. You can also go with Brady, Hasselbeck, et. al., in rounds 4-5 or so. This way you get a good consistent performer, and you don't have to worry about the "who do I start?" question every week. The only problem with this league is QB's often go early, (4 by the 2nd round, 7 by the 4th) so the only real value is in the later rounds. I refuse to reach for a QB2-7 that early.

(HULK),

Thanks, I guess. I'm not really sure what that means.

 

[fornov]

JayMan,

This is exactly what I am talking about. You are correct that I picked an extreme example, but it was very pertinent to my experience as that was who I actually had last year! You can probably sense the frustration I felt.

I would love to read the thread you mentioned - where can I find it?

I'm new to the message boards, if you haven't figured it out already.

 

[wraith5]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way...

Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter).

Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

 

[JayMan]

JayMan,

This is exactly what I am talking about. You are correct that I picked an extreme example, but it was very pertinent to my experience as that was who I actually had last year! You can probably sense the frustration I felt.

I would love to read the thread you mentioned - where can I find it?

I'm new to the message boards, if you haven't figured it out already.

There it is...link

Welcome aboard!

 

[JayMan]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way... Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter). Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

I agree... except that I don't think StDev is the way to measure consistent players (you can see why I think that way above)... A guy scoring 2 little points every week is the most consistent there is according to StDev - but not exactly the guy you want on your team... consistency rankings are the way to go, in my humble opinion...

 

[The Jerk]

On the other thread, another poster made the comment that the choices are not Manning vs. 2 guys late. You can also go with Brady, Hasselbeck, et. al., in rounds 4-5 or so. This way you get a good consistent performer, and you don't have to worry about the "who do I start?" question every week. The only problem with this league is QB's often go early, (4 by the 2nd round, 7 by the 4th) so the only real value is in the later rounds. I refuse to reach for a QB2-7 that early.

Hi again, fornov.I mentioned the Brady/Hasselbeck option in the AC forum. To a certain extent, drafting a QB is a luxury and it comes with a price. You forfeit an advantage at another position to be stronger at QB. If you are unhappy enough with the idea of using a QBBC platoon, then you have to pay the price to avoid it.To me, every player -- and therefor every QB -- has a price. Once I know my draft slot, I project which QBs I would consider drafting in each round. Most of the time, the QB is off the board by the time that round is reached. However, once in awhile one of these players last long enough, and as long as I am happy enough with my other positions, I'll draft him. Last year, it was Brady with the 53rd overall pick.The best complete advice is to list what round you are comfortable drafting each of the top 6-7 QBs, i.e., that clearly do not fall under QBBC. Then, analyze the QB 8-15 grouping and find two guys you feel you can work with. Remember, it's not automatically QBBC at that point. One can emerge as the clear starter for most weeks.You're already on the right track. You are developing a comprehensive strategy and you understand the draft tendencies of your league. All that is left is to put it all together into a draft plan. Bottom line: assuming a top QB doesn't fall to your desired round, you either are willing to pay the price to avoid QBBC, or you are not, but the best way to handle this uncertainty is to prepare for both of them. Then when the draft comes, all you need to do is assess the value and which way you want to go.

 

[Men-in-Cleats]

If you want to evaluate whether drafting Manning has any value why not evaluate the teams from those 20 leagues that actually drafted Manning. If the Manning drafters have a lower average finish in your leagues then it could say one of a couple of things: either that drafting a QB early gives you a disadvantage by forcing you to reach at other positions or that the type of drafters that take a QB early on average don't draft as good of teams. If the Manning drafters finished higher than you on average then a couple of things could be true as well such as: you are correct and it doesn't put you at a disadvantage as advertised or the fact could be that you aren't as good as you think you are.

I will tell you one thing already and that is that the Manning owner has never won any conventional league I've been involved in.

 

[Doug Drinen]

fornov,

I think most of the concerns you're having are not addressed in the article.

One issue is that the article addresses a "typical QB1," and we're all mentally plugging in Manning there. Manning doesn't really resemble the typical QB1, which has included Warner, Culpepper, and McNabb at various times in the past five years. Manning is the only one of those guys who has never put up a real stinker of a season. And he's not likely to put one up this year either.

That certainly does have some value, but it's tough to put a finger on how much it is.

