Soft Cap Calculation questions

[mulleto22]

Hi everyone,

I am currently playing another GC East run on Manstein with 1.20 rules (using 1.21 patch). Using the Softcap, I am planning tight in each scenario which units I can use so that I get max prestige. Being in Stalingrad now, I had to go a little higher with my Unit costs, as I want to win DV on every scenario. I am noticing though, that I get less prestige than I expected which brings up the question, whether I understood Soft Cap rules wrong.

Ok, here is what I understood, based on this thread: viewtopic.php?f=125&t=42325

If my average "unit value" is for axample at 500, the Soft Cap coefficient will be calculated as followed (using default gamerules.pzdat):
(800- 500) / (800-400) = 300 / 400 = 75%. This means, I would get 75% of the earnings when capturing cities etc.

To plan exactly, I set up an excel sheet which calculates the unit values for me, and in order to max out prestige, I am planning carefully how much I overstrength etc. I am currently in GC E42 Stalingrad Ruins, and the average value of my units is 440 points. Using the above formula, I should be getting 90% of normal prestige, like 45 prestige for capturing a non-VH city.

Instead, I am only getting 76%, though. I have no explanation for the reasons. I checked whether SE Units are being accounted for, which is not the case, disbanding them makes no difference. Disbanding some of my core units does make a difference, its just that I don't understand where I am calculating wrongly, or where my misunderstanding in softcap rules is located.

My unit values are being calculated base cost / 10 * strength, so a unit with cost 300 and 15 strength is being counted with 450 prestige. I checked my excel again and again, no mistakes here for sure. Average value of deployed units is 440. Ok, let's suppose for a moment that I made a mistake: to get to a 76% soft cap, I would need an average unit value of 495. Thats 55 extra value on 37 units deployed, or a total of 2035 prestige value in my units. No way I could accidentally have an error this large.

So I guess I am misunderstanding something in Soft Cap calculation. Is anyone here able to point out the point that I am missing?

[Tarrak]

Are you taking into account the value of your transports attached to your units? If you don't their cost can easily add up and cause a quite big deviation. After all considering you are using the first halftrack transport it's 100 additional prestige per unit using it.

[mulleto22]

Hey Tarrak,

yes, I did consider the transports. This issue has caught my attention and I did some further testing. Indeed, I found out that there is an ever increasing deviation from the value that would be calculated from the Softcap formula that was presented in the post explaining the 1.20 changes.

Referring to the formula presented in that post, for example it would be expected, that if you deploy units worth 600 prestige, you would get 50% of normal prestige. Instead, it is just 22 percent!

The following table shows the percentage values for a set of different deployments which I expreimented with:

Unit value -- Expected % -- REAL %
402 -- 99,5 -- 98
431 -- 92 -- 82
444 -- 89 -- 76
523 -- 69 -- 50
569 -- 57,8 -- 32
588 -- 53 -- 28
607 -- 48,25 -- 22
626 -- 43,5 -- 20

It is quite obvious that something must be wrong here: Gamerules.pzdat state that the min koefficient is 0,2 and should be reached at 800 unit value. I am attaching a savefile from my GC42 game. The most expensive units are some fw190 wurfrahmen. Their unit cost at strength 14 is 798 and 714 respectively. So it is quite obvious that I cannot get to an average value of 800, as there are no such expensive units at this point. The average value of units used in the savefile is 626, and if you capture a city hex with this savefile, you'll see that you only get 10 prestige instead of 50, meaning the coefficient is already at 0,2 when it should be 0,43 (watch out, savefile is deployment phase).

To me, this starts to look like a bug. But once again, even though calculation of soft cap is not rocket science, maybe I am overlooking something. But at least, it does not work according to the formula that has been presented to us.

[Tarrak]

Could you attach your excel file with your calculations as well so i can have a look at it? As you said there may be something you oversaw but i don't rule out a bug in the game as well. I don't think anyone went as far as you are doing it now with testing. Something is rotten in the state of Denmark ... aehm ... in the calculations of the soft cap.

[mulleto22]

Hey Tarrak,

thanks for looking into this, I appreciate that. Heres my excel file, feel free to contact me if you have questions. I added some comments, this is one of these over-time-grown files where only its creator understands what is in there and for which reason it is in there

[alkafluence]

My unit values are being calculated base cost / 10 * strength, so a unit with cost 300 and 15 strength is being counted with 450 prestige.

IIRC - That isn't correct. Each point over 10 in overstrength doesn't cost the same with the new rules. It gets exponentially higher in prestige cost with each additional overstrength point added.

[Tarrak]

My unit values are being calculated base cost / 10 * strength, so a unit with cost 300 and 15 strength is being counted with 450 prestige.

IIRC - That isn't correct. Each point over 10 in overstrength doesn't cost the same with the new rules. It gets exponentially higher in prestige cost with each additional overstrength point added.

