Upgunning the soviets - the IS-2

Yes but spreadsheets can't observe why while having good stat the IS2 is barely used. So maybe its because other unit's stats, maybe because of the meta, or build order, or cost etc...
I take again the Sherman example, it has on paper good AT stat, but once you compare it with armor it faces they suddenly don't look like so powerful.

Let's think about the parallel between Tiger and IS2. On paper they have the same AI DPS according to MMX, but both tanks don't have the same speed, could it be possible that the Tiger being able to close the distance faster and disengage faster is used more aggressively and thus regularly at medium range while the IS2 being slower is more likely to shot at max range making the difference people are seeing. Or that a good part of IS2 AI dps comes from the dushka which is modified by the constant late game yellow cover all around the map.

My point was only there, Stats are indeed a good 1st point to take in account but you can't stop by it and say everything is fine or broken because stats are ok or not.

This is not contradicting anything that calculatuons can tell you.

If other units can fill the gap/role of the IS2, then the calculations might actually tell you why. The calculations will also give you a pretty decent picture of the Sherman's performance against other vehicles. If there is e.. a Brummbar meta, stats will tell you what units perform worse or better at countering it.
And if MMX's calculations about the Tigers's and IS-2's AI say they are pretty close, but you in game think they are, it'll at least tell you that it is because of other factors like moving too much etc. He doesn't overstate the explanatory power of his sim, he clearly explained them for what they are: artificial, standardized setups with very defined parameters that come close to the in game situation. No one claimed they will tell you every aspect of the AI of these units. And I fully agree, theydon't give you the full picture of a unit. But single posts of a hand full of usera do not either. And doing the calculations also prohibit anyone from critical reasoning and adding other information. But they give comparable numbers that - if done well - reasonably represent what you should see in game if you try to replicate the setup.

This is not contradicting anything that calculatuons can tell you.

If other units can fill the gap/role of the IS2, then the calculations might actually tell you why. The calculations will also give you a pretty decent picture of the Sherman's performance against other vehicles. If there is e.. a Brummbar meta, stats will tell you what units perform worse or better at countering it.
And if MMX's calculations about the Tigers's and IS-2's AI say they are pretty close, but you in game think they are, it'll at least tell you that it is because of other factors like moving too much etc. He doesn't overstate the explanatory power of his sim, he clearly explained them for what they are: artificial, standardized setups with very defined parameters that come close to the in game situation. No one claimed they will tell you every aspect of the AI of these units. And I fully agree, theydon't give you the full picture of a unit. But single posts of a hand full of usera do not either. And doing the calculations also prohibit anyone from critical reasoning and adding other information. But they give comparable numbers that - if done well - reasonably represent what you should see in game if you try to replicate the setup.

I never said contradiction, I said its not enough to set conclusion.

Why is SU76 the almost never built for Soviets?

never built and people think it's fine because they see it being made


what a pitiful vehicle, deserves more of a thread than the is-2 and i wanted to include it as well, but i already made one and it would have gone nowhere, but i digress

Yes but spreadsheets can't observe why while having good stat the IS2 is barely used. So maybe its because other unit's stats, maybe because of the meta, or build order, or cost etc...
I take again the Sherman example, it has on paper good AT stat, but once you compare it with armor it faces they suddenly don't look like so powerful.

This is true, of course, and should be obvious enough for everyone to not warrant a disclaimer. What spreadsheets, or any amalgamation of stats for that matter, allow you to do is to get insight into things you can't (or only with great difficulty) deduce from gameplay alone. Sure you can get a rough idea how likely it is to pen a PzIV with a Sherman, but without looking up the exact value from some external source you'll likely never be able to narrow it down precisely due to the huge influence of RNG. But values like chance to pen, chance to hit, scatter, etc. are essential for any sort of reasonable comparison to be drawn, after all that's exactly why we have and need all kinds of stat pages, spreadsheets and calculators. That this can't give you all the answers needed to explain why, for example, the IS-2 sees less use than the Tiger is undoubtedly true, but no one (at least not me) claimed anything like that in the first place.

