Did some more testing in a save i setup purely for testing purposes (still no editor used, I can't get past that I consider it a cheat, sorry). I took a team in Swe-2 IF Sundsvall. I saved right after a game that had all players, except the goalies and 1 player who wasn't played much, at low enough Con that they wouldn't cap at 100%. I then setup 17 practice schedules using different variations, I gave "Light" a numeric value of 1, "Medium" a numeric value of 2, and finally "Intense" a numeric value of 3. I then made 3 categories, one where all schedules had the total "value" of 13, this is where "Tech", "Tech2" and "Phys" fit in. Another with the total value of 14, and finally one with a total value of 15. The light con + light ska schedule from the 15 category is the exact same as the one from the 14 category, since you can't get more than 2x light + 4x intense. Thus only 17 schedules instead of 18.
Once I had it setup I started by checking if the con gained back is variable, it isn't. Player x will always gain back the same amount of con on the same schedule, barring outside influence. It's got a set value, not a range of values. This made testin a lot faster, since I only had to record the gain on each schedule once, rather than the planned 10 times, making it require 17 restarts instead of 170, which was nice.
What I found was that the effect of different schedules on con return from day to day is very limited. The most lenient schedule, named "13-S" short for 13 (the amount of total numeric value) and S (Slack), which looks like this:
Con: Lig
Ska: Lig
Tac: Med
Sho: Int
Off: Int
Def: Int
and the most extreme schedule, 15-E (Evil), which looks like this:
Con: Int
Ska: Int
Tac: Med
Sho: Med
Off: Med
Def: Med
was rather small, in fact all the 15 point schedules except the two most lenient had the exact same con return on all 20 players. And the only change in those two was that a single player gained 1 more con back in the two most lenient. The other 19 were totally unaffected by the difference.
The only place where a somewhat noticable change could be seen between schedules was in the 13 grouping, where the most extreme schedule 13-E, also known as "Phys" saw an average Con return of 5.1, whereas the most lenient, 13-S, which is the first one I typed out above showed an average return of 5.57 con per day. Which means that the difference was essentially that half the team got back 1 more con per day on the most lenient schedule, which isn't exactly a world of difference.
I still need to take it to extremes by testing a schedule with all training modes set to light, and ones with several set to "none", as well as the old Malhotra all "int" schedule. To see if there is any more difference there. If there is that means that training isn't a zero sum game as some have hinted at in the past. What I mean by this is that if you have all training set to "Intense" it means it will split the training even between all of them, essentially making all "lig" all "med" and all "int" perform nearly identical, whereas a schedule with 3 lig and 3 int would have the 3 lig on a lower level than that, and the 3 int on a higher level than that, the fact that the attribute gain of players is so different on "Tech" and "Tech2" despite the only difference being that i swap the "lig" in ska with the "med" in tac between the two heavily points towards it being said "zero sum game", the shuffling of them internally means that even the parts of the schedule you don't change gets their ratio of the total training altered. Since a "Med" in Ska takes up about twice as much as one in Tec it leaves less training for all the others, regardless of if they are set to "int", "med" or "lig", I'll try and explain why in the next paragraph.
I noticed that a lot of schedules that were seemingly very different, had the same effect on con return. There were 9 schedules that gave an average return of 5.1 con, the two most extreme from the 13 category, the 3 most extreme from the 14 category, and the 4 most extreme from the 15 category. This means that, as expected, having Med/Int in Conditioning (and probably Skating too, an ingame hint points towards it at least) has a much larger inverse effect on your con return than having Intense on the technical/mental based training. It's close to, but not quite, 2x as taxing with training in the physicals as in the others. Due to my chronic illness I'm currently dealing with, doing math problems is a wee bit of a problem. So I didn't manage to find a perfect fit, but it was very close to having no overlap in the training difficulty ranges between the schedules that gave 5.1 and the ones that gave 5.14 when I used 2x value for all physical training, so Int went from 3 to 6, Med from 2 to 4, and lig from 1 to 2 in the numeric value.