Good spotted. Yes the overstregth costs are rising with each point. Can you check if this can be the cause of the deviation muletto22? I won't be able to have a deeper look at it before saturday due to work and real life commitments.

[mulleto22]

Good spotted. Yes the overstregth costs are rising with each point. Can you check if this can be the cause of the deviation muletto22? I won't be able to have a deeper look at it before saturday due to work and real life commitments.

I had checked this in advance, since this was one of the possible explanations. But it is definately not the reason for the deviations. There is a difference between overstrength costs and the unit values used for soft cap calculations.

First, it is clearly stated in the patch 1.20 notes that a unit with strength 15 ist valued at 1.5 its price (see link above, where it says: "So, a normal 10-strength unit at full OS is 1.5 of its normal cost.").

Second: Yet I checked whether the explanation is wrong and calculated soft cap using overstrength costs. This leads to enormous values. The core provided in my savefile would have a value of 994 (!!!) prestige, which would mean the softcap would kick in with the minimum value of 20%, and not the 76% that I am actually getting. So even IF overstrength costs would be the base to the calculation, there would also be an error in the results.

Third: as shown in the table in my second post, deviation from expected values to real values increases with rising unit value. Whatever the base formula for calculating unit values may be cannot lead to such results. If OS costs where the reason, it would simply raise soft cap penalty to another level, but not create an exponential function.

I won't be able to have a deeper look at it before saturday due to work and real life commitments.

Tarrak, take your time, no problem. Just one hint: in my savefile, simply deploy more or less Wurfrahmen to change values quickly, then just start the scenario, occupy one of the open city hexes and you'll see my point that you are getting too little prestige.

[Horst]

I've also done a prestige cap test in the past with following results:

UseSoftCap 1
NormalPrestige 400
MaxPrestige 800
MinKoff 20

Here's an overview which average unit cost will give what prestige when capturing a victory hex. It should give a good impression how quickly the prestige will drop by just passing the NormalPrestige barrier with medium standard tanks.

Cost: Prestige gain by capturing victory hex (vanilla unit values)
0: 100
1: 100
400: 100 (~ Panzer IIIL)
450: 77
500: 56 (~ Panzer IVJ)
550: 39
600: 25 (~ 15-strength Panzer IIIL)
650: 20 (~ Panther A)
700: 20 (~ Tiger I)
750: 20
800: 20
850: 20
900: 20
950: 20
1000: 25 (~ Tiger II or 15-strength Panther G)
1050: 39 (~ 15-strength Tiger I)
1100: 56 (~ Maus)
1650: 452 (~ 15-strength Maus)
2000: 900 (Battleship)
3000: 3025 (15-strength Battleship)

At a certain point you start to gain more prestige again if you climb too far ontop of the MaxPrestige setting. You are only getting 20 points with 650 average although you haven't touched the 800 yet.
Well, something seems not right about the soft cap calculations. I haven't delved much further into this and simply changed the settings to my mod's 10-strength T-34-85 cost as NormalPrestige and 15-strength Maus cost as MaxPrestige setting. End of story for this soft prestige cap stuff for me.

[mulleto22]

Horst,

thanks for the info, your results are mirroring mine. And they gave another info which leads to the solution of our mystery formula... in a second.

Tarrak, since Horst and I have the same results, I think you don't need to check out whether I made a mistake in my testing.

And now the solution: It was so obvious, I already mentioned an exponential function in my former post, but I needed to see Horsts numbers to realize how the calculation is made:

While the post explaining 1.20 changes explaines the formula like this:

(MaxPrestige-AvgUnitValue) / (MaxPrestige-normalPrestige) = Coefficient

the REAL formula that does the calculation magnifies (is that correct? - sorry I'm german) the result on both sides. And the really used formula is:

((MaxPrestige-AvgUnitValue) ^ 2 ) / ((MaxPrestige-NormalPrestige) ^ 2) = Coefficient <== See the exponent on both parts of the calculation

Put this formula into an excel sheet and you'll get the results Horst and I had. Even better, look at the file I attached, fill out the yellow field. I am sure this formula is what someone will find who delves into the sourcecode that implements the softcap calculation.

Now - this leaves a question: Maybe its been intended to work like this all way long. Which would mean, the SoftCap would hit its max around 630 Unit Value, which can easily happen early in 1942 if someone does not plan his core. In 1943, with all the upgrades necessary to go on successfully, even a carefully built core will hit 630 and the max SoftCap penalty, meaning almost everyone would have to play 3 years at minimum income. Actually, from all that was written in the 1.20 patch notes about the soft cap, it was intended to reach max penalty at 800, not at 630. Otherwise, the maxPrestige setting in gamerules.pzdat would be erroneously used.

Tarrak, I never posted a bug report to the developers, I suppose you have a channel to post to?

[Tarrak]

Tarrak, I never posted a bug report to the developers, I suppose you have a channel to post to?