Let's think about the parallel between Tiger and IS2. On paper they have the same AI DPS according to MMX, but both tanks don't have the same speed, could it be possible that the Tiger being able to close the distance faster and disengage faster is used more aggressively and thus regularly at medium range while the IS2 being slower is more likely to shot at max range making the difference people are seeing. Or that a good part of IS2 AI dps comes from the dushka which is modified by the constant late game yellow cover all around the map.

My point was only there, Stats are indeed a good 1st point to take in account but you can't stop by it and say everything is fine or broken because stats are ok or not.

Saying that both tanks have roughly the same AI DPS/KPS was of course a gross oversimplification; in reality there are indeed various ways both tanks differ in terms of their AI performance based on squad formation and model count, distance, cover, alpha-strike capabilities and so on and so forth.
However, it should also go without saying that a complex analysis like this would be quite far out of scope for a simple reply as to why I think both the Tiger and IS-2 are roughly equal in terms of AI. (Again, if you are really interested in the details, such as what difference MGs make in yellow vs no cover fights or which tank is more effective in killing models with the first shot you can look it up over here).

I fully agree to the last part, though. Nonetheless I still prefer to make assumptions and judgements based on stats that may paint an inherently incomplete picture than to dismiss these kind of objective data alltogether and rely fully on game sense or intuition instead.

Last time I checked IS-2 had a frag round, which makes any weapon crew half dead and it also sports lots of armor, while facing units that are having lower penetration by desing. But wait a minute, Tiger is 5% faster! Is2 is also could be covered by t34-85, but hey, tiger has PIV.
I'm yet to hear what superior supportive structure Axis has that makes Tiger better then IS2...which is a SOV unit....a unit within one of the most complete factions in the game
Show Spoiler

I'd really like to have had the IS2, Tiger 1 and Pershing all have their damage buffed(200) and rof nerfed to better differentiate them as heavy break through tanks. With the same kind of damage they all over lap with the rest of their factions vehicles. It be nice if they had a bit more burst damage especially against heavier vehicles. Anti infantry damage could be readjusted so they don't completely overshadow tanks like the KV2, or Brummbar.

I'd really like to have had the IS2, Tiger 1 and Pershing all have their damage buffed(200) and rof nerfed to better differentiate them as heavy break through tanks. With the same kind of damage they all over lap with the rest of their factions vehicles. It be nice if they had a bit more burst damage especially against heavier vehicles. Anti infantry damage could be readjusted so they don't completely overshadow tanks like the KV2, or Brummbar.

A slight nitpick: burst damage does depend on high rof.

I think you're talking about alpha damage, which is a large front-loaded hit.

Anyway, yeah I like your idea there. Especially thinking of IS2 with better AT.

I never said contradiction, I said its not enough to set conclusion.

I also didn't say you did, but I think you are slightly underestimating what theoretical calcs can do if read properly. But to be honest, I think we pretty much agree, there is no real point in arguing in the first place.

What spreadsheets, or any amalgamation of stats for that matter, allow you to do is to get insight into things you can't (or only with great difficulty) deduce from gameplay alone. Sure you can get a rough idea how likely it is to pen a PzIV with a Sherman, but without looking up the exact value from some external source you'll likely never be able to narrow it down precisely due to the huge influence of RNG.

Just to drive this point home for everyone:
Assume you try to deduce how good a Sherman penetrates frontally at medium range vs an Ostheer P4. You have 20 players discussing over their experience with the Sherman. We don't know what they understand by "mid range", and even if everyone understands roughly 20 meters, there will be variations from probably at least 15-25 meters being classified at mid range, variations because of players driving their tanks, factoring in missed shots, reduced accuracy due to moving, increased penetration due to engaging at an angle and getting a rear armor hit.

Let's put this into a test setup: 20 meters distance (120 penetration vs 180 armor), flat terrain, only frontal shots possible, only actual hits counting. We count 50 hits, assuming that the players in the discussion above would have their last roughly 50 shots in mind when thinking about the engagements. We do 20 tests to form each player's experience.
Basic statistics tells us that the real penetration chance is 66.7%, meaning 50*(120/180) = 33.33 shots should penetrate with a standard deviation of 3.33 shots (10% of the mean in our case). This means that almost one third of the players discussing will have had the Sherman penetrate less than 30 times or more than ~37 times. One player will have had the experience of the Sherman hitting less than 27 times or more than 40 times and probably be screeching about the trash Sherman or saying it were OP.
Even if all discussion was fully rational (good luck with that), players will regularly report a penetration chance anywhere between less than 60-75%.