Should be an interesting thread, though.

 

[The Jerk]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way... Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter). Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

I agree... except that I don't think StDev is the way to measure consistent players (you can see why I think that way above)... A guy scoring 2 little points every week is the most consistent there is according to StDev - but not exactly the guy you want on your team... consistency rankings are the way to go, in my humble opinion...

We all want consistently high scorers. Very little standard deviation with high mean. Numbers generally don't work that way. I like your idea concerning "point threshold" consistency. This is more meaningful than pure standard deviation. However, I'm not too worried about pursuing the numbers because it is mostly hindsight with very little foresight (as radballs alluded to in the linked thread). FWIW, even if you had players with lower standard deviation vs. higher standard deviation, I'm not sure it translates into a better team.If the means are the same (and that's really the issue), then there are two factors to consider:1. If two more inconsistent players tend to go off on different weeks, then their combined mean is not deviating very greatly.2. Timing is everything, and if a given team is inconsistent, it can actually help win as many games as it costs, i.e., winning vs. better teams AND losing vs. weaker teams.It's all hypothetical, but the simple way to describe these factors are as a "cancellation effect" for lack of any better term. Add the fact that consistency itself is much better viewed in the rear-view mirror, and I think it's not that worthwhile to pursue consistency too deeply.

 

[jestertj]

For me, since there aren't as many QBs started as there are RB and WR, I prefer to wait and load up elsewhere. It's much easier picking a QB off of the wire if I completely strike out on the 3 I draft.

 

[fornov]

Men In Cleats,

That is a great idea to go back to see how the various Manning owners did. Unfortunately, I don't have complete results and draft rosters that go back that far. I have only been keeping track of my own results (selfishly, I know). I hope I did not come across as arrogant by listing my leagues and successes; I was only trying to demonstrate that my strategies seem to be working so far. And I do not think a Manning owner has won any of those leagues, either.

And yes, the fact that Manning is such a sure thing as a QB1 sets him apart from other QB1's of the recent past. There is never an absolute "sure thing" in anything in life, much less sports, but he is as close as you can get, for this argument's sake. Meanwhile, stud RB's bust routinely, and that is the problem. Yes, you can hedge your bets by going RB, RB, RB, but eventually you are losing value elsewhere.

The Jerk (BTW, I think your personality seems a bit underrated - you're not that much of a jerk),

I like your idea of listing QB's you'd be willing to take at each pick in the draft. That way you can decide if you are willing to pay for that luxury of having a non QBBC guy or not. Then, if not, I need to do better on my QBBC guys by 1. getting lucky and landing a stud, or 2. getting 2 or 3 guys who have good and/or complimentary schedules to plug and play throughout the year. Despite my disappointment with Plummer and Brooks, I can't help but think they will be value guys again this year, and then I'll be right back where I started! -ARGH-

The idea of consistency balanced with production seems very promising to me. Some sort of product of expected production and historic consistency might lead to better preseason rankings. I still need to read the other thread...

 

[wraith5]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way...

Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter).

Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

I agree... except that I don't think StDev is the way to measure consistent players (you can see why I think that way above)... A guy scoring 2 little points every week is the most consistent there is according to StDev - but not exactly the guy you want on your team... consistency rankings are the way to go, in my humble opinion...

That's a valid point - and would have to be figured into the equasion. Otherwise the top of the list would be all the guys who score ZERO points every week of the season. They are the most consistent, after all... Like "Jerk" said - Very little standard deviation with high mean. Take for example D. Mason - in my PPR league, he ranked #18 last year. But scored only three touchdowns. I have to go all the way to #36 (Engram) before I find another receiver with that few TD's. Now CLEARLY, "not scoring TD's" is not a typical trait I look for in a player. But the fact that Mason was #18 without them is an indicator of consistency - and that I like. Compared to #17, McCardell, who's total points were much more tied up in TD's in a couple of games.

The only thing I don't like about the system you describe is that it draws a line in the sand and creates a yes/no system - did they make the "minimum," or didn't they? It doesn't deal in the shades of gray and percentages that I like to see, that help differentiate one player from another who is closely ranked.

