Some Information on Scaling

[Aetius]

Someone may have already figured this out, but it's new to me and pretty important, especially for players on Legendary, so I thought I should post it.

tl;dr - Add small, weak units to your army roster in order to reduce battle scaling.

My hypothesis is that the "battle value" for each unit is calculated by multiplying unit strength by an experience multiplier. My initial, very rough guess on these multipliers is 0.8 for zero-star units, 1.0 for one-star units, and 1.2 for two-star units - not sure on three-star units. The scaling is based on the average battle value of all the units in your army. This means that a 2,000-man two-star brigade (bv: 2400) and a 500-man zero-star brigade (bv: 400) are treated by the scaling system as bv 1,400, or nearly half what the value would be with only the 2,000-man brigade. Thus by adding these "ballast" units, you can drag down your average battle value and remove thousands of troops from the enemy roster without firing a shot. These units don't even have to go into battle, although it would probably be beneficial if they did, because casualties would reduce their size, further pulling down the average battle value.

With the caveat that these are still very rough numbers, and I haven't done a thorough investigation:

 

[Fred Sanford]

Very interesting.  Although I'm not sure if "battle value" is really a thing, or if it is simply a matter of numbers.  i.e. I wonder if a 500 man 3 star acts the same as a 500 man zero star so far as scaling.

[Bramborough]

Another thing I've wondered is how artillery is treated in scaling.  Star quality aside, is a 24-gun battery seen simply as a 600-man unit, or equivalent to a 2500-man brigade?

[Aetius]

1 hour ago, Fred Sanford said:

Very interesting.  Although I'm not sure if "battle value" is really a thing, or if it is simply a matter of numbers.  i.e. I wonder if a 500 man 3 star acts the same as a 500 man zero star so far as scaling.

I don't think it does, but it's definitely worth investigating. I did some earlier testing while trying to figure this out. 4 two-star brigades with 1450 triggered scaling. However, disbanding and re-creating those units as one-stars permitted me to bump them up to 1750 before the scaling hit. It's definitely possible that the multipliers aren't fixed, but are instead just based on the unit's base xp.

30 minutes ago, Bramborough said:

Another thing I've wondered is how artillery is treated in scaling.  Star quality aside, is a 24-gun battery seen simply as a 600-man unit, or equivalent to a 2500-man brigade?

As near as I can tell, it works the same way, but I think they are siloed - at least, adding large numbers of infantry didn't add any additional guns in this video, and removing the one 3-gun unit did add a gun to the scaling. I think artillery might be harder because the units are smaller and more incremental, but I'll have to test it.

[Aetius]

After some testing, it appears that the weighted average only applies to minor battles. For grand battles, the system only appears to consider the total number of troops, and possibly the number of brigades. This results in a pattern where you create and disband the ballast units as needed. Also, artillery is indeed a separate category, and works the same way.

[LAntorcha]

About scaling... Legendary CSA Newport News (3 Brigades allowed):

If you wear no Arty nor Cavalry, Enemy Arty and Cavalry stays low. With minimum 500 strength 3x Infantry Brigades, Union deploys about 9k troops, Infantry Brigades being 1100 men and arty units are 8 6Pdr Field Guns.

If you max out 3x 2500 Men Infantry Brigades, arty stays the same, but Infantry Brigades goes to 2900+- strength, Overall goes up to 13k. Fixed scenario troops also count towards the scaling, so there is a minimum.

If you swap an Infantry Brigade by a 12 gun arty, their arty goes to up to 20 gun per battery. If you bring a Cavalry unit, their Cavalry comes up in strength also.

Every time you go back to Camp and make changes, IA makes changes too.

Not sure about exact strength ratios... but Newport News is the best scenario to experiment, being the first that you are allowed to create units. Save at Camp, and disband units, create new ones, etc... Go to battle with 0 Brigades and check strength...

 

So my theory is having low strength unit in the rooster does not help. IA takes in account your Total Order of Battle strength (this means, the strongest Force you can bring). @Aetius is correct. Ballast units help downgrading scaling.

Edited March 25, 2018 by LAntorcha

[pandakraut]

I've gone through the early part of both the Union and CSA campaigns on multiple difficulties and ballast units seem to be of limited usefulness prior to Shiloh. I suspect that either the number of brigades or total force size limits are so low early on that ballast units aren't worth it. Going with smaller size units early definitely can keep scaling down though.