Funny side note: The skill level doesn't matter. This one player could be Luvnest or VonIvan, not just your level 2000 rando. Those 50 shots probably represent the last 5 games, which represents 3 hours of games or 1-2 evenings of playing CoH2 with this faction and this unit. This in turn means that it might all the experience you get for 1-2 weeks playing CoH2 if you also play all other factions equally. This one player, high skill or not, will say that the Sherman were OP or trash. And if they are known as a good player, their opinion will have a larger impact although it actually should not. And we can prove that with numbers.

Back to topic: On top of that, all this assumes that this sample of players actually gets a "representative set of RNG". If I just hit refresh on the numbers, I regularly see that even all their "experience data" pooled together will both misjudge the real penetration value by a couple of %, as well as the standard deviation varying decently (often 7-13%, meaning players will either be more in agreement if it gets smaller or disagreement if it is larger). Is this huge? Not huge, but we've seen penetration and armor values being decreased by 10-20 regularly, which depending on shooter and target often lead to only a couple of percent penetration chance. So yes, we're still in the range of actual balance discussions being influenced by RNG.

If you want a really reliable discussion where the values only vary by a one or two percent, you'd need 100 players. That would be 5 pages on this forum if everyone just states their experience, no discussion involved yet. And don't forget: Everyone just tested the test setup, there still is no variation due to a real game.

[...]
Just to drive this point home for everyone:
Assume you try to deduce how good a Sherman penetrates frontally at medium range vs an Ostheer P4. You have 20 players discussing over their experience with the Sherman. We don't know what they understand by "mid range", and even if everyone understands roughly 20 meters, there will be variations from probably at least 15-25 meters being classified at mid range, variations because of players driving their tanks, factoring in missed shots, reduced accuracy due to moving, increased penetration due to engaging at an angle and getting a rear armor hit.

Let's put this into a test setup: 20 meters distance (120 penetration vs 180 armor), flat terrain, only frontal shots possible, only actual hits counting. We count 50 hits, assuming that the players in the discussion above would have their last roughly 50 shots in mind when thinking about the engagements. We do 20 tests to form each player's experience.
Basic statistics tells us that the real penetration chance is 66.7%, meaning 50*(120/180) = 33.33 shots should penetrate with a standard deviation of 3.33 shots (10% of the mean in our case). This means that almost one third of the players discussing will have had the Sherman penetrate less than 30 times or more than ~37 times. One player will have had the experience of the Sherman hitting less than 27 times or more than 40 times and probably be screeching about the trash Sherman or saying it were OP.
Even if all discussion was fully rational (good luck with that), players will regularly report a penetration chance anywhere between less than 60-75%.

Funny side note: The skill level doesn't matter. This one player could be Luvnest or VonIvan, not just your level 2000 rando. Those 50 shots probably represent the last 5 games, which represents 3 hours of games or 1-2 evenings of playing CoH2 with this faction and this unit. This in turn means that it might all the experience you get for 1-2 weeks playing CoH2 if you also play all other factions equally. This one player, high skill or not, will say that the Sherman were OP or trash. And if they are known as a good player, their opinion will have a larger impact although it actually should not. And we can prove that with numbers.

Back to topic: On top of that, all this assumes that this sample of players actually gets a "representative set of RNG". If I just hit refresh on the numbers, I regularly see that even all their "experience data" pooled together will both misjudge the real penetration value by a couple of %, as well as the standard deviation varying decently (often 7-13%, meaning players will either be more in agreement if it gets smaller or disagreement if it is larger). Is this huge? Not huge, but we've seen penetration and armor values being decreased by 10-20 regularly, which depending on shooter and target often lead to only a couple of percent penetration chance. So yes, we're still in the range of actual balance discussions being influenced by RNG.

If you want a really reliable discussion where the values only vary by a one or two percent, you'd need 100 players. That would be 5 pages on this forum if everyone just states their experience, no discussion involved yet. And don't forget: Everyone just tested the test setup, there still is no variation due to a real game.

