Matrix Games Forums

Hello, I want to report a visual bug. In my server multiplayer game I had this situation on the start of my turn playing as Soviets and spotted an opportunity to attack the 9=9 inf division in a clear hex in the east. I gathered the forces, 30 v 9 cv with one hex attacking across the river so 25 v 9 should be ok since lead by Rokossovsky with good support and the result is this battle I will not moan about the performance of the German elements because this is not the main topic, but how does this division all of the sudden have 17.3 CV? I assumed it was a division with high CPP which actually has around 4.5 CV so the lvl 1 fort bumps it up to 9, but it looks like that is not true at all... From what i can see the level one fort wasn't included in the defensive CV values shown to me (which I'm pretty sure it should).

After the attack, even though we busted the fort, the unit is still a 9=9 Also if you look at the western part of the first image you will see that for some enemy divisions i see x=x in a lvl 1 fort, and for some i see x=2x. So the fort effect is sometimes shown, sometimes not from what i can tell. Is this a bug? Is it a known one? Also, how can i evaluate what unit am i dealing with? Should I manually double the CV if i see a fort or not? If this is not something that happens to everyone, only thing specific about my setup is that the game started on an older patch, I'm not sure if that can cause some issues.

I would be very grateful if some of the moderators can respond about this issue

I just found some cavalry units volunteering to scout the German defenses and as you can see that when i attack both the 10=10 and 8=16 divisions they actually in fact both have 8-10 = 16-20 CV

Not a Moderator but a Mod(d)er, so ~60% applicable.

Welcome to your first experience of the unpredictability and variance of MAP CV / Combat Predictor Tooltip CV.
This is one of Gary's many layers in the game that makes it impossible to know with certainty how combats will end (in this case even how they will start).

Your initial combat has a variance of (assuming the middle of the 90s) 1-(95/172)=0.448 so -44.8% variance (using the actual CV as the base).
This is a somewhat normal variance to observe and may even get to greater levels.
In cases where you do not contact a Unit during your logistics but will contact it after movement, this variance will then on average be even larger.
You can observe the same off'ness if you make followup attacks against a retreated unit, but in those instances it appears the CV prediction will always be way too high (from my experience - some call those cases the Pinoccio Predictions). Similar as in your case, while "scouting" the actual CV the Tooltips do not "update" to your new intel with any accuracy afaik, that's normal.

One of the CSV exports one can make in the Editor shows a Recon/Scout value for units, not sure if this is still being used or not. It appears to me that MOT/Tank formations achieve better predictions, but do not quote me on that.
It is possible (I do not know) that certain terrain/fort/weather situations can impact the variance in a "static/predictable" way as you suggest it for the Fort Levels in your example(s). I do not think that this is the case but worth keeping a look out for it.
Other possible factors could also be the amount of friendly units in adjacent Hexes, Terrain, Detection Levels, Supply situations, Leader Rolls, Weather, Ground Weather and the like (I have no idea).

I invite you to contribute to it in this Thread and with many more examples we may figure if there is some pattern to see.

PPetar wrote: Mon Nov 24, 2025 3:15 pm I assumed it was a division with high CPP which actually has around 4.5 CV so the lvl 1 fort bumps it up to 9, but it looks like that is not true at all... From what i can see the level one fort wasn't included in the defensive CV values shown to me (which I'm pretty sure it should).

I am sure you know this but from what you wrote this seems not 100% clear to me, so I just add it here, just in case.
The CPP are only affecting the attacking CV (left number) while the Fort Level is only affecting the Defensive CV (right number).

PPetar wrote: Mon Nov 24, 2025 3:15 pm If this is not something that happens to everyone

Everyone.
As said, one of many layers of randomness we have to go through in each individual combat in hope to be able to anticipate/predict a combat result.

PPetar wrote: Mon Nov 24, 2025 3:15 pm I will not moan about the performance of the German elements because this is not the main topic

You have attacked with 6(+2RESERVE?) Divisions. There are 2 things to consider.
1. The more Divisons/Corps (on MAP units) you attack with, the less chance the individual element will have to shoot (of your own units). This modifier starts shortly after/above 3 Divisions.
2. The more elements you bring (not sure if Elements, manpower ...whichever it is), the easier it will be for the enemy to hit you, so this improves German performance.
This is the point where the topic of CPP then comes into play.

A fort level 1 doubles the defense value only. A 100 CPP doubles the attack value only. So a unit with 100 CPP in a fort 1 and clear terrain will show equal attack and defense CVs. That's probably what you are seeing. Not as sure what to say about the combat CV versus display CV as I've forgotten too much about how these are handled. Wiedrock usually provides good insight into what's going on.