While the standard rules normally hold true(ballast units are useful in side battles, not useful in major battles, and more numerous smaller units can keep enemy unit sizes down) I would recommend experimenting with every method on all battles. You'll notice very quickly if they are having an impact and sometimes the normal rules are not followed.

I've hit side battles where ballast units trigger scaling. I suspect this is due to building up for a major battle and then being over some total force size limit for the next side battle. The multi day battles that only show the starting strength are particularly problematic since you can't probably tell if you're triggering scaling.

I've also hit major battles where ballast units somehow do reduce scaling. I've done extensive testing on Union Antietam MJ and ballast units never helped. But on my last run through of Antietam as CSA on legendary they did reduce scaling. I've even seen cases where adding men to core units reduces scaling when adding a ballast unit would increase it. Maybe it has something to do with the army size you come in with or how much you have depleted the opposing manpower pool coming in, but I haven't been able to find anything consistent.

 

[LAntorcha]

45 minutes ago, pandakraut said:

Maybe it has something to do with the army size you come in with or how much you have depleted the opposing manpower pool coming in, but I haven't been able to find anything consistent.

 

 

Me neither, that's why i wrote it didn't help at first glance. Then I edited.

No clues yet.

[pandakraut]

Reposting my data on Antietam as the Union from reddit

 

[LAntorcha]

Tried this for the Ballast theory. Didn't work 100% denying Scaling.

Or else I hit the minimum cap at Legendary CSA Siloh (roughly 2 vs. 1 disadvantage)...

I think the problem is the fixed scenario 2000-3000 men brigades from Bragg & Breckinridge.

Edited March 27, 2018 by LAntorcha

[pandakraut]

The scaling at Shiloh and other battles that feature allied units definitely seems to be impacted by the allied unit sizes. You can absolutely see it in the size of enemy units even if numerical scaling is kept down.

Using small unit sizes(1100 men each) I was able to get a bit better than a 2:1 ratio on CSA Legendary. 39359 men 113 guns vs 72902 men 208 guns. This was with the realism mod installed in case that has any impact outside of ease of getting kills and reducing manpower.

 

[LAntorcha]

Not directly related to Scaling but found this interesting info related to AI behaviour :

 

 

[Grimthaur]

@Aetius We discussed this in Oktober alrdy. On Legendary the AIs army scales up and down depending on your armies strength. My point at that time was, that you shouldnt use up all available menpower and weapons because the AI will counter the numbers. So if you scratch together your last reserves with farmers or M42s, you will essentially weaken your army compared to the AI cause it adds the numbers scaling to your troops but with better rifles. Obviously it works the other way around - not that that would be needed to win legendary.

Problem is: You have to know the minimum army-size, the ai will deploy each battle, because the scaling only comes into play beyond that point. The displayed number between battles doesnt help, its almost always way too low. But if you knew these numbers,  you could really abuse legendary (but why would you?).

 

 

[Aetius]

On 4/1/2018 at 7:09 AM, Grimthaur said:

@Aetius We discussed this in Oktober alrdy. On Legendary the AIs army scales up and down depending on your armies strength.

I've established that the current scaling does not work that way for minor battles. And pandakraut's tests show that it doesn't quite work that way for major battles either. My last Brock Road battle is a great example - before the battle I was able to reduce the enemy forces from 61,000 troops (above the intelligence report max size) to 41,000 troops by adding ballast brigades and converting unneeded line brigades to ballast. You'll also notice that in the end of battle results, I outnumbered the Confederates 50,000 to 34,000. If the scaling was based simply on numbers, this would be impossible. I increased my total army size to reduce the number of deployed enemy troops.

This changes the dynamic. For minor battles, you need to retain a lot of cheap weapons to build ballast brigades to reduce your average brigade size. For major battles, you should bring as many troops as you can up to the minimum enemy size (regardless of what they are armed with), and additional troops as you feel justified. Ideally, you want to fight minimum-size enemy forces with the largest forces you can muster. This limits your casualties and sharply reduces the number of free zombie troops the AI gets to replace losses, which eases some of the strategic problems with the AI having unlimited reinforcements.