Actually no there is no such channel i am aware of. Just make a post in the open beta or the main forum with a proper and descriptive title and post the explanation or a link to this thread and hope it get spotted. You may as well send a private message to Rudankort, who is the main programmer of Panzer corps and point him to the thread to rise the chances. As far i know he is really busy atm tho with the Warhammer 40k game due to deadlines so it may take a while until he can have a look at this.

[mulleto22]

Hi Tarrak,

ok, I'll try my luck. Thanks for lending me an open ear in this mess!

[captainjack]

which can easily happen early in 1942

I've been affected by the soft cap in 1942, and I suspect it may have been affecting me in 1941 in that playthrough.

It sounds like you've picked a tough challenge (mathematically as well as tactically) - good luck!

[ThorHa]

I love people who test things. The results of Horst clearly indicate why the soft cap is discussed so often - it is much worse than its description indicates. The real cap is not 800 but 650, which is just the value of an ordinary strength 10 Panther. Furthermore Horsts published values clearly indicate the SoftCap has no linear curve but sets in with a venegeance as soon as you cross the 400 treshold.

For PC users this can be remedied, ipad users are stuck with the game as is.

Regards,
Thorsten

[Horst]

I pity all "console" gamers who can't play with mods this game.

I recommend turning the cap off for playing the game. Even if the soft cap mechanic would work as intended, most campaigns, vanilla or custom, are designed with <=1.14 rules in mind. The DLC 45 scenarios are already harsh enough in the end without it if you face hordes of 3-star, 12+ strength tanks later.
What people seem to forget is that the overstrength costs were also highly increased with 1.20. Before, you only paid 100% of 200% elite replacement unit cost for each OS point, now you need to pay 100/150/200/250/300%. Good for the AI that it doesn’t really care about this cost with preplaced units! This will be much more noticeable later when you need expensive, heavy tanks to survive and prevail. Campaigns are typically designed with an increasing average experience for all AI units and will later throw 3+ star, 12+ strength units against your core. Then you are thankful for every prestige point you have saved earlier.

[ThorHa]

Horst, I am in the middle of the 44 campaign now and only losing a bit of money while overstrengthening and upgrading to Tiger II. A still healthy 44 k bank of prestige in hand.

That having said I only managed to do that by good gameplay, abusing game mechanics, planning to be there from 39 forward and using the SE units more or less excluisively for high level tanks. Without the common wisdom amsassed in the forum I would already struggle (in my first playthrough until dlc 43 I only (!) managed to reach 28 k in prestige at the start of 43, consecutively abandoning the campaign and starting from a 39 save game).

Which means - the devs have simply forgotten that there are newcomers to the game without the built in instinct to browse all of the forae for every bit of information available. Right now the 1.20 ruleset is broken for any difficulty level equal or above colonel. Main reason being soft cap.

Regards,
Thorsten

[captainjack]

I play without soft cap. Even then 44East can be hard going - so much so that I have not yet made it through to 45 East, although I have completed 45 West on the path that doesn't involve Sealion 45. But I have at least learned what I'm doing wrong (mostly mis-timing my counter attacks and over-estimating the defensive qualities of 4* grenadiers with defence heroes) and have some idea what to do about it next time round.

With soft cap on I struggled in 43 East and ended up with a net loss of prestige in every scenario so that I was facing 44 East with no hope of success. Even allowing for some improvement in play style since I disabled the soft cap, that's quite a big difference for a reasonably experienced player.

[mulleto22]

Right now the 1.20 ruleset is broken for any difficulty level equal or above colonel. Main reason being soft cap.

This is my main concern also. For any newcomer who does not know what he has to expect, it can become quite a disappointance to find a game that becomes pretty much unplayable from DLC E43 on.

I have contacted Rudankort via private message on this issue, don't yet know what to expect. If anything of interest comes up, I'll post it here.

[Horst]

It's indeed a huge prestige saver when you (re)play campaigns and know what to expect where and when.

Allied Corps is a campaign that should work fine with the default cap settings, as you start getting absurd amount of prestige from mid to end that you could theoretically loose and repurchase all your units in every scenario.
I’m already curious if Soviet Corps will be designed and tested with the cap. I bet many will find having mainly KV tanks in the core very appealing from 1941 on that could trigger the cap mechanic earlier than playing as Axis or Western Allied power.

[BiteNibbleChomp]

It's indeed a huge prestige saver when you (re)play campaigns and know what to expect where and when.

Allied Corps is a campaign that should work fine with the default cap settings, as you start getting absurd amount of prestige from mid to end that you could theoretically loose and repurchase all your units in every scenario.
I’m already curious if Soviet Corps will be designed and tested with the cap. I bet many will find having mainly KV tanks in the core very appealing from 1941 on that could trigger the cap mechanic earlier than playing as Axis or Western Allied power.

If that occurs then this is what I'll do:

Min Cap Prestige: 99998
Max Cap Prestige: 99999

- BNC