Updated... enjoy!

First...huge thanks to just4me for compiling all of this (and to chargingryno for organizing the league in the first place). I started the TWISL (Testing What If Sports League) concept about a year ago for this exact kind of purpose - setting up conditions that would allow us to test different hypotheses. For example, this past year we showed - definitively, as far as I am concerned - that in SLB, low K pitchers do NOT get more plus plays from their defense, all else being equal, than high K pitchers. This is despite many assumptions to the contrary that have been posted in these forums over the years. I could not be happier to see other owners picking up the charge and designing leagues to test other propositions. Huge thanks to both of you.

With that out of the way, I am going to gently disagree with some of the conclusions posted above. Disclaimer: I ran the Fatigue is Just a State of Mind team that is referenced above. My strong conclusion from this league is that it is not a good strategy to run a fatigued pitching staff. If the league is at all competitive, you will not gain enough in offense and defense from the extra salary spent on position players to offset the poorer performance from the fatigued pitching staff. My team had the best offense in the league (by a small margin), had the best defense in the league (by a WIDE margin), was one of only 2 of the fatigue teams to play better than .500, and still WAY underperformed by control team. Isn't the obvious conclusion that one is far far better off spending the money to keep your pitching staff reasonably well rested? None of the fatigue teams made the playoffs. Only two finished over .500. The very best of these teams still finished ~10 games worse than the control team that was run by the same owner. I mean, doesn't that suggest that EVEN IF you really know what you are doing and EVERYTHING breaks your way, that it doesn't pay off?

My second point of disagreement (and I suppose at this point I am living up to my user name), doesn't it seem like fatigue is a VERY big deal even at low levels? Pitchers who were in the second tranch, averaging 96%, had ERAs 0.52 higher than pitchers who were fully rested. That's an enormous difference. Pitchers in the third tranch deteriorated even faster...another 0.6 increase in ERA for that group, which averaged 92% fatigue. To me this is VERY clear evidence that one should attempt to stay as close to 100% as possible with one's pitching staff, and that allowing even a little fatigue is a bad idea. The difference between a 3.37 ERA (for rested pitchers) and 4.49 (for the same pitchers at 92% fatigue) is enormous. Heck, I'd much rather draft worse pitchers and keep them rested than allow good pitchers to lose more than 1.00 runs off their ERA.

And remember, those ERA differences were in spite of having very strong defenses overall. In other words, I look at the performance in this league as sort of the "best case scenario" for what you can expect from a fatigued pitching staff. I shudder to think what would have happened had most of us not spent most of the incremental salary on players with A+++ range. To me, this league is close to definitive proof that it does not make sense at all to run a fatigued pitching staff in a competitive SLB league.

Here's the final pitching stats from my team:
Regular Season Pitching SimStats (Totals)

Player	SN	T	G	GS	CG	SHO	W	L	SV	SVO	IP	H	R	ER	HR	BB	SO	OAV	OBP	SLG	WHIP	ERA
Maddux, Greg	1996	R	77	62	0	0	16	22	5	6	325.7	386	216	187	44	53	134	.287	.319	.465	1.35	5.17
Halladay, Roy	2011	R	78	41	0	0	20	19	5	6	314.7	397	203	178	42	87	159	.303	.352	.450	1.54	5.09
Lee, Cliff	2010	L	73	59	0	0	8	18	2	3	283.0	372	218	192	55	62	124	.311	.348	.541	1.53	6.11
Fister, Doug	2011	R	86	0	0	0	9	2	4	5	87.7	107	45	43	8	14	42	.294	.325	.418	1.38	4.41
Wilson, Alex	2015	R	88	0	0	0	11	3	1	2	85.0	104	52	44	17	21	19	.297	.339	.494	1.47	4.66
Atchison, Scott	2014	R	114	0	0	0	6	2	4	6	83.7	90	62	53	12	25	35	.274	.326	.442	1.37	5.70
Doolittle, Sean	2017	L	90	0	0	0	2	3	4	7	68.7	77	40	35	12	19	34	.280	.325	.436	1.40	4.59
Weilman, Carl	1912	L	67	0	0	0	7	2	3	5	68.0	87	52	41	17	8	14	.300	.321	.552	1.40	5.43
Wagner, Billy	2008	L	90	0	0	0	5	2	6	7	65.7	61	31	29	7	20	53	.239	.300	.388	1.23	3.97
Rivera, Mariano	2002	R	83	0	0	0	3	2	0	1	62.0	63	45	42	13	22	24	.260	.325	.459	1.37	6.10
TEAM TOTALS	--	--	846	162	0	0	87	75	34	48	1,444.0	1744	964	844	227	331	638	.293	.334	.474	1.44	5.26