Couldn't have explained this any better! Can we have that as a sticky disclaimer for the balance forum?

I also didn't say you did, but I think you are slightly underestimating what theoretical calcs can do if read properly. But to be honest, I think we pretty much agree, there is no real point in arguing in the first place.


Just to drive this point home for everyone:
Assume you try to deduce how good a Sherman penetrates frontally at medium range vs an Ostheer P4. You have 20 players discussing over their experience with the Sherman. We don't know what they understand by "mid range", and even if everyone understands roughly 20 meters, there will be variations from probably at least 15-25 meters being classified at mid range, variations because of players driving their tanks, factoring in missed shots, reduced accuracy due to moving, increased penetration due to engaging at an angle and getting a rear armor hit.

Let's put this into a test setup: 20 meters distance (120 penetration vs 180 armor), flat terrain, only frontal shots possible, only actual hits counting. We count 50 hits, assuming that the players in the discussion above would have their last roughly 50 shots in mind when thinking about the engagements. We do 20 tests to form each player's experience.
Basic statistics tells us that the real penetration chance is 66.7%, meaning 50*(120/180) = 33.33 shots should penetrate with a standard deviation of 3.33 shots (10% of the mean in our case). This means that almost one third of the players discussing will have had the Sherman penetrate less than 30 times or more than ~37 times. One player will have had the experience of the Sherman hitting less than 27 times or more than 40 times and probably be screeching about the trash Sherman or saying it were OP.
Even if all discussion was fully rational (good luck with that), players will regularly report a penetration chance anywhere between less than 60-75%.

Funny side note: The skill level doesn't matter. This one player could be Luvnest or VonIvan, not just your level 2000 rando. Those 50 shots probably represent the last 5 games, which represents 3 hours of games or 1-2 evenings of playing CoH2 with this faction and this unit. This in turn means that it might all the experience you get for 1-2 weeks playing CoH2 if you also play all other factions equally. This one player, high skill or not, will say that the Sherman were OP or trash. And if they are known as a good player, their opinion will have a larger impact although it actually should not. And we can prove that with numbers.

Back to topic: On top of that, all this assumes that this sample of players actually gets a "representative set of RNG". If I just hit refresh on the numbers, I regularly see that even all their "experience data" pooled together will both misjudge the real penetration value by a couple of %, as well as the standard deviation varying decently (often 7-13%, meaning players will either be more in agreement if it gets smaller or disagreement if it is larger). Is this huge? Not huge, but we've seen penetration and armor values being decreased by 10-20 regularly, which depending on shooter and target often lead to only a couple of percent penetration chance. So yes, we're still in the range of actual balance discussions being influenced by RNG.

If you want a really reliable discussion where the values only vary by a one or two percent, you'd need 100 players. That would be 5 pages on this forum if everyone just states their experience, no discussion involved yet. And don't forget: Everyone just tested the test setup, there still is no variation due to a real game.

True. At the end of the day, this game is a numbers and chance game. While pure numbers can't stand on their own due to the different scenarios in the game (eg. put a jagdtiger on a small, tight 1v1 map and all of it's "greatness" disappears), they can still paint a nice picture. And in the end, numbers don't lie. Heck, I could have said that the raketen was severely UP 3 or 4 games ago because it missed on my AAHT 4 times in a row, and last game it hit and penetrated each and every shot on a Pershing and AAHT and allied Comet, therefore OP. Don't know why people try to counter statistics and numbers with personal feelings. Sure you can put some context around it, like game mode and map, but concerning something as basic as penetration there really isn't a lot to discuss.

True. At the end of the day, this game is a numbers and chance game. While pure numbers can't stand on their own due to the different scenarios in the game (eg. put a jagdtiger on a small, tight 1v1 map and all of it's "greatness" disappears), they can still paint a nice picture. And in the end, numbers don't lie. Heck, I could have said that the raketen was severely UP 3 or 4 games ago because it missed on my AAHT 4 times in a row, and last game it hit and penetrated each and every shot on a Pershing and AAHT and allied Comet, therefore OP. Don't know why people try to counter statistics and numbers with personal feelings. Sure you can put some context around it, like game mode and map, but concerning something as basic as penetration there really isn't a lot to discuss.

this might be a wild guess, but since raketen is too low, the scatter shots tend to hit obstacles and terrain more often compared to other AT guns. This may rob it from lucky scatter hits on the armor. Although I'd imagine that the difference should not be that big.

if SU76 is effective at wiping german infantry. i ll gladly build them.

back to topic.

i remember back when IS-2 had 240dmg but it takes 8-9 secs reload and seriously bad scatter just to balance it. If RNG is on your side, it can wipe squads left and right and each shell have huge AOE. Even if the shell miss, it would damage the infantry squad badly. And yes, it has about the same front armor as KT.

its a breakthrough tank afterall.

