Writers of Pro Football Prospectus 2008

19 Aug 2013

Gronk GP and Replacement

How many games is Gronk projected to play in the KUBIAK projections?

In order to properly value a player who is projected to miss games, he should receive replacement level value for the games he is projected to miss. I would like to make that manual adjustment, but I need to know how many games he is projected to miss.

Posted by: aron7awol on 19 Aug 2013

24 replies , Last at 27 Aug 2013, 12:26pm by jjjk

Re: Gronk GP and Replacement
by GBS :: Mon, 08/19/2013 - 3:55pm

I don't have access to the spreadsheet right now, but I believe it is in a comment attached to the cell with his name.

Re: Gronk GP and Replacement
by Aaron Schatz :: Mon, 08/19/2013 - 4:39pm

He is projected to play 10 games.

Re: Gronk GP and Replacement
by MJK :: Mon, 08/19/2013 - 8:59pm

No, I disagree that he should receive replacement level value for the games that he is projected to miss. If you draft him, he's eating up a roster spot that you could more valuably use on someone else. So it's not just like "draft Gronk and you'll have a replacement level TE for six games"...it's "draft Gronk and you have a replacement level TE for six games AND be down a roster spot for those same six games". Depending on the size of your bench in your league, that could be a big deal, especially if you have another good player that ends up injured in that time, or face a lot of byes towards the end of that six weeks.

Re: Gronk GP and Replacement
by aron7awol :: Mon, 08/19/2013 - 9:39pm

MJK, I acknowledge that you make some good points. You're right, it's not as simple as that. However, while making no adjustment at all grossly underestimates his value, making my adjustment only slightly overestimates his value. Since even in shallow leagues the last bench spot is usually used on a riskier player with upside, having that spot taken by someone who should be a top 10 player once active is tremendously valuable, especially when you consider you should have him for the playoff weeks. I don't know exactly what the ideal adjustment is, but I do know I'm close even if I completely ignore the roster spot he'll eat for the first 6 weeks.

Re: Gronk GP and Replacement
by aron7awol :: Mon, 08/19/2013 - 9:32pm

Ahh, I missed the cell comment. Thanks.

I defined replacement level as the best non-starting TE (11th ranked TE in a 10 team league) and gave Gronk credit for 6 games at that level of scoring. That moved him way up in the rankings, I won't post here exactly how much, but he's a huge steal at his current ADP of 40+.

For those wondering why I made this adjustment, consider two players: one that scores 100 points in 16 games, and one that scores 100 points in 8 games. It is obvious that the second player is much more valuable, but the current KUBIAK spreadsheet would value them equally. The second player must get an adjustment for the fact that you can start someone else in his place for the other 8 games.

Aaron, you should really consider making this adjustment in the KUBIAK sheet so it can properly value players projected to miss time.

Re: Gronk GP and Replacement
by Zheng :: Tue, 08/20/2013 - 10:07pm

Isn't that what the risk adjustment setting is for?

Re: Gronk GP and Replacement
by Paydro :: Tue, 08/20/2013 - 11:00pm

Risk only makes a difference of 5-10% in the projection. Going from 10 games to 16 is, obviously, a MUCH larger difference.

It's an interesting point, and does seem like something that could be integrated, at least as an option. As it stands, anybody expected to miss games is getting undervalued to some degree.

Re: Gronk GP and Replacement
by Sporran :: Wed, 08/21/2013 - 1:15pm

If we knew that Gronk was going to miss exactly 6 games, then this would be true. However, he has been injured quite a bit in his career, and remains an injury risk beyond those 6 games.

Also, the normal use-case for the "Risk Factor" is for players whom we suspect may be injured at some point but we don't know when. Foster is a great example. He may play a full 16 games, or he may miss a couple. The worst-case scenario is that he starts a game, but then leaves after 2 or 3 carries and doesn't return. Replacement-value players can't help you in that case.

In other words, Gronk may be slightly under-valued based on what we know at the moment, but there is considerable downside risk based on what we don't know, so using Kubiak's value for him seems reasonable even without adjustment.

Re: Gronk GP and Replacement
by Paydro :: Wed, 08/21/2013 - 8:18pm

Sure, he remains an injury risk. But most players, even ones with injury risk, are projected at 16 games and then the risk factor is applied. In theory these two things are not incompatible; a setting that gives you a "plus replacement value" could be applied as well as the red "risk factor."

