Something odd with the hit odds

[7thGalaxy]

So I've got a battleship dueling a few cruisers at about 12 km. I've got 3 quad 15 inch turrets. 

 

The odds of hitting per barrel are purportedly about 4%, giving combined odds of one hit per salvo at about 40%. I'd expect, therefore, a hit just over every three salvos. 

I'm actually getting a lot lower than that: about every 6. I get that there is rng involved but the odds are rolled so often that it should tend to even out. Either the calculations of the odds are wrong, or the simulation of the shells.

I suspect the later: I get that these are moving targets but the computation on the fire control was pretty sophisticated, and the guns accurate. 

[Cpt.Hissy]

There's no shell simulation, just hit chance roll and if RNGesus said you miss, you will miss no matter what.
Fast and smol boats seem to have some additional evasion buff for just being smol and fast, that may be not accounted for in your hit% reading

[7thGalaxy]

2 hours ago, Cpt.Hissy said:

There's no shell simulation, just hit chance roll and if RNGesus said you miss, you will miss no matter what.
Fast and smol boats seem to have some additional evasion buff for just being smol and fast, that may be not accounted for in your hit% reading

In a sense this is the case even with a detailed simulation surely?

 

[Cpt.Hissy]

20 minutes ago, 7thGalaxy said:

In a sense this is the case even with a detailed simulation surely?

 

What i mean, with proper simulation you'd have a projectile going in random direction, and then it depends on if enemy is there when it arrives.
Here if it's decided that you miss, shell will purposefully aim past the target no matter where it is or what it does.

[SPANISH_AVENGER]

Same here. Sometimes I am facing another Battleship at a 5 km range, with a 60% chance of hit or more... and I can perfectly fire 7 salvos without scoring a single hit. Not only that, but the shots go on entirely wrong directions, so it's not even a disperssion matter. Like, shots fall 300m ahead of the target and stuff like that. Very odd.

Edited February 8, 2021 by SPANISH_AVENGER

[The_Real_Hawkeye]

I posted this in April last year:

 

15 to 20% to-hit chance, 3 BBs, each with 4 x 2 12-inch guns. With this, I'd expect to get some 3 to 5 hits with each 3-ship-broadside. As is, I get perhaps a single hit per 3-ship-broadside - if I'm lucky.

As I see it, there are 2 possibilities. 

Either the game is blatantly lying to me or the RNG is complete and utter ***** - which, come to think of it, isn't unheard of in games.

 

I stand by what I said back then. The odds of RNG rolling low _that_ consistently (IIRC for about 20 salvos or some 500 shots) are so abysmal low, my chance of being hit by lightning while I cross the street would be higher.

[arkhangelsk]

Personally, I just use it as a guide - anything under 10% means hits are luck and I need to get closer or just be very patient. Over 10% I can expect hits with regularity. Over 20% and I expect to get a hit almost every salvo, and so on.

[The_Real_Hawkeye]

The thing is, if the game gives me a percentage, I kinda expect that percentage to be reflected in my actual hit-rate.

If it doesn't, just don't give me a number and just list the hit-chance as abysmal, very poor, poor, average, good, very good, exceptional.

When done like that, no-one can complain - of course, no-one can check if things are actually working as intended either.

[coalminer]

Its quite silly to have ranges <10km where the barrels and turrets are pointed in the right direction but when the guns fire, the shells veer way off like the gun barrel is bent, firing round balls with no rifling. its like playing FPS games where the guns can hit the far corners of the screen without even moving the crosshair.

[Right]

logically and actually it doesn't metter what system is used, the rng gods are in play and sometimes you get it all while other times you lose it all just like in real life.

don't go there, just accept that luck if we like it or not is part of the game. your role is to try and min-max it to your favor.

I did a test to check if the info is right, after hours of testing it is, the chances to hit are as what shown. if you would follow the log closely for a long time and check you would see that too.

[madham82]

37 minutes ago, Right said:

I did a test to check if the info is right, after hours of testing it is, the chances to hit are as what shown. if you would follow the log closely for a long time and check you would see that too.

