Is the Fatigue Code Broken in the Sim? Topic

Are you Grizzly's father?
3/23/2013 1:35 PM
Posted by boogerlips on 3/23/2013 1:35:00 PM (view original):
Are you Grizzly's father?
Ha! No, but way back when during one of my tests on the viability of what eventually became the fatigue strategy, Grizzly was in the league and was interested in what I was testing. After the league ended he sitemailed a few questions on how the strategy could be implemented with winning in mind more than testing, and he perfected it beyond what I even imagined from the tests at that point. There were a few others that sitemailed with similar questions around that same time, but the only one (out of that group) that made any real success in using the fatigue strategy (as far as I'm aware) is Grizzly, he's definitely the master of the fatigue strategy.
3/23/2013 1:55 PM
The student has now become the master. Yay!

3/24/2013 12:45 PM
Posted by boogerlips on 3/21/2013 4:34:00 PM (view original):

"Linear Curve" is an oxymoron. Its just a curve, and one that is probably based on curved value of OBP like their pricing system. I think it would be correct of them to do it that way. The pricing system prices OBP relative to the number of runs a team would score if all 9 members had that same OBP.

The OBP curve very roughly shows that as OBP increases 100pts or so, scoring doubles. So if you wanted to penalize a .500OBP guy 20%, making him a .400OBP guy would be a gross error as that is actually a 50% reduction in value.

I give you that example to help you appreciate the slowness a curve needs to start with in order to be correct. If they start turning .300 hitters into .270 hitters when they are at 90% fatigue, the way to then game the system would be to only ever play 100% players. I'd prefer to have the flexibility of playing fatigued players if I need to, with an appropriate penalty rather a disproportionatly high one.

Great post Boog.  But am I the only person who doesn't think fatigue is broken?

I almost never play a guy below 100% but when I do I can accept the results as fair on average.  You point on OBP (and more generally RC27) is a good one and is exactly why at mid caps or higher (\$60M+) trotting out fatigued players on a regular basis is unwise.  I think you are right, the way to "game" the system is to use players only when they are 100%.
3/25/2013 1:25 AM (edited)
Posted by uncleal on 3/22/2013 9:34:00 PM (view original):
Posted by pfattkatt on 3/22/2013 6:41:00 PM (view original):
I have been a bit disappointed in my performance in the game. One of the things that I believe has hurt me has been my reluctance to play fatigued players. No question, fatigue isn't as big a draw back as it could be. But, it isn't the fatigue model, per se, that i think is the issue. I think it is the pricing system that prices PAs too high. The stats we can purchase using ~550 or PAs are quite a bit better than a 700 PA player. At 90%, he is still a much better player. That is the glitch, IMO. '95 Maddux or '00 Pedro at 90% are still very good pitchers. If they are priced accurately, than IPs are too expensive, and the stats that make the superstuds into superstuds need to be priced differently. I think the fatigue model works reasonably well. It is the pricing of the stats that lead to fatigue that need tweaking.

A PA cap for cost might be nice.

While theoretically it's possible to require 884 PAs to stay at 100%, that involves max offense every game which isn't very likely in the SIM. In my mind, a more reasonable number might be somewhere around 663. That is to say, make it so that all PAs beyond the first 663 are free. (That's 4.5 PA/g over 162 games, which is midway between "max offense for fatigue purposes" and "minimum PA for a full game", with the 10% "bonus" accounted for as well.) This would take off some of that edge, because unnecessary PA can make some players overpriced...

Another option would be to curve the \$/PA model, changing it at various thresholds, to make guys in the "nearly enough" range cost more, but I think this would almost certainly end up overcorrecting.

FWIW, I typically will allow hitters to play to 95, but no lower, unless it's the playoffs, in which case I might let them play lower on a case-by-case basis.

At one point (I think it was still there in 2007) there was a cap on PA.  Basically, the cost didn't increase (or the cost increase was severely reduced with regards to \$/PA) once a player got to 600 PA.  The problem with that is it kinda eliminates the platoon strategy for position players.
3/24/2013 10:53 PM (edited)

I'm on board with frazzman here....

If I pay for 550 PA.... Wow!  I get a bonus of 10%.  That's about 605 PA.