This actually touches on a problem I have with any set of season projections I see - how many PPG are being predicted? When we say RB x is going to have 160 points for the season, is that 10/game for 16 games, or 20/game for the 1/2 season he's predicted to play? That clearly (at least to me) has an effect on draft strategy. I end up working backwards to figure PPG and consistency (not StDev, though - I'm just not that smart, frankly) from overall projections.

 

[Warpig]

Hi fornov

The only problem with VBD (and don't get me wrong - it is the best best to assign value to players) is that it looks at the overall fantasy points each player is suppose to score - for the entire season... while we all know, since most leagues are H2H, that is it the good consistent players that helps you win your league.

You need a WR that will get his usual 6/7 receptions for 80 yards and the occasional score... not the guys that gets a 2/30/0 on the first weekend and 8/143/2 the next - especially if they don't follow the "predetermined" matchups!... both guys might score the same amount of points at the end of the season (thus VBD is equal for both)... but the first one is far more valuable...

I think this is what is reflected in your posting - you have selected an extreme case (with P. Manning verus J. Plummer and A. Brooks) because Manning is the ultimate consistently good player (never misses a beat - usually apdat very well to matchups) - while Plummer and Brooks are known for their inconsistent play...

The excerpt below is from a post in a thread where someone was asking about Standard Deviation to evaluate consistent players... I think it corresponds directly to your issue...

This is why I think the future of FF analysis relies on the consistency rankings I mentioned below to match with VBD to really determine who would be on your roster to win your league - This is just my humble opinion, don't take it for granted, far from it!

--------------------

... and the other issue with using Standard Deviation (StDev) to find if a player is consistent or not is an incorrect way to look at this.

As you might know, StDev shows the dispersion to the mean from a dataset... but the influant variable here is the "mean"... For example, one player can score 2 points every week of the year - thus is StDev will be 0 - he's the most consistent FF player there is - but he's constantly poor!

On the other hand - it is a well known fact that we are looking for consistent players - i.e. guys that can score a good amount of point week-in week-out to help our teams win H2H games... and it is also proven that a group of "consistent good players" will beat a group of "inconsistent good players" over 70% of the time... (but as radballs mentioned - it is very difficult to predict the consistent players just by looking at historical data)...

What should be used instead to figure out if a player is consistent or not - is the percentage of games that he scores at least the number of points the "last starter" from his position scored on every week...

You can even cut into those percentages and state that if a guy was in the best 25% of the starter at his position for a given week - he had an "elite week"

(i.e. if you start 12QBs... then the best 3 QB performances on week1 gives an "elite" week - while the top12 gets a "starter" week)... gathering that information, by position, for every week of the year...

Depending on your scoring system - that % can vary obvisouly, but for example in '05 - the most consistent RBs were:

name----- elite --- starterAlexander - 12 ------ 14Johnson --- 10 ------ 12James ------ 8 ------ 12Barber ----- 9 ------ 11Tomlinson -- 8 ------ 11With that in mind - it can help you identity guys that can consistently score enough points to warrant a starting spot on your FF roster...My 2 cents

HULK NO LIKE BIG WORDY NUMBERS! HULK SMASH!

Actually .

This is exactly how I won a 16 team dynasty in my first year (full 53 man rosters!). I put players in my lineup that scored consistent good points consistently...not necessarily stellar points but good points. My best players were Hasselbeck, Deuce (injury riddled), Mewelde Moore, and Marion Barber, Donald Driver, Joey Galloway, and whatever 3rd WR I could plug in. All of my guys were pretty much middle of the road players. But when they played, I knew I could depend on them for a certain amount of points. They may not have the gaudiest numbers at the end of the year like those players that would score 2 points, 10 points, 6 points, and then explode for 60 points in various weeks. The players that tend to explode at any given moment are the ones that have point totals that are misleading (as was mentioned above).

 

[fornov]

So......

I guess studies have shown that a given player's consistency from one year to the next is not predictable. In other words, consistency is not consistent. Hmm, that makes things a lot easier.

JayMan brings up the idea of adding up the number of times a player was "starter" quality or "elite" quality. (aside - we need a term or acronym for these stats) Really, this is giving us a rough measurement of a player's quality and consistency. This is useful for top players, but not as much for lower, backup types. But here we are talking about studs, so it applies to this conversation. My next question is then, is this reliable on a week to week basis? If StDev is not, I would guess that these measurements are not, but I don't have the means to test that hypothesis.