If you're down on him or whatever, that's fine, I'm just backing up aron7awol to say it would be helpful to have that calculation available. Personally, I don't even use "risk factor" because I think it's too subjective... I'd rather know what KUBIAK thinks will happen if he stays healthy, and adjust downward subjectively myself.

But here's a better example. Justin Blackmon is suspended the first four weeks of the season. Blackmon's been hurt, but he played 16 weeks last season and is coming into this one at 100%. If I draft him, I have no real reason to suspect he'll miss time the rest of the season. In the meantime, I don't have to let him sit there as a zero in my lineup. Therefore Blackmon's VBD doesn't really show how much he could help your team... with my personal custom auction values, this is a difference of ~$6, which is kind of a lot for a player at that (relatively low) caliber.

Re: Gronk GP and Replacement
by aron7awol :: Wed, 08/21/2013 - 11:52pm

Gronk may even miss less than 6 games, which would boost his value even more. Sure, he's at risk of re-injury, but I think the chance that he misses less than 6 games offsets that risk.

Yes, Blackmon is the other relevant guy affected by this. After adjustment, he became a good undervalued target on my board.

It's pretty much impossible to logically argue that the current approach doesn't severely undervalue players in this situation.

Re: Gronk GP and Replacement
by Sporran :: Thu, 08/22/2013 - 3:37pm

The latest update has reduced the amount of time Gronk is expected to miss to 3-4 games, per Aaron's latest post on the other thread.

You're arguing for a points per game evaluation, which is fine, but not as useful as you might think. Gronk and Blackmon are essentially useless for 1/4 of the season. You don't think that depresses their value? If I'm choosing players for the entire season, I want to know how many points to expect for the season. If I'm choosing players for a specific week, I want to know how many points to expect for that week.

Consider draft strategy. If I draft Gronk, I have to draft a backup TE. There is an opportunity cost there -- drafting a solid backup prevents you from adding another lottery ticket (Gio Bernard, say) that could potentially help out more.

Blackmon is a different issue, since he essentially is a backup / lottery ticket. You're gambling on upside with him. KUBIAK by nature doesn't deal with upsides, but rather averages. If you use KUBIAK to rank players in terms of possible upside, then you are using the wrong tool.

Re: Gronk GP and Replacement
by aron7awol :: Thu, 08/22/2013 - 11:07pm

I think you're misunderstanding the adjustment I'm making. I'm not arguing for a points per game evaluation at all. In the games that Gronk is projected to play, he's getting his projected points. In the first few weeks that he's not, I'm assuming I get backup TE level production, which is likely exactly what I'll get.

Let me pose this question to you: Would you rather have a player who scores 160 points in 16 games or one who is suspended for 8 games and then scores 160 points in 8 games? I would hope you would say the second player. At the very least, you should admit that their values are drastically different, while KUBIAK would rank them exactly the same.

Yes, he eats a roster spot for the few weeks that he's out, but then he becomes an elite talent in your starting lineup and is there for your playoff run. That vastly outweighs the value of whoever the last man on your roster would be, someone who was a ~25th round pick or whatever your roster size is. The only cost of rostering Gronk for a few weeks is losing out on that guy, and then you gain the roster spot back to use it on the same guy or someone else.

Re: Gronk GP and Replacement
by milas :: Sat, 08/24/2013 - 1:51am

I've spent entirely too much time thinking about this same question as to how to properly value Gronk this year, but I think your kind of asking the wrong question and the example you provided isn't really an accurate picture of what's going on here.

In the example above, obviously you would want the guy who scores 160 points in 8 games as he is twice the player as the other one, but that's not what is going on in the case of Gronk this year. In the example we would not only be getting 8 games of 160 point production from this player, but we also get another 8 games of production from the replacement player, so our actually point total would be 160+whatever replacement player scores. But this isn't how it works when you are valuing players based on FPOB and not just on total points. In an actual fantasy league if we draft Gronk and assume we get normal Gronk production for 12 games, this means that we are only getting replacement level production (0 points above baseline) for the other 4.

The more relevant analogy is something more like: Would you rather have a player who scores 160 points over the baseline at his position in 16 games, or a player who scores 160 points over the baseline at his position in 8 games, while in the other 8 games you get replacement level production (0 points over baseline) out of whoever you are playing in that spot. In both of these cases, at the end of the day you are getting 160 points over baseline production at that position over the course of the season, so IF the regular season was all that mattered for the most part you should be relatively indifferent to either option in my example.