Let's put this into a test setup: 20 meters distance (120 penetration vs 180 armor), flat terrain, only frontal shots possible, only actual hits counting. We count 50 hits, assuming that the players in the discussion above would have their last roughly 50 shots in mind when thinking about the engagements. We do 20 tests to form each player's experience.
Basic statistics tells us that the real penetration chance is 66.7%, meaning 50*(120/180) = 33.33 shots should penetrate with a standard deviation of 3.33 shots (10% of the mean in our case). This means that almost one third of the players discussing will have had the Sherman penetrate less than 30 times or more than ~37 times. One player will have had the experience of the Sherman hitting less than 27 times or more than 40 times and probably be screeching about the trash Sherman or saying it were OP.
Even if all discussion was fully rational (good luck with that), players will regularly report a penetration chance anywhere between less than 60-75%.

This is human nature - remembering anything that seems out of the ordinary. I once had 5 shock troops killed by a single Panzershrek shot because they walked around a corner and were stacked on each other when it hit. Clearly the Panzershrek should've been nerfed! I just couldn't motivate myself to start a thread over it so justice wasn't served.....

Normal people (meaning everyone that didn't have to take a statistics to get their degree) often underestimate how common "uncommon" results are. Take the normal heads or tails. Theoretically it's 50/50. Would you ever expect to roll 9 tails in a row? You can simulate this in Excel by putting "=rand()" in cell A1 and "=round(A1,0)" in cell B1, copy that down 99 cells. Rand will recalculate every time you hit F9. If my Sherman had a 50/50 chance of penetrating a Tiger and I fired 100 shots, I would've got 9 bounces in a row on my third recalculation. Streaks of 9 are somewhat uncommon, but 5-7 are really common.

Excel will also simulate your test. Replace the Round function with "=IF(B1>0.667,1,0)" and add "=COUNTIF(B1:B100,1)" to any cell near the top. Hit F9 a bunch of times and see how bad or good your Sherman is. In short, Excel is OP.

The two counter arguments I'd make against just looking at the numbers in Excel are about the effect of alpha damage and time to kill being irrelevant. In a RTS, alpha damage has an outsized effect because it greatly increases the risk of losing a squad. That was a lot of the problem with the old IS2. It seemed to go miss, miss, wipe on squads a lot, but sometimes it started with the wipe. The other problem is that people like to argue about the time to kill. I've seen multiple comparisons where people are comparing TTK times in the 20-40 seconds and trying to use that to make a point. In an actual game, 20 seconds may as well be an eternity, as nobody has reactions that are that slow. The bazooka was once nerfed because Relic thought the TTK was too short.

Back to original topic - Based on how often the commanders are picked, it doesn't seem like the majority of the top 200 think the IS2 is trash. It's just that the commanders have fallen into the "B" tier. Trash is more like Conscript Support, Tank Hunters, any Partisan commander, etc.

[...]
Normal people (meaning everyone that didn't have to take a statistics to get their degree) often underestimate how common "uncommon" results are. Take the normal heads or tails. Theoretically it's 50/50. Would you ever expect to roll 9 tails in a row? You can simulate this in Excel by putting "=rand()" in cell A1 and "=round(A1,0)" in cell B1, copy that down 99 cells. Rand will recalculate every time you hit F9. If my Sherman had a 50/50 chance of penetrating a Tiger and I fired 100 shots, I would've got 9 bounces in a row on my third recalculation. Streaks of 9 are somewhat uncommon, but 5-7 are really common.

This is a good point IMHO. While getting 7 bounces in a row isn't exactly likely with a bit under a 1% chance, it's not astronomically improbable, either. Yet the perception about how rare such occurrences are often depends if people find themselves on the receiving end of RNG or not.

