Jump to content


ZWK

Member Since 14 Jul 2008
Offline Last Active Today, 12:49 AM
*****

Posts I've Made

In Topic: Thinking Of Starting A Dynasty League Next Year

31 August 2015 - 03:34 PM

What if you had an unlimited taxi squad instead of an unlimited roster? Unlimited both in how many spots in has, and how long you can keep a player on the taxi squad.

 

You'd need some restrictions like: a player can't be put back on the taxi squad after he has been activated in your regular lineup, and only rookies taken in the rookie draft can be put on the taxi squad (or players from another team's taxi squad which you acquired through trade).


In Topic: ZWK's Dynasty Rankings (updated Aug 2015)

30 August 2015 - 06:58 PM

 

 

Regarding Stafford: I worked on my own value formula over the summer that's basically just another flavor of what ZWK is using. It's (PPG - Baseline PPG) * Games Played, but I'm pretty sure my baseline is much lower than his. Over the last three years, my baseline came out to QB21, QB22, and QB21 in terms of PPG average; the lower baseline is a huge help to guys who stay healthy every year like Matt Stafford.

 

And using my extremely-friendly methodology, and a relatively passer-friendly scoring system (1/20 passing, 5pts per TD), Matt Stafford was the 12th-most-valuable fantasy QB over the last three years. Behind Robert Griffin III. And trending down, to boot, (50.88, 44.64, and 10.72 points over baseline).

 

This is despite Stafford ranking top-5 in pass attempts each of those three seasons, (and despite 2012 shattering the NFL record for pass attempts by 36 attempts). Which means there's not a lot of upside, volume-wise. Very little room to grow, but plenty of room to shrink.

The VBD numbers I gave are just from PFR, since that is a nice easy-to-communicate default.

 

The VBD-like-thing that I actually use is also based on ppg, but I use a last-starter baseline. Baseline-level ppg is set at a level so that the number of games played by players at or above baseline is equal to the total number of starts needed at that position (which would be 12x17=204 games for QB, if I'm using full-season stats). Then I do the same: (PPG - Baseline PPG) * Games Played. That is equivalent to giving a player credit for baseline-level performance for each game that he misses, and then doing season-long VBD.)

 

There is a case for setting the baseline lower, since players who score slightly below the "worst starter" level have some use and will occasionally end up in your lineup ahead of even worse players. But there is also a case for setting baseline higher, since players who score slightly above the "worst starter" level sometimes stay on your bench behind even better players. The "baseline should be lower" issue (playing below-baseline players) happens more often than the "baseline should be higher" issue (benching above-baseline players). On the other hand, good fantasy teams face more of the "baseline should be higher" issue (benching above-baseline players), and fantasy points matter more when you have a good team since that's when you're in contention for a championship. So, on the whole, I think the baseline I'm using is pretty close to where I want it to be.

 

Ideally, I would use some nonlinear curve for the relationship between fpts and something-like-VBD instead of just subtracting off a single baseline. And I think I do something like that intuitively when I'm making roster decisions. For the number-crunching part of the process, the VBD-like-thing that I use seems good enough.

 

 

I wrote a couple of articles this offseason about how I'm calculating player value: 

http://subscribers.f...lueOverBaseline

http://subscribers.f...tadVBDBaselines

 

For non-subscribers, the first one just laid out the case for calculating VBD on a per-game basis and introduced the idea of splitting it into two forms- value over "worst starter" and value over "average starter".  The second one trolled through a bunch of MFL leagues to measure what qualified as actual, real-world "worst starter" production. That's how I got the QB21/22 baseline in PPG, because in actual MFL leagues, that's about what the least-productive QB team was producing. Think of it as "in the absolute worst case scenario, this is what you can expect to get at the position". "Average Starter" baseline was calculated similarly- by going to actual MFL leagues and recording what the actual average ppg at the position was across the entire league.

 

I think the two different baselines provide a useful snapshot of value over absolute worst-case scenario and of the actual advantage a player is giving you over your entire league. Especially if we go with the idea that "anyone under baseline scores as a zero", I think it's more justifiable with a baseline that represents true "replacement level". QB12 still provides some value, because at least you aren't forced to start Geno Smith. QB30 doesn't really provide value, though, because you could have always started Geno Smith instead.

 

(In terms of Value over Average Starter, Stafford hasn't produced anything since 2011. Of course, there are only 14 total QBs who gave *any* value over average starter over the last three years.)

 

Regardless of how you calculate it, the fact remains that Stafford has produced little value despite high volume and is trending down to boot.

 

 

That's good research.

 

If you're really serious about trying to upgrade VBD into something that comes closer to tracking what we actually care about, I do think you're going to have to go nonlinear.

 

An elite WR who scores 8 ppg above "baseline" gives you 8 extra fpts every time he's in your lineup (we'll say), and he'll be in your lineup all 15 times.

 

A decent WR who scores 2 ppg above "baseline" gives you 2 extra fpts every time he's in your lineup, and he might only be in your lineup 10 times.