I've seen no evidence salvo accuracy is correct. It is easy to get it to 100%, but still fire 4 or more salvos without a hit. The other accuracy number (under the actual gun display) might be more accurate.  

[Right]

1 hour ago, madham82 said:

I've seen no evidence salvo accuracy is correct. It is easy to get it to 100%, but still fire 4 or more salvos without a hit. The other accuracy number (under the actual gun display) might be more accurate.  

whenever I had 100% all shots did "hit", sometimes they don't do damage because there are other factors in play and sometimes I thought I had 100% but the movment of the ai to dodge my shots got him to get away from it. there is a minimum spread for each gun and if you have for example 3 barrels, sometimes 100% would miss 2 of the 3 just because of that spreading too big to hit with all at the ship.

The decision if something hit happens only when the shell reach the target because there are 2 checks for a hit, when it leave the gun and if it reach the target, the % is only when it leaves your gun.

so you can argue it isnt fair for a game to show you 100% and then something happens along the way that changed it but it sounds very realistic to me.

 

Edited February 10, 2021 by Right

[madham82]

38 minutes ago, Right said:

whenever I had 100% all shots did "hit", sometimes they don't do damage because there are other factors in play and sometimes I thought I had 100% but the movment of the ai to dodge my shots got him to get away from it. there is a minimum spread for each gun and if you have for example 3 barrels, sometimes 100% would miss 2 of the 3 just because of that spreading too big to hit with all at the ship.

The decision if something hit happens only when the shell reach the target because there are 2 checks for a hit, when it leave the gun and if it reach the target, the % is only when it leaves your gun.

so you can argue it isnt fair for a game to show you 100% and then something happens along the way that changed it but it sounds very realistic to me.

 

I don't think the game is simulating like you are thinking. This goes back to what @Cpt.Hissy said about the game is deciding before the shells are fired if there will be a hit. His explanation of why the shots go wild, matches what many have seen at very close ranges with the shell going in completely opposite direction. There are arbitrary penalties/bonuses being done by the game that clearly show something funky is going on (i.e. the rudder position was giving a full maneuvering penalty despite the ship not moving).  

That said, we are all guessing how things are implemented b/c the Devs have never clearly stated how the gunnery is being calculated. But if I have 12 barrels with 100% salvo accuracy, that implies at least one hit out of 12. Which certainly isn't happening in the game. 

[Right]

13 hours ago, madham82 said:

That said, we are all guessing how things are implemented b/c the Devs have never clearly stated how the gunnery is being calculated. But if I have 12 barrels with 100% salvo accuracy, that implies at least one hit out of 12. Which certainly isn't happening in the game. 

I do speculate but I have a gift to recognise patterns, the reason why I find the ai so lacking, and I can't say that I'm surly right about it but after watching tens of thousands of shells in the game that what my observation tought me.

Again, I might be wrong and we would only know for certin when the devs would decide to share the information but I can see no other logic in a computer simulated environment, because as you said 100% not always result in a hit, still it is hardly logical that such a simple formula wouldn't work properly.(I have a programming degree)

[Cptbarney]

On 2/9/2021 at 4:54 PM, coalminer said:

Its quite silly to have ranges <10km where the barrels and turrets are pointed in the right direction but when the guns fire, the shells veer way off like the gun barrel is bent, firing round balls with no rifling. its like playing FPS games where the guns can hit the far corners of the screen without even moving the crosshair.

Ye tried to explain that in the alpha 10 thread, guess i didn't write it properly (that maybe reading fail idk), especially when you have two pre-dreads about between 1-4km and you see both shells land either side of the other ship as if the guns were pointing in those directions despite the barrels pointing straight ahead.

Which shows the shells themselves aren't simulated, but generated individually based on loads of random calculations with in set ranges of each calculation. So we need a proper gunnery system even if it is basic, moving away from RNG is always a good thing.