The only pitcher on the team for whom appearance fatigue should have been a significant issue was Atchison. Importantly, none of the 3 starters appeared in anything close to the number of games required for appearance fatigue to be a big issue. Given how in-game fatigue works in WIS, had I increased the pitch counts of the bullpen (most of whom were RL relievers with low IP/G), their performance would have been even worse, leading to more overall pitches for the staff and an even worse death spiral from fatigue. The basic problem here - and IMO the real risk from this strategy - is that if you get in trouble early (even a few extra inning games) you can't get out of it. There is no buffer.

Posted by contrarian23 on 12/19/2020 10:00:00 PM (view original):
The only pitcher on the team for whom appearance fatigue should have been a significant issue was Atchison. Importantly, none of the 3 starters appeared in anything close to the number of games required for appearance fatigue to be a big issue. Given how in-game fatigue works in WIS, had I increased the pitch counts of the bullpen (most of whom were RL relievers with low IP/G), their performance would have been even worse, leading to more overall pitches for the staff and an even worse death spiral from fatigue. The basic problem here - and IMO the real risk from this strategy - is that if you get in trouble early (even a few extra inning games) you can't get out of it. There is no buffer.

Appearance fatigue is weighted by pitches thrown relative to allotted pitches. For pitchers above their allotment or trending above their allotment (as all of the fatigue teams were), 70% of games is no longer the threshold, that begins to work back to lower percentage of acceptable team games. I haven’t worked out that side of the formula like I have for the other side of it (in my appearance fatigue thread). However, I’d guess that threshold was around 55-60% of team games for some of your guys, and they’d have been projecting beyond that frequently throughout the season.

This was the unseen factor I kept bumping into when trying to get 100+ starts and 1,000+ Innings out of Clarkson. I understand that relationship now, but I didn’t when trying to make that happen.

The very best of these teams still finished ~10 games worse than the control team that was run by the same owner. I mean, doesn't that suggest that EVEN IF you really know what you are doing and EVERYTHING breaks your way, that it doesn't pay off?

This is actually an excellent point.

I wouldn’t argue a fatigue team is better than a non-fatigue option, but that it’s still possible to win with fatigued pitchers. Think of it more as a backup plan if things don’t break your way, or for certain progressive teams, or certain themes... or for maybe 1 pitcher in a rotation... a play in your playbook, not necessarily the best play if all plays are available.

My goal with my observations was to help give an understanding of where things broke for the teams that weren’t successful to give a better understanding of the variables within the fatigue types and the relationship between pitching, defense, and offense.

Posted by just4me on 12/19/2020 10:30:00 PM (view original):

The very best of these teams still finished ~10 games worse than the control team that was run by the same owner. I mean, doesn't that suggest that EVEN IF you really know what you are doing and EVERYTHING breaks your way, that it doesn't pay off?

This is actually an excellent point.

I wouldn’t argue a fatigue team is better than a non-fatigue option, but that it’s still possible to win with fatigued pitchers. Think of it more as a backup plan if things don’t break your way, or for certain progressive teams, or certain themes... or for maybe 1 pitcher in a rotation... a play in your playbook, not necessarily the best play if all plays are available.

First off, thank you for all this research. I'm still digesting the parts I fully grasp and just nibbling at the parts I really haven't gotten my head around.

That said, I've had a few themes where IP are limited (a 2020 draft league we just did and the ongoing Cooperstown Historical Replay, where you get your team's actual IP whatever they are). My strategy sometimes is to let the worst pitchers get more fatigued and protect the good ones, because I can set them to pitch more in blowouts and I really am not concerned if a guy with a 6 ERC# is putting up an ERA of 7 or 8 or 9. He's never going to be effective anyway.

Based on what I've seen here, I definitely won't worry if those guys are pitching frequently in the 80s and might be fine in the 60s and 70s if that's what's necessary to make it all work. If it helps me keep the good pitchers at or near 100, I think that's a worthy tradeoff that shouldn't cost me many games, if any.

