All Forums > SimLeague Baseball > SimLeague Baseball > Is the Fatigue Code Broken in the Sim?
6/19/2013 3:14 AM
My thoughts are simple: it can't be done.

Here is the thought experiment I use.  Take 2 teams with the exact same amount of drafted PA/162, and the exact same amount of salary spent on position players.

Assume Team A is a much better offensive club than team B (either because Team A is poorer defensively and therefore more of their salary reflects their offensive value, or - and more to the point for this experiment - team A is just much better designed offensively).  Assume Team A scores 25% more runs than Team B.  I think this is a very conservative estimate of how much better an offense can be, even holding PA and salary constant, but we'll stick with 25%.  Obviously they will accumulate many more PA than team B.  But they have NOT - in any way - gamed the system compared to Team B.

Any fatigue system that punishes Team A, relative to Team B, for this improvement in performance is a non-starter to me.  Asking Team A to draft even more PA (for the bench), or to rest their players more often than Team B, effectively punishes Team A for being more successful than Team B.  I actually think WIS has solved this very well if they have chosen to make the performance impact of fatigue very light in the 93-99 range. 
6/19/2013 9:24 AM
The flaw is that this only works because you can effectively play your offense at 92-99%. If you could not, then your team wouldn't be effective anymore and you would have to draft a more realistic team according to the proper amount of PAs you need to keep your team at 100% or you would run the risk of seeing negative performance for your fatigued players. The fact that you can play a fatigued player and still be effective is the problem...there is NO RISK of DOWNSIDE to playing a player from 92-99% right now. 

Besides, if the team is "much better designed" for offense, then you can design them to not be fatigued as well. The problem is that under the current fatigue structure WIS ALLOWS you to build a team with far too few PAs to get through the season with players at 100% because fatigue in the 92-100% range doesn't matter.

6/19/2013 9:58 AM
Posted by ncmusician_7 on 6/18/2013 6:20:00 PM (view original):
Posted by frazzman80 on 6/18/2013 4:01:00 PM (view original):

Again...I agree that the fatigue system works for hitters and pitchers from 92% down...it's the 93-99% that is broken. In this league, all of the fatigued players were BELOW 60% for the entire season. They should have gotten killed by fatigue (and they did). The league tested true fatigue and proved it worked.

What it did not test was the 92-99% range that I still feel is not penalized enough. IF 99% acted like 89% then I wouldn't type another word about it.

We agree that 93-99 fatigue doesn't affect hitting enough (if at all), and that severe fatigue affects hitting, pitching and fielding to the point that it isn't a viable strategy.  I'm not sure that 80-92 fatigue affects hitting enough.  Under the current system, I think the ideal team uses their good pitchers at 95+ and their position players at 90+ (with a DH being lower than that).
I see this strategy used more and more...teams just don't rest their hitters until they are below 90%. Their starters just keep starting games because the way the fatigue structure is currently setup there's no real reason to rest their hitters who are above 92%.
6/19/2013 10:06 AM
I think we've reached the point at which we'll need to agree to disagree.  I appreciate your thoughts on this and your input to the discussion.  I see few problems with the current model, and much bigger problems with the solutions that have been proposed to fix it.
6/19/2013 12:41 PM (edited)
Fair enough...the last thing I can do is try to run a test league much like the one biglen setup but with less extreme fatigue. However, I have our first child on the way any day now so I can't play commish right now.

My thoughts would be to set up a standard 24 team with league with the following parameters. Using the 2012 averages of 6139 PA/team and 1445 IP/team as a baseline, teams would be made up of the following:

"Light Fatigue" teams would draft a maximum of 5,000 PA (81.5% of ML average) including their bench with the rule that no position player can start a game at less than 80% (this is the cutoff where people say fatigue starts to affect hitters. This also removes the ultra fatigued scrub as a starter strategy). They would draft a maximum of 1250 IP (86.5% of ML average) with the rule that no pitcher can start a game at less than 85% (this is the cutoff where people say fatigue starts to affect pitchers. This also removes the ultra fatigued scrub as a starter strategy). Breaking the starter percentage rules would result in a forfeits for the entire series in which the player appeared and owners who forfeited 3 series would be playoff ineligible.