[Right]

1 hour ago, Cptbarney said:

Ye tried to explain that in the alpha 10 thread, guess i didn't write it properly (that maybe reading fail idk), especially when you have two pre-dreads about between 1-4km and you see both shells land either side of the other ship as if the guns were pointing in those directions despite the barrels pointing straight ahead.

Which shows the shells themselves aren't simulated, but generated individually based on loads of random calculations with in set ranges of each calculation. So we need a proper gunnery system even if it is basic, moving away from RNG is always a good thing.

mybe the barrels and the shells aren't simulated together, which for visual effect needed, but the shells are working as physical calculation by themselfs, probbly in a liniar back calculation made behind the actual visual effect.

I do agree and would be happy if somehow the rng would be put down a notch but  doubt that would happen. still think it is a good idea.

[Cptbarney]

1 minute ago, Right said:

mybe the barrels and the shells aren't simulated together, which for visual effect needed, but the shells are working as physical calculation by themselfs, probbly in a liniar back calculation made behind the actual visual effect.

I do agree and would be happy if somehow the rng would be put down a notch but  doubt that would happen. still think it is a good idea.

Im not sure if the barrels are, they are probs just an object generator so a empty child is put into the barrels and the projectile 'fires' when certain conditions are met. I could be wrong however, i know the barrels are animated, but the way the shells act it seems like they are calculated as separate objects from start to finish, which if a new parameter was added in like different shot types, we might see more consistent shell patterns idk.

But the system seems pretty basic, we need rangefinding, predictions and co-ordinates to make it more realistic than what it is atm. I guess it's ok for now, but they could update the system later on.

Too be fair without the source code we can only guess really.

[Cptbarney]

Also i forgot to mention, we may see this occur during this year as i doubt game-labs is happy with the current iteration entirely. Seems to be prototyping atm, advanced but still prototyping.

[CapnAvont1015]

6 hours ago, Cptbarney said:

Im not sure if the barrels are, they are probs just an object generator so a empty child is put into the barrels and the projectile 'fires' when certain conditions are met. I could be wrong however, i know the barrels are animated, but the way the shells act it seems like they are calculated as separate objects from start to finish, which if a new parameter was added in like different shot types, we might see more consistent shell patterns idk.

But the system seems pretty basic, we need rangefinding, predictions and co-ordinates to make it more realistic than what it is atm. I guess it's ok for now, but they could update the system later on.

Too be fair without the source code we can only guess really.

You know now that were talking about this I want bring up something that's been bothering me for a while and that is I don't think the shell damage match where they actually hit. Just today I got hit on the side of the bow way above the water line and I still got flooding. I don't know if I'm just crazy or what but it's very weird.

[Cpt.Hissy]

12 minutes ago, CapnAvont1015 said:

You know now that were talking about this I want bring up something that's been bothering me for a while and that is I don't think the shell damage match where they actually hit. Just today I got hit on the side of the bow way above the water line and I still got flooding. I don't know if I'm just crazy or what but it's very weird.

Yep they don't match sometimes, I also had hits to the bow and damage in the stern, but this might be some timewarp-related visual glitch with about the same probability as actual "feature"

[7thGalaxy]

I was thinking about this yesterday when trying to torp a battleship with some light cruisers. Is it worth taking evasive action? With longer range battles the flight time of the shells is enough with a nimble ship to have moved considerably from where the enemies fire computer would have calculated. 

But if the hit odds are rolled at the firing point how does that work?

[Right]

1 hour ago, 7thGalaxy said:

But if the hit odds are rolled at the firing point how does that work?

Simple, they don't, the camplain is about not seeing the % come to life as predicted, when a ship makes a shot it is just a prediction, the hit is calculated only when and if the shells get to the target, just like torps.

In a way the % is just a computer on the ship that tries to tell you what the chances are to hit but it cannot predict 100% for the object that it tries to calculate is a moving self controlled object, so you can say the odds are "off" but they are just an indicator to begin with.

Using evasive moves is very beneficial, the system even shows you how much it change the chances to hit before the shells leave your gun, it still takes effect  and change the actual % after they left.