In my Union Legendary Minimal playthrough, I found that the pre-battle estimates in minor battles were pretty good. Pre-battle estimates in major battles were sometimes wildly off, primarily because they seemed to ignore reinforcements.

Edited April 7, 2018 by Aetius

[Grimthaur]

Yeah in the linked topic we discussed scaling on major battles.

Im not sure, how reliable pandakrauts data is regarding upsacaling at these battles. At antietam he had at about 40k vs at about 41k in all tests and he gives no information about the AI strength displayed in the campaign-screen. Yes he may have varried in brigade-size and -composition but it would have been more interesting to actually change your armysize by quite a margin (for example: with 55k i had to fight 67k, why not test 40, 45, 50, 55, 60?) to try to find out tresholds or diminishing-return mechanics.

Im still sure, scaling is based on numbers. There are just some mechanics that interfere. There is a minimal armysize regardless of how many AI-soldiers u have killed in earlier battles. This treshold will also prevent ballast-abuse to work on major battles if it is reached (and it should almost allways be reached). There has to be a max armysize since scaling is based on increasing or decreasing the AI-brigades size and not their numbers. The game seems to attribute a certain "normal" average brigade-size to you and if you go below, for example with ballast brigades pulling down your average, the AI scales down. If you go up, the AI scales up.

The up-scaling can be justified by providing a decent challange if legendary is too easy. But it is just disheartening and makes your decisive victories meaningless and sometimes they even become phyrric victories. Atm (union legendary) my recruit-pool is 65k before 2nd Winchester with 3 fully equipped corps and i see absolutely no use in adding more brigades in my 4th corps. The problem is not that the AI brigades scale but that your weaponry doesnt. Ofcourse i could equipp my 4th corps with M42s and Tylers but do i really want to? And if i add more brigades with good rifles, i cant refill them even if i am not limited by gold or recruits because there simply arent any. So despite a big recruit-pool i am bound to field an average-sized army.

The down-scaling is another thing. It does defnitly not work as intended, because it was - i think - aimed at players having a hard time and suffering way more casualties than normal players do. It is - at least imho - not meant for disguising as a general scratching togehter his last reserves while bunkering a big recruitpool and 50k rifles.  Ofcourse one can "abuse" this mechanics with ballast-brigades but it is even more counterintiutive than the up-scaling and it just lowers your challlange. Imho one shouldnt need to cripple the AI outside of battles. If there would be another patch this could easily be delt with if scaling down is bound to your army and your recruit-pool.

[pandakraut]

@Grimthaur The AI strength on the campaign screen was reported in my post: Intelligence Report: 64-69k, 42-47% training, 38-43% armory. These numbers get set after the last completed battle and won't change no matter how I modify my current army. I suspect when scaling was implemented that whatever calculation generates the intelligence report numbers wasn't updated to apply the scaling factors.

My goal with these tests was to see how many men I could put into my army without triggering scaling using different brigade sizes. For every test case if I added more men I would have triggered scaling, so testing larger army sizes wasn't of interest at the time. I just ran two new tests. One by adding several 2500 man brigades to the 1750 man test case and the other with maxing out units with the available weapons I had also using the 1750 test case.

Test 1: CSA 63834 men with 144 guns. Union 66469 92 guns. 

Test 2: CSA 75329 men with 172 guns. Union 78933 178 guns. 

I'm somewhat surprised that the CSA scaling didn't increase more though I suspect this would be less favorable on Legendary. The scaling ratios for Antietam seem to favor allowing the union to mostly match the AI size. Playing as CSA the Antietam ratios seem to force around a 2:1 disadvantage from what I've seen.

In general I like the scaling because it allows for multiple styles of play to be successful in terms of army creation. It certainly has it's realism issues, but it makes for a fun campaign. I'm currently doing a Legendary CSA play through where my normal unit size is between 1100 and 1250 and that has worked out fairly well through Chickamauga.

[Aetius]

15 hours ago, Grimthaur said:

The down-scaling is another thing. It does defnitly not work as intended, because it was - i think - aimed at players having a hard time and suffering way more casualties than normal players do. It is - at least imho - not meant for disguising as a general scratching togehter his last reserves while bunkering a big recruitpool and 50k rifles.  Ofcourse one can "abuse" this mechanics with ballast-brigades but it is even more counterintiutive than the up-scaling and it just lowers your challlange. Imho one shouldnt need to cripple the AI outside of battles. If there would be another patch this could easily be delt with if scaling down is bound to your army and your recruit-pool.

