CDG question

[Doc]

Has anyone noticed fairly large changes in doller values in the CDG weekly if not daily? I run it pretty much every day just to have it when evaluating and looking at trades and pick-ups and I think I am going crazy. One day John Danks is $2 the very next day he is -$3. Looking at a trade offer for lowell. CDG has him at $12 and my own 3rd baseman rolen at $20, reject. Two days later lowell $18 and Rolen $19. When guys get injured it is even worse. C.young $14. Gets hit in face $10 , is going to be out about 3 weeks , $4 now. I guess I should try to move him, but I am afraid he comes back and his value at the ASB is $14 again. Some times it is only $1 here and a $1 there, but many,many values can fluxuate $6 to $8 daily or weekly. So much so that it is affecting my trading and FA pick-ups. I know I put the exact same parameters in the CDG so what is it? thanks

[RAY@HQ]

To state the obvious, there are two variables: playing time and production. Both can be changing... playing time as injuries/roles change, and projected production as we continually adjust the blend of historical v. in-season influences on the projections.

In short, you're taking something that's going to be dynamic by nature, just because of the variables involved, and asking for it to be static.

[RAY@HQ]

By the way, what are those league parameters you use? Because what you're inputting, for instance, for that 17-team keeper league will affect the dollars differently than what you'd see in Playerlink for the AL- or NL-only values.

[Doc]

I understand all that Ray, and I input the same parameters into the CDG so that i do get the same results. I have been a member for awhile and i never remember values changing so dramaticaly before. And I understand a doller here and a doller there. I can understand moving slightly in one direction, for example Kubel listed at $21 day one in the season, now $10. That i can understand since it was a slow progression down. What I don't understand is May 22nd Mike lowell $12, david wright $51. May 29th Lowell $18, Wright $45. These are just two examples, but there a ton of them. So much so that a trade that looks good one week, 5 days later looks bad. I challenge you to run the CDG for a 20 team league one week then run it one week later and check the differences. I don't know what kind of formula is being used, that is why I asked the question.

[Doc]

Here is just an example: John Danks

May 22nd

Wins-7 IP-106 K-88 ERA-4.42 Whip-1.37 $2

May 29th

Wins-6 IP-108 K-90 ERA-4.58 Whip-1.39 -$3

So in one week his pIP and his pK went UP pERA and pWhip went UP and his value dropped $5
One week when deciding who to try to trade or drop for an injury, Danks looks like a keeper, the next week, he looks like a drop.

[RAY@HQ]

Doc, I understand where you're coming from. The only point I was making, which you clearly understand from your most recent post, is that the $ values are just a representation of the underlying stats... and it's the way the stats are adjusting that your questioning. There's a post somewhere else (the Cano projection thread, maybe?) from Ron (search for recent posts by Ron@HQ) that talks about how he's adjusted the in-season data model to be more responsive to early-season trends. I believe what you're seeing is a result of that change.

[RobR]

Here is just an example: John Danks

May 22nd

Wins-7 IP-106 K-88 ERA-4.42 Whip-1.37 $2

May 29th

Wins-6 IP-108 K-90 ERA-4.58 Whip-1.39 -$3

So in one week his pIP and his pK went UP pERA and pWhip went UP and his value dropped $5
One week when deciding who to try to trade or drop for an injury, Danks looks like a keeper, the next week, he looks like a drop.

I keep decent, but certainly not daily, archives of our projections and they're definitely not for Doc's 17 team 6x6 league. I don't see the 5/22 values quoted above for Danks. They seem to be more from approximately 5/8/08 when maybe Ron was working on the model change Ray mentions above. I show on 5/11/08 the projection wasn't very different from today.

For Chris Young the changes must be specific to Doc's league settings because I only see one change around his injury for our normal valuation parameters. You should expect his projected value in the future to increase once he's off the DL. Right now, he's projected for less IP than his counterparts and his value should be lower than those with roughly the same skills, but after he's back in the lineup, he'll be projected for about the same IP as his counterpart and you'd expect his value to return about the same level as the pitchers with about the same skills.

For Lowell, I see we increased his expected playing time around the time of the change, which is again around the dates I mentioned above, not between 5/22 and 5/29.

Doc, if you specify the exact league parameters and single out any more changes you can't explain I'm happy to look into them and try to help explain them.

[Doc]

Interesting you should mention the injury values, rob, since i do have a question. A guy like young who may not be projected to come back for 4 weeks. His value now might be $4 and a guy like slowey $8. But, maybe a week AFTER young comes back his value could be like $10 and slowey could still be $8 right? Is this correct? Or am I missing a piece. If this is so , would it not be more helpful to show how much young will be "worth" from his estimated time back on instead of basing his value on a month of zeros?

[RAY@HQ]

On the injury question: when a guy gets hurt, his playing time for the balance of the season gets cut. That's obviously going to drop his value. We do this buy %, you can see these on the team depth charts.

When a guy returns, as you propose it's likely that he might get a playing time increase... maybe we projected him for 70% balance of year playing time while he's out, then maybe that jumps to 90% when he actually comes off the DL.

Ideally, that adjustment should probably happen in small increments even while he's on the DL but getting closer to returning, something like 5% a week. To be honest, I'm not sure how our playing time analysts handle that situation right now. But this is an almost unavoidable side effect of doling out playing time in percentages.

[RobR]

Interesting you should mention the injury values, rob, since i do have a question. A guy like young who may not be projected to come back for 4 weeks. His value now might be $4 and a guy like slowey $8. But, maybe a week AFTER young comes back his value could be like $10 and slowey could still be $8 right? Is this correct? Or am I missing a piece. If this is so , would it not be more helpful to show how much young will be "worth" from his estimated time back on instead of basing his value on a month of zeros?

I think you have it right if we assume Young was $10 before his injury reduced him to $4. Assuming their skills don't change and the rest of the leagues skills don't change as soon as Young gets knocked out his relative value will drop and it should converge back to the previous level once he's projected for the same percentage of playing time as before. Ray covered the mechanics of this on the site well.

Mainly, when Young got knocked out, I want to see his relative value take that drop because that's what I need to know about who's the best for my team for the remainder of the season. I could see some secondary use in some sort of a rate R$ stat that would tell me what his R$ per projected start (or something along those lines) is, more so in leagues where the waiver wire is thinner.

[Doc]

In my league we have 4 dl spots so, unless you have some major injuries it is easy to keep a guy until they come back. How would you recommend I use the values then? Since I don't want to trade young if he comes back and becomes the 3rd best pitcher on my team.

[RAY@HQ]

Well, if you have DL slots that don't cost you anything, you want to just stash guys in those slots while they're DL'd and not trade them during that time, right? In general, a guy's market value is going to plunge while shelved (because they offer no immediate help) then spike when they come back healthy... in some sense, that kind of mimics the behavior of our $ values you describe above (unintentionally, of course).

As the guy gets closer to coming back, his dollar value is going to get more accurate in terms of balance-of-year value to you.