I didn't pay for 650.  Or 700.  I paid for 550.

If I go on pace to exceed the 605 (with bonus), the player should start struggling big time.  I WANT the first drop to be steep.  Going from 100% to 99% should be a big deal.  Why?  Because you're gaming the system to get more than you paid for.  While it shows up as only 1%, I'd like to see it a 10% drop.. Go from a .300 hitter to a .270 hitter when you're a bit tired.  And then I'd drop another 3% for every 1 percent drop in fatigue below 99.  By the time a player hit's 60%, you;d have better luck putting a quadrouple amputee in the batters box.

And yeah, I want to see injuries start faster too.

Why such strict wants?  Because anything else is gaming the system, getting more than you paid for.

I always hear the argument that "No MLB player is ever 100%".  Yep.  Most likely true.  But it has absolutely zero do do with this game.  WIS is not MLB.  It is a computer game. The 100% is really not a real person's fatiuge state, it's a rating that says "Player x produced stats at this rate for 605 PA".    If you exceed the PA that you paid for, then you should expect that player's stat accumulations to drop off.

It doesn't happen.  But it should.

3/25/2013 11:21 AM (edited)
If a team full of 90% hitters only scores 90% of the runs they normally would, how is that gaming the system?
3/25/2013 11:36 AM
I wish admin would just publish the fatigue curve.  That would end all this nonsense.
3/25/2013 3:35 PM
Posted by boogerlips on 3/25/2013 11:36:00 AM (view original):
If a team full of 90% hitters only scores 90% of the runs they normally would, how is that gaming the system?
I think the point is some people think you get the same performance over the same PAs by drafting better players and playing them fatigued.

i.e. to simplify your obp example the argument is:
It is cheaper to draft 550 PAs of a player with a .444 obp and play him down to 90% performance, effectively getting a .400 obp over 660 PAs (550 *1.20) than it is to draft 600 PAs of a .400 obp guy and use him at 100% for 660 PAs.
3/25/2013 3:44 PM
the only way i would agree with frazzman's model is if it went back to the old way of once you exhaust all your PA's then you severely fatigue but I think also that in the playoffs everyone should be back to 100% and you get a % of your PA's because those are really unanticipated pa's and ip's because totals are based on regular season only

3/25/2013 4:11 PM
Posted by zubinsum on 3/25/2013 1:25:00 AM (view original):
Posted by boogerlips on 3/21/2013 4:34:00 PM (view original):

"Linear Curve" is an oxymoron. Its just a curve, and one that is probably based on curved value of OBP like their pricing system. I think it would be correct of them to do it that way. The pricing system prices OBP relative to the number of runs a team would score if all 9 members had that same OBP.

The OBP curve very roughly shows that as OBP increases 100pts or so, scoring doubles. So if you wanted to penalize a .500OBP guy 20%, making him a .400OBP guy would be a gross error as that is actually a 50% reduction in value.

I give you that example to help you appreciate the slowness a curve needs to start with in order to be correct. If they start turning .300 hitters into .270 hitters when they are at 90% fatigue, the way to then game the system would be to only ever play 100% players. I'd prefer to have the flexibility of playing fatigued players if I need to, with an appropriate penalty rather a disproportionatly high one.

Great post Boog.  But am I the only person who doesn't think fatigue is broken?

I almost never play a guy below 100% but when I do I can accept the results as fair on average.  You point on OBP (and more generally RC27) is a good one and is exactly why at mid caps or higher (\$60M+) trotting out fatigued players on a regular basis is unwise.  I think you are right, the way to "game" the system is to use players only when they are 100%.
I also don't think fatigue is broken in that respect whatsoever. I think fatigue is broken in that the way they prorate and calculate fatigue (especially in respects to early-season and late-season fatigue, and the DAMP effect), the fatigue model can be gamed. Likewise, the model doesn't seem to effect all stats equally; and as such, can be gamed. If fatigue were to be calculated more linearly with the same effect curve and equal distribution of the penalties, then I'd think it were an excellent system as is.
3/25/2013 11:37 PM
Posted by just4me on 3/25/2013 11:38:00 PM (view original):
Posted by zubinsum on 3/25/2013 1:25:00 AM (view original):
Posted by boogerlips on 3/21/2013 4:34:00 PM (view original):

"Linear Curve" is an oxymoron. Its just a curve, and one that is probably based on curved value of OBP like their pricing system. I think it would be correct of them to do it that way. The pricing system prices OBP relative to the number of runs a team would score if all 9 members had that same OBP.