Thanks for all the great responses!

 

[rascal]

Didn't Doug do an article that pretty much stated it was impossible to determine from year to the next which players were going to put up consistent production?

 

[Warpig]

I wouldn't say that consistency is inconsistent...if you do your homework.

I like to average out their game point totals over a 3 year period. This gives me something to go buy. You can also see how they perform against a certain team, at home or away...yada, yada.

All I can say is that I'm more of a consistency guy. When in a crunch, I will go with a guy that has upside or prevents some value when I'm comfortable with what I already have.

 

[JayMan]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way...

Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter).

Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

I agree... except that I don't think StDev is the way to measure consistent players (you can see why I think that way above)... A guy scoring 2 little points every week is the most consistent there is according to StDev - but not exactly the guy you want on your team... consistency rankings are the way to go, in my humble opinion...

We all want consistently high scorers. Very little standard deviation with high mean. Numbers generally don't work that way. I like your idea concerning "point threshold" consistency. This is more meaningful than pure standard deviation. However, I'm not too worried about pursuing the numbers because it is mostly hindsight with very little foresight (as radballs alluded to in the linked thread). FWIW, even if you had players with lower standard deviation vs. higher standard deviation, I'm not sure it translates into a better team.If the means are the same (and that's really the issue), then there are two factors to consider:

1. If two more inconsistent players tend to go off on different weeks, then their combined mean is not deviating very greatly.

2. Timing is everything, and if a given team is inconsistent, it can actually help win as many games as it costs, i.e., winning vs. better teams AND losing vs. weaker teams.

It's all hypothetical, but the simple way to describe these factors are as a "cancellation effect" for lack of any better term. Add the fact that consistency itself is much better viewed in the rear-view mirror, and I think it's not that worthwhile to pursue consistency too deeply.

I agree with most of what you are saying - especially the unfortunate hindsight part of consistency and this is why I havn't loss any sleep on the subject either... but I think it's worth a shot - to try and figure out if there are players that remains consistent (weekly) from year to year - by using the "point threshold"...If one can figure out that Edgerrin James is consistently putting up starter or even elite weeks when playing after his bye week - or playing the second time agaisnt his divisions foes - I think it's worthwhile to look at it...

I know that you'll say that it's overanalysing - and you probably are right... but, what if?... (I'm not asking you to be my right hand on this one - I know you have better things to do!)

I had found an article by Doug (can't seem to find it on the net right now) that showed that - probably through Monte Carlo simulation - with equal means, a team consisting of "consistent" players would beat a team of "inconsistent" players (StDev consistency)... this is why I think it might be helpful to pursue this issue a bit (looking at "point thresold" consistency though)

Finally, concerning your "cancellation" effect... I'd rather have the two guys that score 10 points each week - rather than two that are scoring either 20 or 0 - 50% of the time... Why?

The first pair assure me of 20 points each week... while the second will give me either: 0 - 20 - 20 - 40 (and this is assuming I make the right call on who to start)... equal you say?... might be, might not be... since the 0 week obviously doesn't help you much - and the 40 week? other players might have carried the load and you had beaten your opponent already...

Obvisouly, this is all theoretical - if I knew on a particular week that Plaxico would get 9/143/2 - I wouldn't play Derrick who scores his usual 7/91/0... (and this is my evil plan - to figure this out! )

 

[The Jerk]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way...

Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter).

Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

I agree... except that I don't think StDev is the way to measure consistent players (you can see why I think that way above)... A guy scoring 2 little points every week is the most consistent there is according to StDev - but not exactly the guy you want on your team... consistency rankings are the way to go, in my humble opinion...

We all want consistently high scorers. Very little standard deviation with high mean. Numbers generally don't work that way. I like your idea concerning "point threshold" consistency. This is more meaningful than pure standard deviation. However, I'm not too worried about pursuing the numbers because it is mostly hindsight with very little foresight (as radballs alluded to in the linked thread). FWIW, even if you had players with lower standard deviation vs. higher standard deviation, I'm not sure it translates into a better team.If the means are the same (and that's really the issue), then there are two factors to consider:

1. If two more inconsistent players tend to go off on different weeks, then their combined mean is not deviating very greatly.