Posted by just4me on 12/19/2020 4:11:00 PM (view original):
The following is the first data breaking down fatigue. This is the totals of each individual game from the game logs of the three starting pitchers on every team and sorted by fatigue % at the beginning of their start. Each line represents all of games within designated fatigue buckets (bucket levels shown on right, average fatigue per game in that bucket on the left). I used the same buckets and coloring (to match WIS) as my last post on fatigue representing Marshall. Form these three pitchers we have 2,319 games totaling 13,514.33 IP:

Games	Avg PC	AVG %	IP	BFP	PC	W	L	SV	R	ER	H	HR	SO	BB	WP	ERA	OAV	WHIP	TYPE
249	104.2	41.10	1299.33	7691	25936	8	167	0	2468	2244	2880	495	331	940	15	15.54	0.425	2.94	0-59%
246	172.2	64.71	1105.65	5192	16745	38	81	11	913	798	1557	209	466	271	8	6.50	0.319	1.65	60-69%
495	79.6	71.71	2755.53	12202	39419	143	168	3	1730	1535	3367	388	1325	497	11	5.01	0.289	1.40	70-74%
171	77.7	76.64	971.03	4064	13290	37	55	3	571	514	1169	113	509	164	6	4.76	0.286	1.37	75-79%
63	82.6	81.40	366.35	1606	5201	18	23	0	182	155	434	37	198	61	3	3.81	0.283	1.35	80-84%
26	85.1	86.73	156.69	673	2212	8	9	0	82	73	164	14	82	36	0	4.19	0.259	1.28	85-89%
120	92.5	92.17	777.03	3385	11095	39	42	0	436	388	918	71	530	125	7	4.49	0.283	1.34	90-94%
104	92.7	96.53	688.00	2932	9637	48	33	1	333	297	769	58	455	101	2	3.89	0.271	1.26	95-98%
845	89.1	99.92	5440.12	22821	75325	399	197	4	2362	2040	5597	369	3774	801	27	3.37	0.255	1.18	99-100%
2319	85.75	80.93	13514.33	60566	198860	738	775	22	9077	8044	16855	1754	7670	2996	79	5.36	0.294	1.47	Total

I had a thought on this chart in regards to "success" of using fatigued pitchers. very few games at 90-100% were by fatigue teams, which mostly had the good defenses behind them. All but 1 start between 70-89% were from the fatigue teams, which mostly had great defenses behind them. The stats at 90-98% are a fairly big drop from 99-100%, but the stats at 75-89% are almost identical to the stats at 90-98% and the stats at 70-74% are only a marginal drop from either 74-89% & 90-98%.

Contrarian's point still stands that anything below 99% has a solid dropoff, but when planning for the dropoff, it certainly seems that defense/ballpark/etc can mitigate that dropoff quite a bit. If the pitchers in the 90-98% groups had the same defense as the pitchers in the 70-89% groups, how much dropoff from 99-100% would there be (obviously accounting for defense at 99-100% would improve that group more, as well)?

Posted by just4me on 12/19/2020 8:23:00 PM (view original):
Observations (not already made above):

• Two Cliff Lee with least IP combined for 81 G, 476 IP, .271 OAV, 1.2457 WHIP, and 4.56 RA/9. Two Cliff Lee with most IP combined for 104 G, 581.67 IP, .272 OAV, 1.2464 WHIP, and 4.75 RA/9. Average fatigue of two with least IP was 100%, while average fatigue of two with most was 74%. That's 23 extra games and 105 extra IP with no real drop in performance.
• The three sets of teams run by a single owner were each nearly identical on offense despite each fatigue version having a top 3 defense and two control versions having a bottom 3 defense and the third a league average defense. The TWISL Fatigue team was also almost identical to the TWISL Control team on pitching, thanks to that extra defense.
• The 85-89% bucket, though the smallest sample, isn't necessarily an aberration, the better stats there over the 90-94% bucket could be explained by defense in that sample since these buckets aren't controlled by defense or matchups, etc...

With fatigue teams, managing work load across the three fatigue types is also crucial to success, as you can see from the results of the teams that treaded into appearance or in-game fatigue on top of their pitch-based fatigue.

Even though this is linked in the other thread this particular point about the multiple types of pitching fatigue and how they all interplay is important to think about in terms of how fatigue impacts performance.

Bump