"No Fatigue" teams would draft a minimum of 6660 PAs (108.5% of ML average) including their bench with the rule that no position player can start a game at less than 99% (I have to set it at 99% because that's the maximum autorest value and all owners would be required to set autorest for all players at 99%). They would have to draft a minimum of 1500 IP with the rule that no pitcher start a game at less than 99% (I have to set it at 99% because that's the maximum autorest value and all owners would be required to set autorest for all players at 99%). The maximum WHIP for any pitcher drafted would be 1.31 (that is the average WHIP of all pitchers for 2012) to prevent teams from loading up on cheap IP with a high WHIP.

NL East would be all "Light Fatigue" teams
NL Central would be a 2 "Light Fatigue" and  2 "No Fatigue" teams
NL West would be all "No Fatigue" teams
AL East would be all "Light Fatigue" teams
AL Central would be a 2 "Light Fatigue" and  2 "No Fatigue" teams
AL West would be all "No Fatigue" teams.

Ah...crap...I'll just set it up in the classifieds but I will definitely need someone to help me monitor for the forfeits as I know I will be away from a computer for multiple days while at the hospital.
6/19/2013 11:42 AM (edited)
OK...I set it up...I apologize in advance if the birth of my baby causes any issues.
6/21/2013 5:56 AM
   An interesting thing happened to me, fatigue wise. I needed to rest my SPs, and I intended to "throw" a game. In mlb 105544, I started 1910 Kirby White. He started the game at 91%, ended at 7%. Of course, his PCs were set to max. He outdueled a 100% '08 Addie Joss, winning a complete game shut out, 1-0. He walked 8, giving up 4 hits. I understand it wasn't a great game, and doesn't really "say" anyything, but in light of this idea of frazzman's, I felt it appropriate to point out.
8/2/2013 9:26 PM
bump for reference.
10/18/2013 9:17 AM
Posted by frazzman80 on 8/2/2013 9:26:00 PM (view original):
bump for reference.
bump...still broken. I would like to point out that in crazystengel's great WIS interview that even he admits that it was a mistake for him to draft so many IP/PAs in the past.

WIS: How has your strategy evolved during your seven years playing SimLeague Baseball?

crazystengel: I draft far fewer innings/plate appearances now. Once upon a time I was nervous about playing anyone under 100%; now if my team ends the season with no one tired, I figure I drafted badly. At some point it hit home that when I draft a player I'm paying for two things: quantity (number of IP/PA) and quality – and the less I take of the former, the more I can have of the latter.

10/18/2013 9:27 AM
Posted by frazzman80 on 6/19/2013 12:41:00 PM (view original):
Fair enough...the last thing I can do is try to run a test league much like the one biglen setup but with less extreme fatigue. However, I have our first child on the way any day now so I can't play commish right now.

My thoughts would be to set up a standard 24 team with league with the following parameters. Using the 2012 averages of 6139 PA/team and 1445 IP/team as a baseline, teams would be made up of the following:

"Light Fatigue" teams would draft a maximum of 5,000 PA (81.5% of ML average) including their bench with the rule that no position player can start a game at less than 80% (this is the cutoff where people say fatigue starts to affect hitters. This also removes the ultra fatigued scrub as a starter strategy). They would draft a maximum of 1250 IP (86.5% of ML average) with the rule that no pitcher can start a game at less than 85% (this is the cutoff where people say fatigue starts to affect pitchers. This also removes the ultra fatigued scrub as a starter strategy). Breaking the starter percentage rules would result in a forfeits for the entire series in which the player appeared and owners who forfeited 3 series would be playoff ineligible.

"No Fatigue" teams would draft a minimum of 6660 PAs (108.5% of ML average) including their bench with the rule that no position player can start a game at less than 99% (I have to set it at 99% because that's the maximum autorest value and all owners would be required to set autorest for all players at 99%). They would have to draft a minimum of 1500 IP with the rule that no pitcher start a game at less than 99% (I have to set it at 99% because that's the maximum autorest value and all owners would be required to set autorest for all players at 99%). The maximum WHIP for any pitcher drafted would be 1.31 (that is the average WHIP of all pitchers for 2012) to prevent teams from loading up on cheap IP with a high WHIP.