 

The elite WR is adding 6x as many fpts to your squad (120 vs. 20), even though he's only scoring 4x as many points over baseline. And moving the baseline around can't fix this problem with the formula. The player value curve (relative to the players' ppg) is a lot flatter near the level that typically gets called "baseline" - there are a lot of players who are moderately useful options to plug in for a few games when you need them, but who probably won't be starting for you for most of the season.

 

For high-end players, a linear relationship makes sense: each additional fantasy point that they score does translate into 1 more fantasy point for your team, because they're in your lineup every week. But that's not true near baseline.


In Topic: ZWK's Dynasty Rankings (updated Aug 2015)

30 August 2015 - 02:33 PM

 

 

 

 

Marcus Wheaton at 85 and Bryant at 33 are upside down.  Wheaton has won the starting job and now Bryant has a four game PED suspension.  Big Ben has been praising Wheaton all off season and he could easily solidify his spot in the next four weeks.  Even if you think Bryant has more long term talent, you would have to drop him some and move Wheaton up into the mid-40s or 50s.

 
I posted these before the suspension news (now being reported as "multiple failed marijuana tests"), but I don't think that news changes things all that much. Bryant has tons of upside and was ridiculously productive when he was on the field last year - in addition to the TDs, he was 3rd in the NFL in yards per route run (behind only AJ Green & Demaryius Thomas). The 4-game suspension means very little for his dynasty value (though the risk of future off-the-field issues does hurt him enough to drop him down a bit).

 

After what we have seen with Gordon and Blackmon, I find your stance that multiple failed marijuana tests "means very little for his dynasty value" to be very surprising.

 

Bryant now carries a very significant risk that was not associated with him or priced into his value before his suspension was announced. If he fails another test, he will be suspended for a year, and we know he has failed multiple tests in the past year.

 

I meant that missing 4 games doesn't matter much. The concern about additional suspensions (or other trouble) in his future does matter, and is a reason to downgrade him. Probably to somewhere on the next tier, but I'm not sure how far down it. I'm trying to remember the details of the new marijuana policy - do they still do a yearlong ban?

 

 

Next suspension is 10 games. After that is a full year.

 

Also worth noting: the new Substance Abuse policy doubled the limits of detection, so testing positive is harder and the "second-hand" argument carries less weight. Also, under the new policy a 4-game suspension doesn't come until the 4th strike, which means Bryant has either failed four tests in the last 16 months, or else he entered the league in the program and then failed three tests in the last 16 months.

 

Either way, there is non-trivial cause for concern.

 

That does sound pretty serious. There is some major risk there for Bryant.

 

On the other hand, Bryant keeps flashing talent. And it's worth buying into risky guys with upside, for the same reason that it's worth packaging competent starters to buy studs: the top guys are what win you championships. (The same logic applies to Blackmon & Gordon.) So I think my post-suspension update to that part of my WR rankings looks like this:

 

Tr  Rk    Play                   (prev: tr,rk)

5    32    Julian Edelman (prev: 5,32)
5    33    Jeremy Maclin (prev: 6,42)
5    34    Michael Floyd (prev: 5,34)
6    35    Devin Funchess (prev: 6,35)
6    36    Martavis Bryant (prev: 5,33)
6    37    Donte Moncrief (prev: 6,36)
 

(with Maclin rising because he has shown some rapport with Alex Smith in the preseason.) Even though 3 spots seems like too small a drop in the rankings for this news, I still think I'd rather have Bryant than guys like Moncrief, Latimer, Wright, or Andre Johnson.


In Topic: ZWK's Dynasty Rankings (updated Aug 2015)

30 August 2015 - 01:22 AM

Regarding Stafford: I worked on my own value formula over the summer that's basically just another flavor of what ZWK is using. It's (PPG - Baseline PPG) * Games Played, but I'm pretty sure my baseline is much lower than his. Over the last three years, my baseline came out to QB21, QB22, and QB21 in terms of PPG average; the lower baseline is a huge help to guys who stay healthy every year like Matt Stafford.

 

And using my extremely-friendly methodology, and a relatively passer-friendly scoring system (1/20 passing, 5pts per TD), Matt Stafford was the 12th-most-valuable fantasy QB over the last three years. Behind Robert Griffin III. And trending down, to boot, (50.88, 44.64, and 10.72 points over baseline).

 

This is despite Stafford ranking top-5 in pass attempts each of those three seasons, (and despite 2012 shattering the NFL record for pass attempts by 36 attempts). Which means there's not a lot of upside, volume-wise. Very little room to grow, but plenty of room to shrink.

The VBD numbers I gave are just from PFR, since that is a nice easy-to-communicate default.

 

The VBD-like-thing that I actually use is also based on ppg, but I use a last-starter baseline. Baseline-level ppg is set at a level so that the number of games played by players at or above baseline is equal to the total number of starts needed at that position (which would be 12x17=204 games for QB, if I'm using full-season stats). Then I do the same: (PPG - Baseline PPG) * Games Played. That is equivalent to giving a player credit for baseline-level performance for each game that he misses, and then doing season-long VBD.)