[...]
The two counter arguments I'd make against just looking at the numbers in Excel are about the effect of alpha damage and time to kill being irrelevant. In a RTS, alpha damage has an outsized effect because it greatly increases the risk of losing a squad. That was a lot of the problem with the old IS2. It seemed to go miss, miss, wipe on squads a lot, but sometimes it started with the wipe. The other problem is that people like to argue about the time to kill. I've seen multiple comparisons where people are comparing TTK times in the 20-40 seconds and trying to use that to make a point. In an actual game, 20 seconds may as well be an eternity, as nobody has reactions that are that slow. The bazooka was once nerfed because Relic thought the TTK was too short.
[...]

TTK is certainly not the only and most important performance figure, but I'd argue it is also far from irrelevant and can be quite suitable to prove a point (at least as long as it is put into the right perspective). It should, however, more be seen as a surrogate for DPS/DPM that also takes things like over damage into account than as a simple measure of 'how long does it take A to kill B'. After all, the fact that firefights usually don't last more than 20 s doesn't mean it can't serve as a good benchmark for comparing the relative performance of different units against a similar target.

In the end, though, you're of course right in that TTK shouldn't be taken as the only measure and things like alpha damage or the variety of mobility and other non-combat relevant stats also play an important role.

This is a good point IMHO. While getting 7 bounces in a row isn't exactly likely with a bit under a 1% chance, it's not astronomically improbable, either. Yet the perception about how rare such occurrences are often depends if people find themselves on the receiving end of RNG or not.



TTK is certainly not the only and most important performance figure, but I'd argue it is also far from irrelevant and can be quite suitable to prove a point (at least as long as it is put into the right perspective). It should, however, more be seen as a surrogate for DPS/DPM that also takes things like over damage into account than as a simple measure of 'how long does it take A to kill B'. After all, the fact that firefights usually don't last more than 20 s doesn't mean it can't serve as a good benchmark for comparing the relative performance of different units against a similar target.

In the end, though, you're of course right in that TTK shouldn't be taken as the only measure and things like alpha damage or the variety of mobility and other non-combat relevant stats also play an important role.

Getting 7 bounces in a row doesn't seem likely, but if you have 100 Sherman shots in a game, the odds of it happening to you are about 50%. Since most people here don't use Excel, I decided to see if I could make the same spreadsheet in Sheets. Here it is:

https://docs.google.com/spreadsheets/d/1Doyd3MrUB83xvzYpcqS3TkeXsWtziEI-uCbJPbqusr0/edit?usp=sharing

It can be recalculated by hitting the refresh button if you're looking in a browser. Scroll down and see the streaks of bounces or penetrations. You can change the odds by editing the "if" statement.

You're right that I overstated the "irrelevant" part. It has relevance when talking about infantry as target since reinforcing costs manpower. There is less relevance when talking about tanks since repairs are free. It can be a substitute for DPS as you pointed out.

IS2 buff, my god. Some people have no concept of reality whatsoever.

I truly believe is 2 must have special rounds that automatically used every two tree shots and when it fired and hit it would delete enemy tank turret it special animation during wich turret goes up split in two parts or more with loud boom sound killing all only ENEMY infantry within 2 screens including buildings

I truly believe is 2 must have special rounds that automatically used every two tree shots and when it fired and hit it would delete enemy tank turret it special animation during wich turret goes up split in two parts or more with loud boom sound killing all only ENEMY infantry within 2 screens including buildings

So kinda like a Sturmtiger?

PS - see my signature

So kinda like a Sturmtiger?

PS - see my signature

No sturmtiger kills in small circle

I wrote 2 screen in fully automatic action no input required

Tried out the IS-2 today. Its actually pretty decent considering it can take damage for whatever units you put behind it (think Churchill). The HE ability is quite strong too. It feels stronger than ISU-152's normal HE shell at the cost of vet and some muni. Near pinpoint accuracy which is very nice when fighting team weapons not in green cover like AT guns or MGs.

By the time you get an IS-2 out, you should have an AT gun or two. Fighting Panthers and other big cats is really a non-issue because you will almost never go up against them alone. Soviets are a combined arms army after all.