2. Timing is everything, and if a given team is inconsistent, it can actually help win as many games as it costs, i.e., winning vs. better teams AND losing vs. weaker teams.

It's all hypothetical, but the simple way to describe these factors are as a "cancellation effect" for lack of any better term. Add the fact that consistency itself is much better viewed in the rear-view mirror, and I think it's not that worthwhile to pursue consistency too deeply.

I agree with most of what you are saying - especially the unfortunate hindsight part of consistency and this is why I havn't loss any sleep on the subject either... but I think it's worth a shot - to try and figure out if there are players that remains consistent (weekly) from year to year - by using the "point threshold"...If one can figure out that Edgerrin James is consistently putting up starter or even elite weeks when playing after his bye week - or playing the second time agaisnt his divisions foes - I think it's worthwhile to look at it...

I know that you'll say that it's overanalysing - and you probably are right... but, what if?... (I'm not asking you to be my right hand on this one - I know you have better things to do!)

I had found an article by Doug (can't seem to find it on the net right now) that showed that - probably through Monte Carlo simulation - with equal means, a team consisting of "consistent" players would beat a team of "inconsistent" players (StDev consistency)... this is why I think it might be helpful to pursue this issue a bit (looking at "point thresold" consistency though)

Finally, concerning your "cancellation" effect... I'd rather have the two guys that score 10 points each week - rather than two that are scoring either 20 or 0 - 50% of the time... Why?

The first pair assure me of 20 points each week... while the second will give me either: 0 - 20 - 20 - 40 (and this is assuming I make the right call on who to start)... equal you say?... might be, might not be... since the 0 week obviously doesn't help you much - and the 40 week? other players might have carried the load and you had beaten your opponent already...

Obvisouly, this is all theoretical - if I knew on a particular week that Plaxico would get 9/143/2 - I wouldn't play Derrick who scores his usual 7/91/0... (and this is my evil plan - to figure this out! )

I applaud your enthusiasm. The best way to do this is to try to come up with your approach, then post the predictions for the 2006 season. Give it a shot. I'll be glad to look at this again when the season is complete. I believe that the more consistent team will win more games and championships in the long run, too. My point on inconsistent teams was that it can be better to be lucky than good, but over the long haul, consistency will win out. However, the best way to have a consistent TEAM is to simply have the better players.

 

[JayMan]

I like the concept of QBBC, but picking the "right" matchups can be difficult - and the rearview mirror can be painful when you miss the hot hand. In leagues where you don't have to select your starters and just get the highest point total for each position, it is obviously an easier and higher-reward strategy. But most leagues don't score that way...

Brooks, despite all the positive hype this year, is inconsistent and lack-luster. His FF value has a lot to do with "garbage time" stats. His concentration and heart are questionable at best. I've drank the Kool-aid one too many times with him. He is poison. (Yes, I drafted him last year, and YES, I'm bitter).

Standard Deviation is a model I'd love to see explored. I think it's key to getting guys who will help you win every week vs. the guys who will single-handedly win 3 weeks and disappear the rest of the time.

I agree... except that I don't think StDev is the way to measure consistent players (you can see why I think that way above)... A guy scoring 2 little points every week is the most consistent there is according to StDev - but not exactly the guy you want on your team... consistency rankings are the way to go, in my humble opinion...

That's a valid point - and would have to be figured into the equasion. Otherwise the top of the list would be all the guys who score ZERO points every week of the season. They are the most consistent, after all... Like "Jerk" said - Very little standard deviation with high mean. Take for example D. Mason - in my PPR league, he ranked #18 last year. But scored only three touchdowns. I have to go all the way to #36 (Engram) before I find another receiver with that few TD's. Now CLEARLY, "not scoring TD's" is not a typical trait I look for in a player. But the fact that Mason was #18 without them is an indicator of consistency - and that I like. Compared to #17, McCardell, who's total points were much more tied up in TD's in a couple of games.

The only thing I don't like about the system you describe is that it draws a line in the sand and creates a yes/no system - did they make the "minimum," or didn't they? It doesn't deal in the shades of gray and percentages that I like to see, that help differentiate one player from another who is closely ranked.

