CDG 2020 and pitcher valuation

[JonE]

I'd like some help in understanding what looks like a radical realignment of dollar valuations in 2020.

I ran the CDG today for a 10-team 4x4 mixed league. Here are the projections and values for two relievers.

                W-L  SV    IP     ERA   WHIP    R$
Stammen, Craig  2-1   0  16.0    3.38   1.06     9   
Hader, Joshua   1-1   9  17.0    3.18   0.94    -1

I know that wins are pretty scarce in this short season, but this makes no sense to me. Stammen is a good pitcher, but Hader is even better. He has a better ERA and WHIP, is projected for one additional inning, and is expected to get 9 saves. Why should he be worth $10 less than Stammen? Does a pitcher actually hurt his team by collecting saves?

[RAY@HQ]

Geez, that's certainly weird.

I think the answer is just that our sample sizes are so small that we're starting to stretch the bounds of the projections engine. For instance, if I do the same calcs for a 15-team mixed, I get Hader $24, Stammen $11... which is reasonable. But if I shrink the universe to 12 teams, the results snap toward yours above.

I believe what happens in the smaller universe is that even though saves are concentrated among fewer teams, there are enough that they're essentially freely available. In PVM terms, Hader's saves don't matter much. But the difference between the 1 and 2 wins matters immensely.

That's that I think is happening. Which is not to say that I think it's correct. That definitely doesn't pass the smell test for me.

[The Shadow]

I got this bizarro world outcome today, too:

Name	age	T	TM	$	MM	ADP	w	l	g	qs	sv	bsv	hld	ip	h	er	hr	bb	k	era	whip	k9	bb9	bfg	hr9	cmd	oob	rar	bpv	xera	MM Code	LIMA
Cole,Gerrit	29	R	NYY	$ 53	54	4	3	1	6	3	0	0	0	37.0	28	12	4	9	49	2.92	1.00	11.9	2.2	23.6	1.0	5.4	262	6.0	173	3.10	5503 BAB	B
Lynn,Lance	33	R	TEX	$ 45	39	105	3	1	7	3	0	0	0	38.0	30	15	4	13	43	3.55	1.13	10.2	3.1	21.4	1.0	3.3	286	3.1	119	3.76	3403 DAA	A
Gray,Sonny	30	R	CIN	$ 42	54	89	3	1	6	3	0	0	0	35.0	28	14	3	14	39	3.60	1.20	10.0	3.6	23.4	0.8	2.8	299	3.6	111	3.60	5503 CAA	A
Sanchez,Sixto	21	R	MIA	$ 34	6	706	3	1	5	2	0	0	0	27.0	28	12	4	6	21	4.00	1.26	7.0	2.0	22.0	1.3	3.5	309	0.6	98	4.25	2101 AFF	F
deGrom,Jacob	32	R	NYM	$ 21	54	9	2	1	5	3	0	0	0	34.0	26	11	3	7	41	2.91	0.97	10.8	1.8	25.8	0.8	5.9	256	6.0	169	3.01	5503 AAB	B
Bieber,Shane	25	R	CLE	$ 19	54	22	2	2	6	3	0	0	0	35.0	31	12	4	6	43	3.09	1.06	11.1	1.5	22.6	1.0	7.2	273	6.0	180	3.04	5503 ABA	A
Scherzer,Max	35	R	WAS	$ 19	48	14	2	2	6	4	0	0	0	35.0	29	13	4	8	48	3.34	1.06	12.3	2.1	22.6	1.0	6.0	273	5.6	182	3.13	4503 DAA	A

Clearly, DeGrom has superior stats to Sanchez in everything except wins, yet Sanchez has a 50% greater value. Based on that single outcome, I examined the entire Pitcher run, and the values were right in line with Wins. All 3 Win projections were of greater value than 2. 2 Wins were all greater than 1. And 1 Win was greater than none. No player with less than 2 wins made my positive value half of the CDG printout for Pitchers. Hader had 1 win, so he was down in the negative value table. Stammen had 2, and he made the positive table. James Pazos came out at $ -83.

If it helps, my settings were: 9 teams, Mixed, 5x5, traditional 14+9, $260 w/ $1 minimum, no bench, no forced positions.

I ran it 3 different times between 6 pm and 11 pm, and I got the same result each time.

In thinking about how Wins totally dominate the Pitcher rankings, I looked at my Batter values, and they seemed to be distribted in a fairly normal fashion. We're looking at the same 4 weeks remaining (approx Sept 1 in a normal year). I've never seen September 1 results this wonky before.

[The Shadow]

It's working for 9 teams, again!!

Thank you, HQ!!

[RAY@HQ]

Hmm, I don't think I changed anything. Did you change a CDG setting?