The OBP curve very roughly shows that as OBP increases 100pts or so, scoring doubles. So if you wanted to penalize a .500OBP guy 20%, making him a .400OBP guy would be a gross error as that is actually a 50% reduction in value.

I give you that example to help you appreciate the slowness a curve needs to start with in order to be correct. If they start turning .300 hitters into .270 hitters when they are at 90% fatigue, the way to then game the system would be to only ever play 100% players. I'd prefer to have the flexibility of playing fatigued players if I need to, with an appropriate penalty rather a disproportionatly high one.

Great post Boog.  But am I the only person who doesn't think fatigue is broken?

I almost never play a guy below 100% but when I do I can accept the results as fair on average.  You point on OBP (and more generally RC27) is a good one and is exactly why at mid caps or higher (\$60M+) trotting out fatigued players on a regular basis is unwise.  I think you are right, the way to "game" the system is to use players only when they are 100%.
I also don't think fatigue is broken in that respect whatsoever. I think fatigue is broken in that the way they prorate and calculate fatigue (especially in respects to early-season and late-season fatigue, and the DAMP effect), the fatigue model can be gamed. Likewise, the model doesn't seem to effect all stats equally; and as such, can be gamed. If fatigue were to be calculated more linearly with the same effect curve and equal distribution of the penalties, then I'd think it were an excellent system as is.
Do you mean fatigue of effectiveness?  Untill at least about 90% fatigue is very linear.  For example, a guy at 90% fatigue is about at 110% of his alotted PAs.
3/26/2013 11:36 PM (edited)
To me, it's the 10% bonus that is throwing off the fatigue model and the real reason why I think fatigue should start dramatically impacting players at 99%. Like biglen noted above, you pay for a players PA/162 and the price of that player's PAs/162 is built into his salary. Then, you get a full 10% additional PAs without paying for them and can play a player at 100% when he is a full 10% ABOVE his real life PAs/162.

To me, this is like real life and counters all of the "real players play fatigued all the time arguements" becuase you are already playing your players fatigued. WIS is just showing you that they are 100% (on pace for 10% above). WIS is essentially saying that "you are playing this guy fatigued right now, but because it's not excessive fatigue (in this case, less than 10% above their real life PAs) we are not going to impose a fatigue penalty for this player."

When that players hits 99%...he should start experiencing negative fatigue effects. He is truly fatigued at this point becuase he has played past the 10% bonus. It should get significantly more negative for each percentage point beyond that. HE IS FATIGUED...not tired, not playing through pain, not bumped and bruised...that was during the 10% buffer.

It works this way already for pitchers. They have 3 seperate messages for pitcher fatigue. You have the roll over text that says the following:

• 100% - (I can perform as expected based on my stats.)
• Tired from recent game activity (Not fatigued...just tired. If my fatigue level isn't below 90, it's a bit of a gamble to pitch me, but my fatigue will get worse and you won't be able to use me effectively for a while.)
• Overworked - On pace for X% more pitches than actual pitches/162 total (I'm slightly fatigued...but still in my buffer range. It's a roll of the dice how I will perform and you'd better rest me or I'll be terrible the next time I play without rest.)
• Overused - On pace for X% more pitches than actual pitches/162 total (I'm fatigued...pitch me and I will more than likely play like crap or be so fatigued that next time you better not even think about using me. Rest me and I'll be usable again).

Why doesn't it work this way for hitters? As of now...it doesn't work this way at all...we have the following:

100% - (I can perform as expected based on my stats.)
100% - On pace for (up to) 10% more PAs (XXX) than PA/162 total - (I can perform as expected based on my stats.) - Read NO PENALTY
99% or more - Overused -- On pace for XX% more PAs (XXX) than PA/162 total  - Exact CS response: (Our fatigue system is non-linear and the effects are percent adjustments off their core ratio stats. A player at 99% is pretty close to 100%, but a player at 60% is a shell of his 100% self. It's an exponential decay.)