But the tricky part is in figuring out how much to boost Gronk based on the fact that we have fantasy playoffs. KUBIAK tries to take this into account with the playoff adjustment factor but I'm a little curious how this was calculated specifically in Gronk's case. Intuitively, the adjustment seems low to me, although I have no numbers to back that up. Clearly, a playoff game is quite a bit more important than a regular season one, and the value of having someone who when healthy is probably a top 15-20 fantasy player for the most important games of the season just feels like it should be worth more than the 8% bump KUBIAK is giving Gronk right now. If Aaron is still following this I'd be interested in hearing from him how the playoff adjustment factors specifically for Gronk (and to a lesser extent Blackmon) were calculated.

So I guess in short, while I think Gronk is undervalued by KUBIAK, he's not undervalued to the point where you can just extrapolate out his stats for a full season and say he's worth a top 20 pick this year. If you feel confident in your ability to draft Gronk and still field a team that will make the playoffs where you can reap the benefits of taking a chance on him, then he should probably be bumped up a bit (but not into the top 20) in your KUBIAK rankings, whereas if you are in a competitive league where a playoff spot is not guaranteed, his value is probably closer to where KUBIAK currently has him.

Re: Gronk GP and Replacement
by aron7awol :: Sat, 08/24/2013 - 2:44pm

While I understand your line of thinking here, that the replacement player is the baseline so his contributions are 0 points over baseline and thus he doesn't improve Gronk's points over baseline, you're not accounting for the fact that this baseline is a "replacement" player who plays 16 games. Here's an example that will explain what I mean:

Let's assume Gronk is a 15 point per game player, and the baseline (replacement) player is a 5 point per game player. The baseline is thus 5*16 = 80 points. Gronk on a per-game basis is 10 points above replacement, so in his 10 games, he is 100 points above replacement. Except his total points are only 15*10 = 150 points, and so he ends up at 150-80 = 70 points above replacement. So he's basically getting his 10 points above replacement per game for the 10 weeks he plays, but he's getting penalized 5 points below replacement for the other 6 games, since he's a 0 points per game player for those 6 weeks while replacement level is 5 per game. That's the whole problem. While you're right that the replacement player during those 6 weeks is 0 points above baseline, that's 5 points per game better than the penalty Gronk is currently getting for those weeks.

I hope that sheds some light on this issue for you. I'm 100% sure that this needs to be done as a starting point. From there, you may want to penalize him somewhat for eating a roster spot for the first few weeks, but leaving it as is is certainly not the answer. In doing so, you are actually assuming you will not start a TE at all during those first 6 weeks, getting 0 points per game from that starting roster spot. If that's what you plan on doing, fine, but that would be a terrible way to run a team.

Re: Gronk GP and Replacement
by Aaron Schatz :: Fri, 08/23/2013 - 8:41pm

I just want to say this is an interesting idea. Next year, I will try to play around with having a separate option for value where players expected to miss games can have "replacement level" value added for the games they are supposed to miss. If we provided this option, it would be just that: an option, not the only values being given.

Re: Gronk GP and Replacement
by Zheng :: Fri, 08/23/2013 - 11:34pm

You could even put in a "games expected to play" column, which modifies the projected numbers accordingly. That could even save you time putting out KUBIAK updates as the inevitable preseason carnage gets underway. Just a thought.

Re: Gronk GP and Replacement
by Phyrre56 :: Sat, 08/24/2013 - 10:29am

Just wanted to say that I've been making this adjustment to my own KUBIAKs (adding replacement level games for injured players) for a few years and while it works -- it tends to grossly overrate injured players relative to ADP. "Most people" haven't figured out this trick yet that Gronk + replacement is probably better than any TE except Jimmy Graham, so his ADP is depressed. Just be careful because if you don't pay attention to ADP this method will lead you to draft injured guys much too early.

Re: Gronk GP and Replacement
by aron7awol :: Sat, 08/24/2013 - 2:53pm

This is a good point, but drafting straight off a value-ranked cheatsheet without accounting for ADP is suboptimal for all players, not just the injured ones. While trying to predict when a player will go off the board and grab him just before that happens is risky, there's little reason to grab a guy 5 rounds early just because he's the highest value guy remaining on your cheatsheet either. You are safe grabbing them a round or two early in most cases, and that will add a lot of value over straight VBD.