Thanks for the reply.

My initial assumption was that the 9=9 was a 100 CPP unit but in both combats i posted initial cv jumped up by x2 which would suggest a fort not being calculated. It would also be really strange for my opponents units from the second picture to have one 0 CPP one 100 CPP since they have both been just sitting there idle for a few turns digging, I would expect them to be on the same level.

Here are some other cases that are probably just wrong displays of defensive cv value due to not counting (partially or fully) fort levels

Here the 8=16 should probably be 8=25 like the division a bit to the east but fort is counted as +1 instead of +2 (and in this case even 100CPP wouldn't really make sense since it should then be 8=12) The panzer division in the last screenshot I know for sure doesnt have 100 CPP or anywhere near that since it was active at the front for a while.

Wiedrock here is talking about the unpredictability of actual cv, so what you see on the counter is an estimate of the actual CV. This affected me too, as you can see a 9 in the first battle was actually 8.6 if you count the fort, in the other two battles the 10 was actually a 9.3 and the 8 was actually a 8.7. But this is again not what I'm talking about, here the actual CV is in line with what the offensive CV shows, its just that the defensive CV sometimes gets doubled due to fort and sometimes not.

I did a bit of testing in my ai vs ai test game and I found similar strange results. Here are images of the predicted cv values with and without fog of war
As you can see there is some variance with the CV due to uncertainty (17.3 -> 20.3 and 7.5->6.8) which is fine, but when the fow is on the defensive CV is just capped to offensive CV, something that also happened in my examples (so it shows exactly x=x). I would expect to see something like 17.3=23.1 and 7.5=9.9, but instead sometimes the defensive CV just skips some calculations and shows random values. These are maybe not the best examples since the units here i think have close to full CPP and i doubt that this is the case with my opponents units in the actual game.

I think i found a case that makes the issue I'm talking about pretty clear.

I'm controlling both sides in a local game. This is the status of the division when i end the turn as Axis. As Soviets then I see this And when i attack it is actually So the fort level is not counted in the defensive CV shown on the counter.

I thought about it being a bug where a newly built fort is not counted, but I checked my aar and the 10=10 we see in the first picture of my original post had a fort level 1 as well one turn before So I think it is clear it is a visual bug as I have stated originally.

PPetar wrote: Mon Nov 24, 2025 7:02 pm I think i found a case that makes the issue I'm talking about pretty clear.
[...]
So I think it is clear it is a visual bug as I have stated originally.

Then it could always be observed I'd say!?

PPetar wrote: Mon Nov 24, 2025 3:15 pm So the fort effect is sometimes shown, sometimes not from what i can tell. Is this a bug? Is it a known one?

This may be it. This explains some of the larger variances.
But it seems to require a certain chain of events (or is random).

Sometimes it is like the prediction (as you say) removes "-1" from the Hex defence CV multiplier (maybe this also happens to terrain modifiers (without Forts)? Have you checked this?
- If it would only be the Forts randomly losing one +1 (100%) multipilier it feels somewhat buggy.
- If terrain also loses it it may rather be WAD.
- If it is (only) applied to recently built forts we may check in which order during logistics things happpen (e.g. are Forts being built before "Ground Recon values are being compared" or so).

Can you reproduce this in the opposite direction (Axis vs Soviet)?

What I also could imagine that the idea at some point was to not give 100% accurate info on the defender's fort level (like not accurate percentages/not giving any levels if a certain intel level isn't reached) but that got skipped but this predictior variance left in the predictor?!

Wiedrock wrote: Mon Nov 24, 2025 8:08 pm - If it is (only) applied to recently built forts we may check in which order during logistics things happpen (e.g. are Forts being built before "Ground Recon values are being compared" or so).

Logistics Phases from 1-2 years ago.
"Unit Build Forts" (1st section) happens before "Auto Land Recon" (next-to-last section).
So this seems to not be it as you have said already.

PPetar wrote: Mon Nov 24, 2025 7:02 pm I thought about it being a bug where a newly built fort is not counted, but I checked my aar and the 10=10 we see in the first picture of my original post had a fort level 1 as well one turn before

Wiedrock wrote: Mon Nov 24, 2025 8:08 pm Then it could always be observed I'd say!?

I said and posted screenshots of it sometimes being there, sometimes not, so no it cannot be observed always, though i see it pretty often as you can see from the screenshots. If you are asking about whether other people see it, i really don't know.

Wiedrock wrote: Mon Nov 24, 2025 8:08 pm

PPetar wrote: Mon Nov 24, 2025 3:15 pm So the fort effect is sometimes shown, sometimes not from what i can tell. Is this a bug? Is it a known one?