So with pitchers, we see the 10% buffer as Overworked (and see a high level of negative performance) but for hitters the 10% buffer shows ZERO fatigue. This is inconsistent at best and rediculous at worst.

Combo in the zubs statement above and you can see how the system is broken:

"I think the point is some people think you get the same performance over the same PAs by drafting better players and playing them fatigued.

i.e. to simplify your obp example the argument is:
It is cheaper to draft 550 PAs of a player with a .444 obp and play him down to 90% performance, effectively getting a .400 obp over 660 PAs (550 *1.20) than it is to draft 600 PAs of a .400 obp guy and use him at 100% for 660 PAs."

If that 550 PA .444 OBP player started seeing a major hit to performance at 99% (and then even larger ones at 98, 97 and on down), he should be so fatigued at 90% that it shouldn't be effective to play him and that resting him would be of utmost importance...just like a pitcher who is fatigued.

That is the point I am trying to make. Zubs hitter example right now is performing FAR too well at 90% based on the current fatigue curve. Owners are not resting that kind of player. He is playing every game and IMO there isn't enough performance drop off to consider resting him with the current fatigue system.
3/26/2013 11:37 AM (edited)
Posted by biglenr on 3/25/2013 11:21:00 AM (view original):

I'm on board with frazzman here....

If I pay for 550 PA.... Wow!  I get a bonus of 10%.  That's about 605 PA.

I didn't pay for 650.  Or 700.  I paid for 550.

If I go on pace to exceed the 605 (with bonus), the player should start struggling big time.  I WANT the first drop to be steep.  Going from 100% to 99% should be a big deal.  Why?  Because you're gaming the system to get more than you paid for.  While it shows up as only 1%, I'd like to see it a 10% drop.. Go from a .300 hitter to a .270 hitter when you're a bit tired.  And then I'd drop another 3% for every 1 percent drop in fatigue below 99.  By the time a player hit's 60%, you;d have better luck putting a quadrouple amputee in the batters box.

And yeah, I want to see injuries start faster too.

Why such strict wants?  Because anything else is gaming the system, getting more than you paid for.

I always hear the argument that "No MLB player is ever 100%".  Yep.  Most likely true.  But it has absolutely zero do do with this game.  WIS is not MLB.  It is a computer game. The 100% is really not a real person's fatiuge state, it's a rating that says "Player x produced stats at this rate for 605 PA".    If you exceed the PA that you paid for, then you should expect that player's stat accumulations to drop off.

It doesn't happen.  But it should.

Maybe the other way to explain my thoughts on the fatigue idea is to think of this way: in my mind, a player that hits 99% should play like a player WIS currently shows at 89%. At this point, he is fatgued and is 11% over his real life PAs. Not 1% (although he is showing 99).

IMO, if WIS just made 99% players play like a player who is 89% under thier current fatigue system, then the whole fatigue problem would be resolved because everyone agrees that a player who is at 89% under the CURRENT fatigue system should be rested. That would resolve the whole problem for me
3/26/2013 11:11 AM
I fail to see why the 10% bonus thing is such a hangup for you.  It applies to everyone, it's not like it's giving a special advantage to some guys and not others.  Everyone is just allowed to play 10% more than they did in real life.  And in spite of your confusion, resulting from the poor way in which things are worded in the manager's center, it works the same way for position players and for pitchers.  For whatever reason, when fatigued/overworked pitchers show their % above their sim-allowed pitch count, while position players show their use as a % above RL PA/162.  However, that sim-allowed pitch count is based on their RL IP*1.1.  In other words, a pitcher who shows up as being on pace to throw "2% over actual pitches/162" when you hover over his fatigue is actually on the same usage schedule as a position player who shows up as being overworked, on pace for 12% over RL PA/162.  Don't know why, that's just the way it's always been.  But the model is actually the same for both.

Again, though, I don't see why this is a problem and bothers you so much when it works the same for everyone.

3/26/2013 2:07 PM
◂ Prev 1|2|3|4...10 Next ▸
Is the Fatigue Code Broken in the Sim? Topic