Also, while your statement "it tends to grossly overrate injured players relative to ADP" is absolutely correct, I'd like to clarify that it's actually ADP that is undervaluing the players rather than this adjustment overvaluing them relative to their actual value. I just don't want the way you chose to word it to confuse some people into thinking that it's the other way around.

Great discussion here, guys, and I hope it helps people out there get even more value out of their drafts.

Re: Gronk GP and Replacement
by Phyrre56 :: Mon, 08/26/2013 - 12:03am

I understand what you're saying, but the truth is a bit more complicated than whether VBD or ADP is "right." A pure VBD approach of Gronk + replacement player is undervaluing the opportunity cost of rostering that replacement player. Maybe that roster spot could have been used to draft this year's Alfred Morris.

I think we can all agree that the value of an injured player is greater than his expected contributions compared 1-to-1 against another player expected to play all 16 games. But his value is also not equal to his production + replacement. We just don't know how to value "lost bench spot" because it's entirely context dependent. The average value of a bench spot is probably small but it's definitely not zero.

It's especially tough to figure out at TE because many people don't even bother drafting a backup TE. Really the question is what's the value of Gronk + replacement - RB/WR late round flier. When you start doing that math, Gronk's ADP seems more logical.

Re: Gronk GP and Replacement
by aron7awol :: Mon, 08/26/2013 - 10:22am

>The average value of a bench spot is probably small but it's definitely not zero.
I agree completely. It is also minimized by the fact that when Gronk is back in just a few weeks, you can drop the backup TE and use that roster spot for a waiver wire flier, which isn't going to be drastically different from a flier you would take in the last round. This is why I haven't made a major adjustment down. What's the difference in EV between a last round of the draft flier and a week 3 waiver wire flier? I can't imagine it's more than a few points. In my case, I drafted Gronk and Jordan Cameron as my backup TE. Cameron has high upside himself, so he's not without his own value as a flier.

My conclusion is still that ADP is grossly underestimating injured players' value.

Re: Gronk GP and Replacement
by milas :: Mon, 08/26/2013 - 12:30pm

I have to say after thinking about this a bit more I'm starting to agree with you. The heart of the issue seems to be that KUBIAK calculates and then ranks individual players based on FPOB, but what really matters for fantasy value is instead the FPOB for a given "slot" in the roster (TE slot in this case). When players aren't expected to miss any time these two values should be the same but they diverge when you're talking about injured players, and you need to adjust the rankings accordingly. One other thing I was thinking about for this was that if you make this adjustment to gronks value, shouldn't that also mean that you need to reduce his playoff adjustment factor? So you have like decent boost in FPOB and then a slight downgrade due to smaller adjustment factor. Also wanted to say that i completely agree that the value of using a back end roster spot on a replacement level tight end, while nonzero, is negligible in comparison to value of getting an under ranked player

Re: Gronk GP and Replacement
by Phyrre56 :: Mon, 08/26/2013 - 12:47pm

You definitely need to reduce the Playoff Adjustment if you turn Gronk into Gronk + Replacement. I'm not entirely clear about the high Playoff Adjustment for injured players but I assume it started based on an observation that Gronk was coming in too low in the rankings. Bumping up the Playoff Adjustment is easier than changing how KUBIAK handles injured players at a macro level.

Re: Gronk GP and Replacement
by aron7awol :: Mon, 08/26/2013 - 2:16pm

Good catch, guys, on the playoff adjustment. Playoff adjustment probably = projected points per week during playoff weeks / regular season points per week (using 16 games as denominator). Since Gronk's regular season points per week (with 16 game denominator) were suppressed, it makes his playoff adjustment abnormally high. So I agree that now that we are bumping up his regular season points per week we need to knock his playoff adjustment back down to a normal level.

Re: Gronk GP and Replacement
by jjjk (not verified) :: Tue, 08/27/2013 - 12:26pm

This should be a function of bench spots. If you've got 8 bench spots, the value of each is miniscule. If you're in a 14 team league with 2 bench spots, well, those have a lot of value. Having the best backup RBs is a huge advantage vs a guy who has to roster a backup TE for 8 weeks. That's 8 shots at say, ben tate getting a start and producing huge value.