Based on limited findings it did seem that having Int con + Lig ska = med con + med ska. There were two schedules in the test that got the exact same numeric values with the 2x physicals, the only difference between the two was one had Int+lig, the other had med+med. Making it likely, but far from guaranteed that it's "the same" for con at least, but from my earlier findings Int+lig gives more attribute gain than med+med does, making it a superior combination.
Note that all this means is that from a con perspective you can have much more taxing schedules than "Tech", "Tech2" and "Phys". It does not mean that using those more taxing schedules actually result in more attribute gain for your players. My current hypothesis, that this is one of the first steps towards proving, is that all players have a minimum and maximum amount of workload. If a schedule is below the minimum, the player will either stagnate or in a worst case scenario even decline. This explains those players who refuse to grow in the summer on the same scheule that had them gain tonnes during the season, it also explains (partially) those players that stagnate/lose if they don't get quite enough game time. They fail to reach their "floor value" since the low ice time+practice schedule doesn't add up to enough workload to satisfy them. On the flipside you have the player that simply will not grow at all on schedules with 2x "int" in the physicals, but grow just fine with 2x "med", but also decline with 2x "lig".
I had that exact player I describe above a few saves ago, Christopher Fish of VIK in Swe-2. When I had him on a 2x med schedule all was well, 2x lig and he lost attributes in everything. Not just con and ska based stuff as you'd imagine. And on 2x int he also lost from everything due to being overworked. It points towards every player having a "sweetspot" where they grow flawlessly, and the further you get to either side the worse it gets, just like if you play a prospect in a league above their level they won't grow (even if they score a lot of points and have a good rating), and if they are in too low a league the same happens. The same system seems to be in place with the practice schedules.
Finally another thing to note on Con return, it seemingly is only connected to the hidden attribute "Natural Fitness", Stamina seems to have no effect, I had a 70 (14) stamina player gain back 7 con from every single schedule, as well as a 49 (9) stamina player gain the exact same (7) on all schedules. I also started off the test by having a physio with relatively high judging player ability (just in case that matters) deliver physio reports on all players to see if it would give me any hints of their Natural fitness, it only reported back on their stamina, workrate and strength though. Based on my findings any player with 35 (7) or above stamina or workrate will be described as being able to do "heavy minutes" <-Stam and "works hard"<-workrate. The next cutoff is at 80 (16), where the return for workrate becomes "Big minutes" (which is really confusing since "heavy minutes" relates to Stamina) and the return for Stamina SHOULD be "Stamina worthy of praise". I say should because the only one who had 80 (16)+ in workrate, strength or stamina was my goalie. He had 81 (16) workrate, 85 (17) Strength and 34 (6) stamina. And he got the "Stamina worthy of praise" and the "isn't strong and it's having negative impact on his game" returned. Which means it's bugged to have it look at strength for stamina value and vice versa, at least for goalies. Can't speak to skaters since I had none with 80+. I've reached the point where I'm finding more than 1 bug per day in this game now, not ideal...
The anomaly at the 13-M schedule isn't due to a typo. One player simply got no con back on that schedule specifically. The number under the average con return is how many players were affected, if the same player were affected more than once he was counted once each time. The same player who had no return on 13-M gained back the same he had on 13-C when 13-L was used, and also got another +1 on 13-S. He had low Natural Fitness (only gaining back 3 per day).