I don't think it is - I see it as a trap for struggling players. These players take high casualties and lose brigades. The game heavily penalizes small inexperienced brigades,  so creating new small brigades those won't trigger scaling but will cause you to lose battles and continue to take high casualties. On the other hand you can beef up existing brigades or create new large brigades, both of which make your average brigade size high even while your total number of troops and brigades is low. This triggers the scaling and tilts the game even further in favor of the AI.

I believe it was intended for players who want to fight with a small elite army instead of huge masses of troops. It doesn't work well on Legendary because the AI has much more experienced troops and the game is so lethal that small brigades tend to get wiped out regardless of experience.

More importantly, it's really unfair for the game to penalize you for units that don't participate in the battle - that's really the most counter-intuitive part of the whole thing. And it's almost unlimited how far the game will tilt in favor of the AI based on your army composition. I see ballast units as addressing that problem, not an abuse.

[LAVA]

1 hour ago, Aetius said:

More importantly, it's really unfair for the game to penalize you for units that don't participate in the battle - that's really the most counter-intuitive part of the whole thing. And it's almost unlimited how far the game will tilt in favor of the AI based on your army composition. I see ballast units as addressing that problem, not an abuse.

Would you not have achieved the same effect by, for example, creating a Cavalry Corps?

Since I am an avid viewer of your videos, I also noticed your preoccupation with minimizing the amount of artillery that was brought to the field by the Confederates. Why is that? Union artillery is far superior to Confederate artillery so their numbers really aren't that important. Your quality will trump their numbers any day.

My feeling (perhaps incorrectly so) is that because you were building an infantry centric army, that the AI merely responded in kind, as it also brings lots of cavalry and skirmishers to the major battles. I'm wondering (because I was guilty of the same sort of army creation) that if one brought a more balanced force perhaps the issue of scaling disappears.

[Grimthaur]

5 hours ago, Aetius said:

More importantly, it's really unfair for the game to penalize you for units that don't participate in the battle - that's really the most counter-intuitive part of the whole thing. And it's almost unlimited how far the game will tilt in favor of the AI based on your army composition. I see ballast units as addressing that problem, not an abuse.

Yes that is a problem, but either way (elite forces or ballast units). Either AI scales up and gets too strong or AI scales down and gets too easy to beat. I call ballast-units abuse because they disguise your real strength so that the game thinks you are a lot weaker than you really are and acts accordingly. It may address the upscaling in major battles, because decisive victories in small battles hurt the AI-armysize in major battles, but it is still using a game mechanics in a way it wasnt supposed to work imho. I watched your minimal playthrough (have watched quite a few of your vids, you deserve a lot more viewers imho) and at 2nd Winchester for example, you didnt want to play the displayed number of units so you added ballast units to weaken the AI. Thats as if you would use a slide controller for armysize, no offense.

Same thing can be done in a variety of games and it allways makes the AI weaker or matches you with weaker opponents than you should get in multiplayer games (for example, in WC3, if anyone remembers that awesome game, you could leave the first 10 games and would be matched with beginners for the next 70-80 games cause you disguised yourself as the ultranewb, in other games you get the rating of the groupleader and can abuse that way, in some RTS you could intentionally stuck the AI and shoot it down without resistance, all that i would call abuse).

@pandakraut My bad, i missed that.

Your new tests are very interesting to me, AI seems to add on a 1 vs 1 basis on MG. As i mentioned, on leg i had 55k vs 67k (122%) at antietam and i was under the impression that i depleted the AI before (67-72k, 91-96% training, 39-44% armory). I reloaded the save and went in with 3 corps and 1.000-men brigades. With 39,5k i had to face 46,7k (118%). Second test: 1500 men, with 58k i had to face 70k (121%). Third test: 2.500er brigades, 90k vs 105k (117%). So it is at about 1 vs. 1,2 at least in my tests. I guess it will become more favourable in numbers the more units you field, because you will reach the cap finally. But you will suffer more casualties naturally and you will still have the same amount of rifles in your weaponry. So one should defnitly not field the most units possible since it will only hurt the player and not the AI. Fielding the right amount would be an armysize to wich the AI would scale to its minimum armycap, ergo you should specialise and field a quite low force to be most cost-efficient (if you do not want to (ab)use ballast units).