[The Shadow]

Hmm, I don't think I changed anything. Did you change a CDG setting?

I needed to get a proper ranking of pitchers, so I ran it with 15 teams. That provided values that made sense. Then, I tried it with 9 teams again, but I noticed I didn't have a number of bench players in for this league, so I added that (changed from 0 to 4). When I ran it for 9 teams, the values all made sense again, and there was no absolute correlation between Wins and value any longer. Odd. 2020, I guess.

[RAY@HQ]

Adding bench players effectively makes the league deeper in the CDG, so it makes it look like the 15-team league. And in my first response in this thread, I noted that deep league values still seemed fine. So that all checks out. Glad you're all set.

[JonE]

Shadow's approach makes sense for a workaround.

But after thinking about it I suspect it's the scarcity of Wins, rather than the relative abundance of Saves, that's making the valuation engine go wonky. In a normal season wins can run from 0 to 20, and the majority are concentrated in a relatively small number of pitchers. The same is true for saves. In 2020, Saves still has something of a spread -- there is a difference between a Hader who will get 10 saves and, say, a Knebel who will pick up 1. But because Wins only range from 0 to 5, each win is going to have a huge effect on pitcher value, especially when you get down to valuing 90 pitchers instead of 135.

That's a valuable lesson for using CDG in small leagues, anyway.

[RAY@HQ]

Shadow's approach makes sense for a workaround.

But after thinking about it I suspect it's the scarcity of Wins, rather than the relative abundance of Saves, that's making the valuation engine go wonky. In a normal season wins can run from 0 to 20, and the majority are concentrated in a relatively small number of pitchers. The same is true for saves. In 2020, Saves still has something of a spread -- there is a difference between a Hader who will get 10 saves and, say, a Knebel who will pick up 1. But because Wins only range from 0 to 5, each win is going to have a huge effect on pitcher value, especially when you get down to valuing 90 pitchers instead of 135.

That's a valuable lesson for using CDG in small leagues, anyway.

Yeah, I think this is spot-on.

[The Shadow]

Shadow's approach makes sense for a workaround.

But after thinking about it I suspect it's the scarcity of Wins, rather than the relative abundance of Saves, that's making the valuation engine go wonky. In a normal season wins can run from 0 to 20, and the majority are concentrated in a relatively small number of pitchers. The same is true for saves. In 2020, Saves still has something of a spread -- there is a difference between a Hader who will get 10 saves and, say, a Knebel who will pick up 1. But because Wins only range from 0 to 5, each win is going to have a huge effect on pitcher value, especially when you get down to valuing 90 pitchers instead of 135.

That's a valuable lesson for using CDG in small leagues, anyway.

I agree, that appears to be it, at least in terms of pushing down values for Closers. Still, Wins is just one category in a 5x5 league. It still seems odd that it could become the dominant factor in a valuation between the likes of Sixto Sanchez and Jacob DeGrom - both starters - where Wins is one of 4 compared categories (both will get zero Saves). DeGrom is as dominant over Sanchez (if not more so) in all 3 of the other Starter relevant categories. Sanchez, with a 50% "bonus" in Wins had a 50% greater value, essentially setting DeGrom's dominance in the other 3 categories to $0. It seems that in a PVM valuation, that shouldn't happen, regardless of smallish league size of limited remaining games.

[Howie M]

I have seen this with SB - but tough to argue with it.

at midseason, a batter with 2 or 3 SB provides serious added value.

my NL-only SB team totals before tonight:

23 19 17 16 15 14 14 13 12 11 10 9

the league leader and runnerup don't get help yet from 2 steals - but every other team does

[JonE]

Another way that Steals resemble Saves is that they are relatively concentrated. There are a handful of guys who have a lot, a number of guys who have 1-2, and an even larger number of guys who have 0. But in 2020, the spread in Wins runs from 0 to 3, and hardly anyone has 3.

[bolton wanderers]

There also seems to be a bug/feature with reliever projections whereas the number of games is less than decisions (wins+loses+holds+saves). Total of 15 players with Jonathan Hernandez: in 4 games he will get 6 holds as well as win one and lose one game. Only effects my points league (and barely). Understand things getting wonky with small size of current season to go as well as played (i.e. NOT paying attention to pERA and pWHIP), but odd to see the impossible coming out possible (more decisions than games played).

[RAY@HQ]

I'll clean that up for Hernandez. If you see other examples let me know.

[bolton wanderers]

The other 14 are: Hader – Gallegos – Montero - Bradley – Wick – Newsome - Hearn – Nelson – Yamaguchi - Paredes – Sneed – Patino - Soto - Alexander


Thanks.