This actually touches on a problem I have with any set of season projections I see - how many PPG are being predicted? When we say RB x is going to have 160 points for the season, is that 10/game for 16 games, or 20/game for the 1/2 season he's predicted to play? That clearly (at least to me) has an effect on draft strategy. I end up working backwards to figure PPG and consistency (not StDev, though - I'm just not that smart, frankly) from overall projections.

This is a valid point... and it is probably unfair to backups - or guys that becomes starter after an injury to their team starters - they couldn't have put up "starter" numbers prior to the injury...I like your idea of percentage - to help differentiate closery ranked players - this is obvisouly something that has to be explored, in my opinion

Thanks for your input!

 

[rascal]

The data base to do this would have to be huge.

 

[JayMan]

I applaud your enthusiasm. The best way to do this is to try to come up with your approach, then post the predictions for the 2006 season. Give it a shot. I'll be glad to look at this again when the season is complete.

I believe that the more consistent team will win more games and championships in the long run, too. My point on inconsistent teams was that it can be better to be lucky than good, but over the long haul, consistency will win out. However, the best way to have a consistent TEAM is to simply have the better players.

That sums it up... now, where is the list of "the assured best players for '06" so I can grab them all?

 

[Liquid Tension]

So......I guess studies have shown that a given player's consistency from one year to the next is not predictable. In other words, consistency is not consistent. Hmm, that makes things a lot easier.JayMan brings up the idea of adding up the number of times a player was "starter" quality or "elite" quality. (aside - we need a term or acronym for these stats) Really, this is giving us a rough measurement of a player's quality and consistency. This is useful for top players, but not as much for lower, backup types. But here we are talking about studs, so it applies to this conversation. My next question is then, is this reliable on a week to week basis? If StDev is not, I would guess that these measurements are not, but I don't have the means to test that hypothesis.Thanks for all the great responses!

All I know is that I plug in my rules 6 pts a TD...and 1 pt for 20 for QB's into the VBD spreadsheet and it says that Manning is the 9th best value. There is NO question in my mind that this is accurate and I believe that based on the tiers Manning should probably be the 5th choice in the draft. I can't speak for years past, but this year there are too many close guys (RB's) from 5-15 and Manning is the much better option.

 

[Doug Drinen]

Didn't Doug do an article that pretty much stated it was impossible to determine from year to the next which players were going to put up consistent production?

No, I definitely have not done anything that made a conclusion that strong. But I have done a lot of playing around with the data and, while I have never published anything along these lines, all my investigations have suggested that past consistency is of no practical value in predicting future consistency. Let me briefly describe one such investigation. STEP 1: use regression to find a formula that maps overall seasonal stats to a predicted number of starter quality games. E.g. if you are a running back and you had 250 fantasy points, then you "should have" had 9 starter quality games. If you had 325 points, you should have had 12 starter quality games. Or whatever. I just made those up. STEP 2: find all guys over the last 10 years that have been over or under their projected number of starter quality games in Year NSTEP 3: check to see if the under guys stayed under the next year and if the over guys stayed over. ANSWER: in general, they didn't. CONCLUSION: If you have two guys who had the same overall totals in Year N, but one of them had more starter quality games, then you cannot conclude anything about which guy will have more starter quality games the next year.Now that's not exactly the same as saying that it's impossible to determine which players are going to be consistent. Maybe there are other things, aside from last year's consistency, team-related factors or the like, that can help you determine who will or won't be consistent in the future. Anecdotally (and yes, I know this doesn't mean anything), I can tell you that Isaac Bruce, Corey Dillon, and Marvin Harrison were all at one point considered poster children for the "I'll never draft him because he's too inconsistent" camp. Now they are considered steady but unspectacular options.

 

[JayMan]

Didn't Doug do an article that pretty much stated it was impossible to determine from year to the next which players were going to put up consistent production?

No, I definitely have not done anything that made a conclusion that strong. But I have done a lot of playing around with the data and, while I have never published anything along these lines, all my investigations have suggested that past consistency is of no practical value in predicting future consistency. Let me briefly describe one such investigation.

STEP 1: use regression to find a formula that maps overall seasonal stats to a predicted number of starter quality games. E.g. if you are a running back and you had 250 fantasy points, then you "should have" had 9 starter quality games. If you had 325 points, you should have had 12 starter quality games. Or whatever. I just made those up.