Edited April 8, 2018 by Grimthaur

[pandakraut]

@LAVA There are at least two scaling factors involved. The base army size will scale and separately each unit type will also scale. For example, if your army only has infantry units your total army size will cause non-infantry units to scale up as well. But the infantry units will scale much more significantly. Large amounts of cavalry, skirmishers, or artillery will have a lesser effect on total army size scaling but you'll encounter noticeably larger sizes of the corresponding unit type. Two easy ways to see this is watching enemy guns decrease or increase as you add/remove infantry units and in some of the early battles using a cavalry heavy composition and seeing how large the cavalry units are in comparison to going infantry heavy. 

I have been playing with relatively small infantry units recently and using more cavalry. I have found that I need far fewer infantry ballast units but that I need far more artillery and cavalry ballast units. There also seems to be some kind of minimum threshold involving small unit sizes that can drive scaling back up. At Antietam CSA Leg I ended up fielding a third corps of small rookie units with muskets because fielding fewer men actually drove up scaling. I basically needed them anyways to hold the southern crossings and didn't have the rifles to add troops to my main units, but it was surprising to see removing units increase scaling in a major battle. There is probably some minimum size to unit number ratio that gets enforced but never comes up on larger unit sizes.

@Grimthaur The sweet spot for minimum scaling seems to be somewhere between 1k and 1.3k depending on other factors. I've found that while ballast units are still beneficial they are far less impactfull then when I was using 1.5 to 2k sized brigades. So if you wanted to avoid using them and don't want to deal with over scaling that's probably the range to go for. These numbers are purely anecdotal but < 400 for cavalry, < 300 for skirmishers and <10 guns for artillery seems to be around the right spot as well. The side advantages of this in terms of fewer weapons and veterans needed is also very nice. I was able to invest the money I saved on rifles into veterans so post Antietam I had two full corps of 2+ star units. The rifles you can get through reputation are arguably a little too strong when playing on smaller unit sizes. 

[Aetius]

On 4/8/2018 at 10:32 AM, LAVA said:

Would you not have achieved the same effect by, for example, creating a Cavalry Corps?

As pandakraut pointed out, no - each type of unit is scaled separately. On Legendary, creating a cavalry corps virtually guarantees that any enemy cavalry units will all be overstrength 1050-man units, even when your cavalry isn't in the battle. Cavalry units also make poor ballast units because disbanding and re-creating them costs money every time for horses.

 

On 4/8/2018 at 10:32 AM, LAVA said:

Since I am an avid viewer of your videos, I also noticed your preoccupation with minimizing the amount of artillery that was brought to the field by the Confederates. Why is that? Union artillery is far superior to Confederate artillery so their numbers really aren't that important. Your quality will trump their numbers any day.

It doesn't work that way on Legendary. As the game progresses, almost every Confederate artillery unit will be three stars and have one of the two large damage bonuses. This turns standard 10pd Ordnance / 10pd Parrots into dangerous guns, and even 12pd Howitzers are deadly if you get too close - and late game you are facing numerous 24pd Howitzers (I captured 32 at Richmond, which meant that over half their 600 guns were 24pd Howitzers). Further, artillery damage in a battle goes up more-or-less linearly with the number of tubes, as you have few options for keeping them from shooting (unlike infantry, which can be suppressed relatively easily). In short, every gun that's added to the enemy side roughly equals X amount of casualties regardless of how well the battle goes, so fewer guns has a direct correlation to fewer casualties on your side. And due to how scaling works, replacing infantry brigades on your side with artillery brigades in minor battles does nothing to reduce the enemy infantry forces while adding enemy guns, making the battle more difficult and increasing casualties. It's not so much a desire for an infantry-centric force as the game's various incentives put pressure on you to create an infantry-centric force.

[pandakraut]

While doing some unrelated testing I noticed that for Brandy Station and Supply Raid your army size is included in scaling even though only allied units deploy in the battle. Somewhat obvious in retrospect considering how side battles normally works, but I hadn't realized this before. Definitely don't forget to use ballast units prior to those battles as the allied units already make the scaling bad enough.

