Posted by schwarze on 8/29/2017 7:13:00 PM (view original):
20 of the top 24 have a pre-ranking of 28 or better

Maybe the tweaked formula is not so bad after all

Posted by schwarze on 8/24/2017 3:47:00 PM (view original):

Posted by schwarze on 8/20/2017 7:37:00 PM (view original):
My teams have a very lucky 34-16 record in 1-run games... similar to asufan04 from last year. Regression caught up with him and he missed advancing to round 2 (he finished 33rd)..

The good news is that my record in 2+ run games is 72-40 for a .642 win%. So if I go .500 in all my remaining 1-run games, and keep the .642 going in 2+ run games, that equates to an average of 97 wins per team going forward. I'll take that all day.

We're still getting extremely lucky in 1-run games... 46-19 (.708), that's 12-3 sine my last update. We're a healthy .630 in games decided by 2+ runs (102-59). Waiting for the regression to hit and hit hard.

My actual winning percentage (.658) is higher than my expected winning percentage (.635) and that includes one mopup blowout win and one mopup blowout loss.

Now 57-36 in 1-run games, so since my last post, 11-17.... now that's regression.

Record in two run games is 145-86, .627. Since the last post... 43-27, .614... holding pretty steady.

Posted by ozomatli on 8/29/2017 7:15:00 PM (view original):

Posted by schwarze on 8/29/2017 7:13:00 PM (view original):
20 of the top 24 have a pre-ranking of 28 or better

Maybe the tweaked formula is not so bad after all

Well, it depends on what the rankings would have been under the old formula...

Current standings compared with the old Pre Rank formula. 15 of the current top-24 would have been ranked 28 (really 35) or better

Rank__	PreRank__	Old PreRank__	Owner
-----	----------	----------	----------------
1	1	1	brianjw
2	5	9	schwarze
3	3	5	rbow923
4	12	37	doctorcc
5	28	39	charliembg
6	20	7	dougpalm
7	2	2	discodemo
7	26	22	mpitt76
9	8	6	mildnhazy
10	9	25	ozomatli
10	18	42	jfranco77
12	6	4	jmissirlis
12	22	10	redwingscup
14	59	45	beauchamp
15	11	12	buddhagamer
15	25	11	justinlee_24
17	76	65	toddcommish
17	66	55	Jdh34
19	27	21	batandball
20	10	13	bigmc
20	43	43	mllama54
20	15	35	Relkcirts
20	24	44	ballantine
24	16	17	crystalao
25	36	41	Fusion27s
25	91	73	knieman
25	21	30	markeking
25	19	20	bighooze
29	39	49	XJchessman
29	4	3	jbohrman
29	62	58	frank_drebin
29	33	19	tigerrott
29	94	52	NebHusker
34	70	67	Jtpsops
34	53	50	DoctorKz
34	30	31	cstrohmier
34	45	24	rmdriskill
38	51	46	magicdreamer
38	80	79	BullMoose
38	73	82	zlionsfan
38	37	34	ledfoot
42	14	15	barracuda3
43	49	61	fooolishfool
43	57	51	kyleb63
45	77	81	dodgebob
45	74	75	greenies
47	40	18	cubbies84
48	7	8	bardin
48	35	28	ejstockman
48	61	70	Chisock
48	41	14	jjgreen14
52	23	26	newarkwilder
52	29	33	daddyzander
52	46	47	redcped
55	69	62	jetscanes
55	72	66	mick33
55	64	74	DarthDurron
58	13	16	pedrocerrano
59	38	29	firesalt
59	44	56	d_rock97
61	32	32	brucel
62	17	27	sford
62	48	38	calhoop
64	93	92	fatboydad54
65	65	64	dlchow
65	34	60	glowguy
67	71	68	thejuice6
68	89	91	scottharra
69	55	53	rickysdad44
69	31	23	Lazerhawks
71	83	80	wink0094
72	82	93	npg32433
73	42	40	hersheybear
74	56	36	paulives
74	54	59	Humdogs
76	84	85	garmansouth
76	79	89	happyhours
76	68	71	nordawg
79	47	83	sjh0825
79	75	84	gigrant
81	58	48	KOTH
81	67	63	hacker7
83	78	78	thebubbaq
84	60	87	slightner
84	81	76	hurst47
86	63	57	trd3
86	52	54	thehuseman
88	50	69	moose821
89	87	86	midknight
90	92	77	brickster44
90	96	96	darthmetool
92	88	88	spikeboots
93	86	90	tridentric
94	90	94	spoonfed
95	95	95	rufus4ever
96	85	72	sohio4ever

Overall, there are hits and misses on both sides, even if there is better predictive performance for the Top 24 (so far) with the new formula.

dougpalm, redwingscup and justinlee_24 seem to have been underrated by the new formula, while doctorcc, jfranco77 and ozomatli would have been underrated by the old formula, and neither formula saw Jdh34, beauchamp or toddcommish coming.

Also, on the other end, 84th place slightner would have been better assessed by the old formula, but that formula ranked current-last-place sohio4ever at 72 (vs 85 from the new formula).

I see why doctorrcc and jfranco are ranked too low by the old formula. doctorcc finished 53rd in 2014 and jfranco finished 68 in 2015. Those poor finishes really hurt their overall ranking. The reason I avoided putting too much weight on the inception-to-date records is because of progressive leagues, etc. skewing the data. But I can see their are flaws with both systems. So far, your formula looks to be more predictive.

I think I'm closer to 33rd than 19th. I don't see why I should move up from 22nd (last year's ranking) after a 61st place finish. Kudos to ozo! BTW, I'm barely in the top 30 this year.

First time checking the standings, as I was afraid of what i'd see, pleasant surprise :)

Here's a stat that shows why I have little confidence in my relief pitching. I spent a combined 84M on it, 30% less than the median amount in my leagues. I don't foresee a reversion to the mean given that. I will continue to struggle to win close games.

Thursday's pm2 set marks the all star break and the next 162 game update.

Posted by schwarze on 8/26/2017 7:34:00 PM (view original):
Mark down the date.... 8/26/2017... right before the pm2 games (where I went 1-5)... the last time I would hold the lead in the WIS Championship.

Back into a tie for first, after a roller coaster ride for each of us.

What is your record in 1-run games, brianjw?

93-62. Not that different to my overall record now.

Wow, you have played 20% more 1-run games than I have.

Thru 77-games, my teams are a combined 77-52 (.597) in 1-run games and 203-130 (.610) in 2+ run games. My actual win% of .606 is pretty close to my expected win% of .619. After a terrible start of this third 162-game period (we were < .500), we've caught fire lately and are now playing at a .565 pace overall during this period.

Posted by brianjw on 8/20/2017 10:31:00 PM (view original):
I'm an absurd 45-14 in 1 run games, so I don't think I'll keep this pace up. I'm barely over .500 in 2+ run games.

So since this post, you are 48-48 in 1-run games.... which is about right.