STEP 2: find all guys over the last 10 years that have been over or under their projected number of starter quality games in Year N

STEP 3: check to see if the under guys stayed under the next year and if the over guys stayed over.

ANSWER: in general, they didn't.

CONCLUSION: If you have two guys who had the same overall totals in Year N, but one of them had more starter quality games, then you cannot conclude anything about which guy will have more starter quality games the next year.

Now that's not exactly the same as saying that it's impossible to determine which players are going to be consistent. Maybe there are other things, aside from last year's consistency, team-related factors or the like, that can help you determine who will or won't be consistent in the future.

Anecdotally (and yes, I know this doesn't mean anything), I can tell you that Isaac Bruce, Corey Dillon, and Marvin Harrison were all at one point considered poster children for the "I'll never draft him because he's too inconsistent" camp. Now they are considered steady but unspectacular options.

And there you have it... Thank you Doug for bringing this up - I knew that you had to have looked at this issue in the past - and won't lose sleep thinking about a way to incorporate this into a cumbersome database to say the least...We cannot reject the null hypothesis - but it does not means that the alternate one is therefore true...

I can say that I must be one of your most loyal fans (must be my Masters in Applied Maths & Statistics speaking!), your many articles at http://www.pro-football-reference.com/ on the FF subject are a treat for a guy like me...

Case closed on consistency... since it would be inconsistent to say that one can consistently find consistencies in the FF world, for future use...

 

[fornov]

Doug,

Thank you very much for joining in on this discussion.

I guess your research pretty much answers the consistency debate - it seems the "consistent" and "streaky" labels have been mostly (completely?) exposed as myths.

Getting back to the original question, how do we account for the so-called "set-it-and-forget-it" production of having Manning on a team? If we go on the assumption that he will hands-down be a top-3 QB this year, isn't that better than past QB1's have provided (Warner, Culpepper, etc.)? And how much extra value does that give us, knowing that we can essentially carry just one QB all year, except his bye week? You touched on that early in your article, but didn't really explore it very far that I can tell. Freeing up a roster space for one more sleeper RB or WR helps, as does the extra time we can devote to the daily and weekly decisions regarding those other positions, knowing our QB is set.

I suppose I'm asking if we can treat Manning, specifically, as a separate case from this discussion, just for argument's sake for now. Is there a way we can re-run the data, using stats for QB1, QB2, and QB3 as the season actually played out, as opposed to the preseason QB1? I don't know if my question is clear or not, so let me try again.

I want to assume Manning will finish the year as QB1, QB2, or QB3. Can we see what we would get from him with this expectation, instead of what we would expect from his preseason ranking as QB1?

Thanks again.

 

[Doug Drinen]

I want to assume Manning will finish the year as QB1, QB2, or QB3. Can we see what we would get from him with this expectation, instead of what we would expect from his preseason ranking as QB1?

As currently constituted, the programs are just not flexible enough to make an assumption like that. Let's simplify it a little, though. Let's assume that Manning was *guaranteed* to be QB2 this year. Over the past six years, QB2 has ranged from about 50 points to about 150 points of VBD and has averaged roughly 100 points of VBD in hindsight after the season.

Now, in preparing some numbers for this article, I found that all first round picks at all positions over the last six years, not counting those that missed more than a game to injury, have averaged almost exactly 100 points. So the average for all first round picks, including the ones who were injured must be lower than 100.

So what does that mean? I think it means that, if Manning were guaranteed to finish the season QB2 (which is essentially the same as your assumption), he'd be worth at least as much as an average first round pick that stays healthy. That's pretty darn valuable. Just off the top of my head, I'd say he'd probably be worth taking at #4 or #5.

 

[finnref]

Consistency has been a subject I have been interested in the past couple of years having lost important games when my "stud" fails to score as per his usual game. Some analysis of the stats for my leagues led me to the conclusion that players that scored the most points also tended to be the most consistent. I posted on this last year. But I am not satisfied that I have been able to come up with any predictable trend.

Standard deviation though mathamatically correct is not helpful since we are not dealing with plain mathe but with people.

Several articles on this subject have been written by Matt Waldman and posted on

fftoday.com. He calls his analysis of consistency the Crank Score.

Check it out and add your comments.