[Grimthaur]

23 hours ago, pandakraut said:

 

@Grimthaur The sweet spot for minimum scaling seems to be somewhere between 1k and 1.3k depending on other factors. I've found that while ballast units are still beneficial they are far less impactfull then when I was using 1.5 to 2k sized brigades. So if you wanted to avoid using them and don't want to deal with over scaling that's probably the range to go for. These numbers are purely anecdotal but < 400 for cavalry, < 300 for skirmishers and <10 guns for artillery seems to be around the right spot as well. The side advantages of this in terms of fewer weapons and veterans needed is also very nice. I was able to invest the money I saved on rifles into veterans so post Antietam I had two full corps of 2+ star units. The rifles you can get through reputation are arguably a little too strong when playing on smaller unit sizes. 

Problem with 1k to 1,3k is that you cant really use detached skirmishers and they are too powerfull to do without imho. I think an average 1,7 to 1,8k would be best then. Btw the efficiency-cap for skirmishers is 250, someone tested that some time ago. And yes, you will have a small but elite force equipped with the best weapons available. But what to do with all those weapons captured and the cumulating recruit-pool (atm at Gburg with 80k recruits available), ignore them? Doesnt feel right :(.

[pandakraut]

I haven't had any issues using detached skirmishers, though maybe I use them differently. I definitely don't expect them to stand up to fire for any period of time outside of cover with the unit sizes I'm using. I'd agree that somewhere between 1.6 and 1.8 is probably the range you want to shoot for if you want larger skirmishers. 

Good to know on the skirmishers efficiency. 250 is what I normally settle on in terms of available sniper weapons anyways. The extra weapons and recruits have never bothered me, but I also tend to horde resources just in case and then never use them. If you're using ballast units you always want at least 5-8k available to scale those up and down. I've never tried the max army size approach, but with anything less than that I think you will always eventually float recruits unless you are taking horrible losses.

[LAVA]

It appears that the only reason to pay attention to scaling is to reduce the number of men to the absolute minimum you have to fight and win battles. Thus, it makes it easier to win battles. Winning battles means you get all the rewards, plus you suffer fewer casualties and increase the efficiency of your units.

But while winning battles is the name of the game... unfortunately, it isn't the whole game. By using your scaling techniques you are more likely to win... but you are less likely to actually do grievous harm to your opponent. For example, I am now up to Nansemond River. In your case Aetius, you were facing a foe who has training in the region of 70%. Because of this every single unit you fight in the battle is a 3 star brigade. In my campaign, the Confederates training and armory for Nansemond River is below 50%. Though I haven't fought the battle yet, I'm pretty sure there won't be a lot of 3 star brigades that I will have to contend with. In the scenario before that, at Supply Raid, I was facing Stuart's 3 star cavalry, but the infantry was only 1 star and the skirmishers actually had no stars.

In reality, the most important factor in the game is totally eliminating enemy brigades because by totally eliminating them, they lose all the experience they have gained from past battles. Why should you want the enemy to bring as few troops possible to a battle that you know you can win easily? The fewer troops they bring the less impact your victory has on the overall enemy army, especially in the minor battles. After slaughtering the Rebs at Gaines Mill and Mulvern Tavern, I saw an almost immediate effect on the quality they were bringing to the minor battles.

The minor battles are a "training ground" for not just your army but for your enemy as well. If they survive, they gain experience. If you use scaling to reduce their numbers to the lowest possible you miss a huge opportunity to deal body blows to their main army. Damn! Let them bring as many as possible! At Kettle Run I wiped them all out. I wiped them all out at Thouroughfare Gap, Crampton's Gap, Iuka, and Parker's Crossroads. I completely destroyed around half of the enemy units at Perryville and Supply Raid (including 4 out of 5 of Stuart's 3 star cavalry brigades). And I totally expect to wipe them all out at Nansemond River. After that I expect to see lots of 1 and 2 star units at that bugger of battle, the Siege of Suffolk. And if that is what I encounter, I expect to wipe them out there as well.

By reducing the amount of troops you encounter at minor battles via scaling techniques, you are actually helping the Confederates maintain a qualitative advantage. Why would you want to do that?