Edit: Just ran the all int, and a few other variations. Results were more or less as expected. But not entirely, running int in both con and skating did indeed have a bigger con return hit than int everything, making it seem like training does indeed have an upper limit. If you have everything on "int", each one of them will be less impactful than if you had a mix of lig, med and int. There are only so many hours of useful training one can do per day, making the "all int" approach impossible to sustain in real life, which is why it's reduced to a possible maximum level of training. All makes logical sense really. It's with all med vs. only med con+ska where things get messy. Here you gain back more con per day on only con+ska than on all med. Which means that medium isn't in fact working out at 0, which I just naturally assumed it to be, it's less than 0, which means that when you are going "very medium" or whatever you want to call having medium in only two different ones, you get a % modifier to a negative value, making it a bigger negative value (duh). That's the only way it makes sense. because if we assume that medium is actually at plain 0; any multiplier to it would have no effect at all. If the modifier was a set number, we wouldn't see the decrease in workload either but rather an increase, so by process of elimination the change when not using all training types at once to super focus on one is a multiplier, which is good to know. So that's the end of that then? No, the freaking all light vs. con+ska light has the exact same con return for both. Which totally ruins that theory, unless it was because all the players hit their maxium possible con return per day (same as on resting) with both schedules, which after further testing is indeed the case. Yes, I should have tested that as the very first thing, but I forgot. This presumably means that if it were possible con+ska lig would return even more con per day than full lig. But that is an entirely theoretical discussion since your players hit a workload low enough to "cap out" their con return per day before that happens in practice (literally). What this means is that if you "need to rest a player" putting them on "resting" is wasting training, since you'd get the exact same con return with a more lax schedule. Exactly where that line in the sand is, I do not know yet, but with some brute force trial and error I could find it quick. I never ever rest my players though, so I don't feel particularly inclined to do so.
The big takeaway i got from all this is that the difference between going all out int everywhere, and going full resting, only results in 1 more/less con returned per day for all but the lowest of Natural Fitness players, I had one guy gain +2 going from 3 to 5 per day, which is still poor. The average in my tests was way lowered due to a few very low natural fitness players, as well as 2 goalies + 1 skater being at 100% con even after playing 3 games in 6 days. So there are always 3/20 players having a 0 gain from any and all schedules every single day. Since all I care about is the difference between the schedules, not the actual average, I didn't bother to cut them from the data. Since it doesn't have any impact on what I actually need to know anyways. The two important factors are: how much did the player who gained the most between the lower/highest workload gain? Answer: 2, how many of them showed any difference from schedule to schedule? Answer: to be found under each column. Generally slightly more than half of them showed any change at all between minimum and maximum for any schedule you would realistically use, which is really insignificant. Enough so that con should have no real impact on choice of schedule. I can't believe I missed that for this long. I just aimed at having a solid 6-8 return on all players, and hit that right away with my schedules (since as I now know, ALL schedules will have good natural fitness players in that range). This won't actually really change anything in how I train players per se, but it does mean I'm now a lot less paranoid about testing higher/lower strain schedules to find some more hidden gems. The downside of this is that I need to rebuild my training spreadsheet to allow for more schedules in use at the same time...darn it!
Doing some other small time thinking leads me to these conclusions:
1. All the 5.14 con return schedules are potentially good schedules for players who have normal Natural Fitness (or w/e attribute decides how much workload they need to improve, I don't freaking know) this is the range where "Tech" is. So far this range only seems to work in 50 game a year leagues.
2. August Gunnarsson, who is the only player I've ever had who realiably, across multiple saves, grows on "Phys"; a 5.10 schedule, should mean that all schedules in that range are potentially good ones for players with high <insert attribute that requires high workload>, but based on the numbers I used to design "Phys" it might be the best of them. I'd consider trying to replace it with 15-E and pray nothing explodes, since Int for Off. Skill is very efficient, and Med for Tac stops it declining. And Phys has proven to not work too well for young players during summers. Nino, does Natural Fitness grow with age at all? I forget.
3. 13-L could prove to be the remedy for players who are too overworked on "Tech", knowing if someone is under/overworked will probably be down to trial and error. If someone plays a lot of 70-80 games a season leagues and wouldn't mind trying it out, It'd be much appreciated. Since that seems to be an environment where players don't grow well on the 5.14 schedules.