 

[fornov]

I want to assume Manning will finish the year as QB1, QB2, or QB3. Can we see what we would get from him with this expectation, instead of what we would expect from his preseason ranking as QB1?

As currently constituted, the programs are just not flexible enough to make an assumption like that. Let's simplify it a little, though. Let's assume that Manning was *guaranteed* to be QB2 this year. Over the past six years, QB2 has ranged from about 50 points to about 150 points of VBD and has averaged roughly 100 points of VBD in hindsight after the season.

Now, in preparing some numbers for this article, I found that all first round picks at all positions over the last six years, not counting those that missed more than a game to injury, have averaged almost exactly 100 points. So the average for all first round picks, including the ones who were injured must be lower than 100.

So what does that mean? I think it means that, if Manning were guaranteed to finish the season QB2 (which is essentially the same as your assumption), he'd be worth at least as much as an average first round pick that stays healthy. That's pretty darn valuable. Just off the top of my head, I'd say he'd probably be worth taking at #4 or #5.

This is good. So if I convince myself that Manning will finish the season as QB2, I just need to determine how many RB's will finish with a VBD>100 to determine where it makes sense to draft him. I looked through the last 10 years, comparing where the QB2's VBD would have placed among the RB rankings. Here is what I found:'96 3

'97 5

'98 7

'99 3

'00 3

'01 7

'02 10

'03 13

'04 1

'05 12

That averages to 6.4 over the past 10 years, but 8.6 over the past 5 years. There are some interesting trends here. First, there is a steady rise from 3 to 13 from 2000 to 2003. 2004 had astronomical years by Culpepper and Manning, such that they both placed ahead of all the RB's. 2005 then had a return to previous levels.

Now when we look at these numbers, we have to decide what to make of them. Does the 10-year sample mean more than the past 5 years? Does that trend from 200-2003 suggest that QB value will continue to diminish over the near future? Was 2004 an anomaly? Is this something that will be cyclical, as are most things in sports, such that we will see a rise in QB value over the next few years?

These are difficult questions to answer. In my opinion, I think the trend from 2000-2003 is very important to take note of. If you go back even farther, the QB2 was an average of 4.3 from 1990-1995. It seems the historical norm is for the QB2 to be approximately equal to the RB4 or RB5. In my opinion, we are currently in the nadir of QB value, and will see an upward trend toward the end of the decade. I do not know how to predict when that upward trend will begin, and of course that is the key, isn't it?

My first insinct is to say that we are correct in valuing the current preseason QB1 (Manning) at around the 12-15 draft spot, based on where the actual QB2 has finished in the most recent seasons. But I still don't think the answer is even this simple. We still have not accounted for the wildly variant nature of the top RB's. I keep referring to Manning as a "sure thing," because throughout his carreer, he has been. We can't say that about the RB's in this discussion. If we want to start naming "sure thing" RB's, the list is Alexander, Tomlinson, and that's pretty much it. Even guys like Portis, Barber, R. Johnson, et. al. do not have the track record of excellence that Manning does. If Barber or Portis finishes as RB15 or RB11 like they have, or if they get hurt and put up a VBD of 0, that precludes them from joining the category of "a sure thing."

For other examples, look at Emmit Smith in '96, or Terrell Davis in '99, or Marshall Faulk in '02. In each of those years, those guys were hands down the number one pick in every fantasy football league. I would guess that there were very few teams that picked number one those years who went on to win their leagues. This is why RB's are so hard to place in the "sure thing" category.

Now, don't get me wrong, I'm not saying we need to take Manning #1 overall, or even in the top 5. In fact, I'm arguing this almost as a devil's advocate, but really to get information. Thoughts?

 

[The Jerk]

This is good. ...Now, don't get me wrong, I'm not saying we need to take Manning #1 overall, or even in the top 5. In fact, I'm arguing this almost as a devil's advocate, but really to get information. Thoughts?

If you really want to get into it, consider that 10 of the top 18 RBs drafted last year were busts to a considerable degree. I think this was an anomaly also, but it does underscore the potential for flopping on your first round pick. The problem is that no one intentionally picks a bust, and you're still more likely to pick a top 10 RB in the first round than any other round, then get a QB late. The combination of early RB and later QB still provides the best blueprint for consistent (that word again, although this time it refers to seasonal) FF success.