 

There is a case for setting the baseline lower, since players who score slightly below the "worst starter" level have some use and will occasionally end up in your lineup ahead of even worse players. But there is also a case for setting baseline higher, since players who score slightly above the "worst starter" level sometimes stay on your bench behind even better players. The "baseline should be lower" issue (playing below-baseline players) happens more often than the "baseline should be higher" issue (benching above-baseline players). On the other hand, good fantasy teams face more of the "baseline should be higher" issue (benching above-baseline players), and fantasy points matter more when you have a good team since that's when you're in contention for a championship. So, on the whole, I think the baseline I'm using is pretty close to where I want it to be.

 

Ideally, I would use some nonlinear curve for the relationship between fpts and something-like-VBD instead of just subtracting off a single baseline. And I think I do something like that intuitively when I'm making roster decisions. For the number-crunching part of the process, the VBD-like-thing that I use seems good enough.


In Topic: ZWK's Dynasty Rankings (updated Aug 2015)

28 August 2015 - 03:24 PM

 

 

 

Stafford, Kaepernick, and Tannehill have all been around for a few years (they're all 27), and don't look like top 5 QBs to me. Bridgewater has significantly worse draft pedigree than Winston or Mariota (the track record of late 1st rounders isn't nearly as good as the top few picks), and he was a below average QB as a rookie.

This is strange considering Stafford has had finishes of 5, 11, 7 & 16 the past 4 years.
Stafford is not a top 10 passer (in terms of NFL efficiency, not fantasy value). In terms of his track record of fantasy value, he had one huge season in 2011, which involved leading the league in passing attempts and a flukishly high number of TDs. Since then has not been a difference-maker, picking up 49 VBD over 3 seasons and averaging only half a fpt per game more than Kaepernick from 2012-2014. And he has had relatively favorable circumstances over the past 3 years - he is 2nd in the NFL in pass attempts over that period and has had Calvin Johnson at WR. My guess is that Stafford will continue to be a low-end fantasy starter as long as Calvin remains elite, and then he'll slip down into fantasy QB2 territory.
 
I've probably gotten more pushback in this thread on Stafford than on anyone else. I was saying similar things about Stafford 2 years ago, and the past 2 years have not given me reason to update in favor of Stafford.
Huh, the last 2 years have confirmed your stance? 2 years ago directly conflicts with your stance, he was QB7.

Last season was clearly a down year for him and there are many factors that led to it. Primarily the loss of Calvin and Det morphing into the #1 D in the NFL. This year however, Calvin is healthy and they lost arguably the best dlineman in the game as well as another pretty dominant DT. They added everyone's preseason darling, Ameer and also have a leading TE breakout candidate in Ebron.

I don't really care if Stafford puts up points because he throws a 1000 passes in a season. I only care about the points. In that regard, you have by and large been wrong about Stafford. He's had a top 5 finish,11 and 7.

 

Stafford has 24 VBD, in total, over the past 2 seasons. He has been 11th in fpt/g over that period. That seems pretty much in line with what I was guessing 2 years ago - slightly worse, actually.

 

 

How has he been wrong about Stafford?  A top 5 season does not necessarily equal a top 5 QB.  Stafford has been hugely dependent upon volume, and still hasn't been elite.  In exactly one season, which was 4 years ago, he provided an advantage at the position over some of your leaguemates.  I think ZWK nailed Stafford two years ago, and is 100% correct in his assessment of him today.  The only potential gripe I'd have about Stafford ranked as QB12 is ranking Brady in front of him, simply due to the fact that while Brady may play 5 more years, he's more likely to play 2-3 more.  And I'd probably take 10 years of Stafford over 2-3 of Brady unless my team was completely loaded.

I think that Stafford has 3-4 years left in his current form (with Calvin) as a low-end QB1, and then another 3-4 where he's basically just Joe Flacco. Whereas Brady has a good shot at another 1-2 top 5 seasons (at least in PPG).

 

Vereen seems low IMO, or maybe it is that Jennings is too high. Or both. I'm surprised to see Jennings a full tier ahead of Vereen, who is four years younger. Vereen could easily outproduce Jennings this season.

 

Remember what happened the last time New England's third down RB moved on. Woodhead went to the Chargers in 2013, and he finished as a top 15 RB. Vereen could easily do the same and is younger than Woodhead was at that time.

My take on the NYG running backs is that Jennings is a good all-around back, Andre Williams is not good, and Vereen is a relatively one-dimensional receiving back. The fantasy value of guys like Vereen is pretty dependent on their team's offense, and I don't get the sense that NY is the sort of team that is going to throw a ton to their RBs. It's possible that I should rank Vereen a little higher just because of the uncertainty about his role. I don't think I should put Jennings any lower - he's basically the last guy in my rankings who has a good shot to be a solid fantasy starter this year (except maybe Ivory, who I should move up after the recent good news about his role).