NL East would be all "Light Fatigue" teams
NL Central would be a 2 "Light Fatigue" and  2 "No Fatigue" teams
NL West would be all "No Fatigue" teams
AL East would be all "Light Fatigue" teams
AL Central would be a 2 "Light Fatigue" and  2 "No Fatigue" teams
AL West would be all "No Fatigue" teams.

Ah...crap...I'll just set it up in the classifieds but I will definitely need someone to help me monitor for the forfeits as I know I will be away from a computer for multiple days while at the hospital.
By the way...this theme idea died on the vine because I couldn't find ANYONE willing to take the No Fatigue teams. I even reduced the thresholds of IP/PAs for the no fatigue teams to try to get interest and no one bit. I heard from several owners that "the no fatigue teams didn't stand a chance of winning".
10/18/2013 1:26 PM
As a infrequent player, I read much of this post, and first want to thank everyone for their contributions because they have given me some better insight about how the SIM works.

On that point -- and this was raised in a different way earlier in these posts -- since the SIM is also based on randomness, it would seem unlikely to be able to differentiate the random statistic effects from the fatigue effects.  To do so, you would need to be able to control the random effect and run multiple leagues to really
figure out the fatigue effect.  The statistical randomness means you can't really have a controlled test because two of the factors are varying, one (statistical randomness) in unknown ways.
10/18/2013 2:33 PM
Posted by radlynch on 10/18/2013 1:26:00 PM (view original):
As a infrequent player, I read much of this post, and first want to thank everyone for their contributions because they have given me some better insight about how the SIM works.

On that point -- and this was raised in a different way earlier in these posts -- since the SIM is also based on randomness, it would seem unlikely to be able to differentiate the random statistic effects from the fatigue effects.  To do so, you would need to be able to control the random effect and run multiple leagues to really
figure out the fatigue effect.  The statistical randomness means you can't really have a controlled test because two of the factors are varying, one (statistical randomness) in unknown ways.
I agree with you 100%. The random nature of the game and the engine will make it very hard to differentiate "poor rolls" from "fatigue", but some data is better than none when testing this theory. I've been in enough leagues to be able to draw some conclusions. Horribly fatigued (players/pitchers in the red) teams lose many games. Light fatigue teams (those with pitchers/players above 92%) tend to not lose many games and in fact are extremely competitive due to the higher quality of their PAs/IPs at the cap level in which they are playing.

As for direct stats and how much is randomness and how much is fatigue, I agree it would be very difficult to quantify this.
10/18/2013 3:25 PM
If you ran the same league simulation many times, you could replicate a normal distribution of statistical randomness.  My estimate, based on sampling software, is that you might have to run such a simulated league between 88-278 times before you have useful results......  Let's see....at $ 9.99/teams with the six team discount, and 12 teams per league, that would be between 1,056 teams to 3,336 teams, or $10,549.44 to $33,336.44 -- wonder if you can get a grant for this kind of statistical test?
10/18/2013 7:48 PM
it takes only one season to see the effects of fatigue (on hitters anyways)
10/18/2013 9:00 PM
I would play a no fatigue team and believe I could be very competitive!!!! Let me know if you still want to run the league.
of 9
All Forums > SimLeague Baseball > SimLeague Baseball > Is the Fatigue Code Broken in the Sim?

Terms of Use Customer Support Privacy Statement

Popular on WhatIfSports site: Baseball Simulation | College Basketball Game | College Football Game | Online Baseball Game | Hockey Simulation | NFL Picks | College Football Picks | Sports Games

© 1999-2014 WhatIfSports.com, Inc. All rights reserved. WhatIfSports is a trademark of WhatIfSports.com, Inc. SimLeague, SimMatchup and iSimNow are trademarks or registered trademarks of Electronic Arts, Inc. Used under license. The names of actual companies and products mentioned herein may be the trademarks of their respective owners.