This may be it. This explains some of the larger variances.
But it seems to require a certain chain of events (or is random).

Yes i said in my first post it really seems like that is the issue. I don't know what causes it, thats why i am here.

Wiedrock wrote: Mon Nov 24, 2025 8:08 pm Sometimes it is like the prediction (as you say) removes "-1" from the Hex defence CV multiplier (maybe this also happens to terrain modifiers (without Forts)? Have you checked this?
- If it would only be the Forts randomly losing one +1 (100%) multipilier it feels somewhat buggy.
- If terrain also loses it it may rather be WAD.
- If it is (only) applied to recently built forts we may check in which order during logistics things happpen (e.g. are Forts being built before "Ground Recon values are being compared" or so).

Can you reproduce this in the opposite direction (Axis vs Soviet)?

What I also could imagine that the idea at some point was to not give 100% accurate info on the defender's fort level (like not accurate percentages/not giving any levels if a certain intel level isn't reached) but that got skipped but this predictior variance left in the predictor?!

I do not work for Matrix or 2by3 and was not involved in the development of the game, nor am i a tester. I do not know why would you assume i know/did those things. I spent a couple of hours investigating this and will not be pressured into investing even more time since i really do not think it is my responsibility to do so. I helped the devs by finding, investigating and reporting a bug and really do not think there is anything more that should be asked from me to do. It is up to you if you want to test these things yourself, but I would really like to speak to and was directing this discussion towards the actual developers of the game. We can spend days guessing what the code is and why does this happen but I really do not see why would that be my responsibility nor why would I spend multiple hours guessing about something that I cannot really know without seeing the code.

From the photo you sent it looks to me like you have this issue too, so I guess its not only on me?

CPPs are gained, and forts are built at the very end of the movement phase (when you end your turn, remaining MPs are used to gain CPPs and build forts).

If you have a save, you can attach it to a post and tell us what unit to look at.

Joel Billings wrote: Mon Nov 24, 2025 5:56 pm Wiedrock usually provides good insight into what's going on.

Is Wiedrock an official go to person regarding these things or am I missing something. How can he help tell what the issue is? Was he involved in writing the code?

Joel Billings wrote: Mon Nov 24, 2025 9:11 pm CPPs are gained, and forts are built at the very end of the movement phase (when you end your turn, remaining MPs are used to gain CPPs and build forts).

If you have a save, you can attach it to a post and tell us what unit to look at.

The unit i showed had 3 mp at the end, so they maybe gained little CPP. I attached the save, look at the division I highlighted in my post, end the turn, check the fort level and shown cv for the Soviets, attack it and check actual cv.

PPetar wrote: Mon Nov 24, 2025 9:10 pm From the photo you sent it looks to me like you have this issue too, so I guess its not only on me?

It can't be seen in any of my "Clear" Hex examples, nor the picture.

Loaded your save, ended the German turn.

With FOW off, it shows 4=7 (fort level 1, 2 CPP), it's real value is 4.05/7.91, which seems correct
With FOW on, it shows 4=4 (FOW can show you incorrect info, so nothing wrong here)

When I run attacks on the unit, it shows up having a starting cv of 79, so it seems to be working. I noticed this save is from June 2023. Are you using the latest version?

The save is some random save I have from a multiplayer game i played couple of years ago. I agree FOW can show you incorrect info, but how is it always around /2 and why is only the defensive CV incorrect, the offensive CV has much lower variation (also why does it always show x=x, not x=1.5x or x=2x or something)? It seems much, much more likely to me that the defensive CV doesn't count the fort in the hex. My game client is currently updated to the latest version, 1.04.12, the save was like i said from an older version. I suggested in my first post that a version missmatch could be the issue since my current multiplayer game started on an older save.

@Joel:

Joel Billings wrote: Mon Nov 24, 2025 9:34 pm FOW can show you incorrect info, so nothing wrong here

As you say.

I don't see it having particularly being related to the Fort level itself.
He is expecting things to be accurate (from an intel perspective) which they simply aren't.

It maybe has a bit of this:

Wiedrock wrote: Mon Nov 24, 2025 8:08 pm If terrain also loses it it may rather be WAD.

And certainly a whole lot of this:

Wiedrock wrote: Mon Nov 24, 2025 5:45 pm This is one of Gary's many layers in the game that makes it impossible to know with certainty how combats will end (in this case even how they will start).

Attached you can see that a variance of 50%+ is also possible in a Light Wood Hex without any Fort, and more...

It is cerainly weird how so many "same = same" are shown, but likely yet another layer of RNG, at least I've not seen a straight forward pattern on that yet (more often appears to happen in low defense modifier Hexes tho).