S1914 Battle Calculator

  • FMD

    Approved the thread.

    Hi, I tried the calculator a bit, but couldn't get it to work.

    1) Issue 1: No 'convoy' or 'troop ship' unit type available for selection. If i choose infantry at sea, I get the error "oops: can't have Infantry at sea."


    2) Issue 2: unit count seems to be capped at 999. It throws an error when the unit count is 1000 onwards (error message "oops: format of unit count (1000) is wrong."


    Thanks!

    Any sufficiently advanced tactic is indistinguishable from magic to the uninitiated.

    Yeah, that would probably be useful for supremacy, where I take it the use of forts is pretty common. I have some info about the mechanics vs buildings, but to model fort damage I would have to add options for all other buildings types and their HP. I suppose I could just apply a portion of the building damage to forts each round which may be better than nothing. In the meantime, one could manually run 1 round at a time and knock off a bit of the fort bonus each round.

    We are now experimentally allowing the fortress to take building damage which reduces its effectiveness each round. Normally, the chance that a fortress is damage and how much depends on other buildings in the province. For now this is ignored and the fortress is arbitrarily getting 1/3 of the incoming building damage per round. Hopefully, this is better than not damaging the fortress at all. If it becomes a problem it will be changed or removed. Allowing other buildings may be added in the future.

    So I've been using this extensively in the last few days, and I must say that it makes a huge difference.

    For some reason unknown to my opponents I always make the right calls when deciding which fights to choose and which to avoid.

    I couldn't play effectively without it.


    I think the current approach to deal with fortresses is not very accurate. In some cases, where there are few buildings other than the fort, the fort will take almost all the building damage. In other cases, if there are few buildings present, the fort will take a small share of damage in the first few rounds of combat, and a large share afterwards. In provinces with a lot of infrastructure, the first phase will take longer.


    I'm not sure if there is an easy way to deal with that. One possibility would be to make the user select the share of damage that goes to the fort. Another option would be to input fort hitpoints and other hitpoints separately and work from there.

    Thanks for the feedback! I'm considering allowing the user to input all buildings and their hit points. Even then the variance of fort damage from battle to battle is much larger than the variance in damage to troops. Each round a building is picked randomly (with forts getting more weight) and damage applied. Thus, by chance the fort could get hit every round or never get hit. If I add other buildings I will probably do some portion of damage to the fort each round based on the expected value given other buildings. This is basically what I'm going now except that I'm assuming some constant number of other buildings. But, what I will do now, even before I add other buildings is to change the behavior when "Simulate Variance" is selected. When this is selected I'll give the fort some chance to take damage and if it's hit it will take full damage. Running this multiple times should give a better idea of the range of possible outcomes. I might be able to work on this later today.


    I have a pretty good idea of how building damage works in COW and assume it is the same or similar in S1914. If there are differences perhaps freezy can let me know :)

    Buildings can now be added to the calc like units. See https://dxcalc.com/share/s1914.info.html#forts for more info.


    As mentioned above a battle with forts and other buildings can have a very high variance so it's probably a good idea to run the battle several times with "Simulate Variance" checked. Without this the calc will give the average expected outcome and all buildings will be damaged proportionally (with forts taking 50% more) each round. For an example of the variance I ran the battle below several times and it varied from side B losing as little as 44% of its total HP to completely losing the battle. Thus fortresses and buildings can cause huge variances in units killed and this probably occurs similarly in-game.


    So I ran the above battle 500 times to get an idea of the distribution of Army B's losses (shown below). The distribution is pretty normal except for a large outlier at 100% (B losses the battle). This is probably because there are a lot of different ways for the randomness to shift the balance in such a way as to cause B to crumble. The meaning of the colored vertical lines are as follows:


    red: the mean of the distribution

    blue: the median of the distribution

    green: the average expected outcome (ie. without "simulate variance")


    Since the expected outcome and the median are pretty close it means the number of battle outcomes that are better and worse than the average expected outcome is about the same on either side. However, the mean being higher than expected indicates that battles that are worse than the expected outcome are more extreme (on average) than battles that are better than expected. This distribution may be specific to this particular battle, but this demonstrates that it's probably a good idea to run several battles with 'Simulate Variance' when you have forts and other buildings to get an idea of the distribution.


    Edited once, last by DxC ().

    Ok. Here is a summary of what I learned from using that tool.


    1. Contrary to my previous belief, fort's don't really matter that much in large land battles.


    Consider this battle:

    What changes if we change the fort level from 1 to 5, but remove the single tank in army A?

    Virtually nothing. The single tank is about as relevant to the defence of army A than a lvl 5 fort would be.


    Armies with tanks and inf shielding can assault forts no problem. They will tear down the fort in the first few turns of the battle, and still have enough attacking power because only irrelevant inf died. The variance doesn't matter here because of the law of large numbers. It's probably more relevant in smaller battles.


    Having more heavy units present is much more important than forts in large land battles.


    In my recent game, a huge enemy stack of 200 inf, 3 heavy tanks, decided not to attack my lvl 5 fort with 35 inf, 2 ac, 1 art in it. They would have obliterated me if the attacked. They instead sent bombers. Not a good idea.



    (note that they would have won if they had sent in more fighters).


    OK, I get it, they didn't know what size of an army I had there. But they were acting under the assumption that they would surprise attack me and force me to divert my main army from elsewhere. Given that situation, they should have attacked with the land force anyways.


    2) Naval infantry makes a huge difference.


    Consider this battle:

    Army A clearly outguns army B but is obliterated. Look at that loss ratio in terms of materials and cash!


    What happens if we change the number of transport ships in army A?


    Turns out they win the battle if they have just a few dozen more transport ships. They win with 35 instead of 15 inf, and the loss ratio in terms of material and cash quickly turns if you add just a few more.




    So this is what I meant when I say that I can make much better tactical decisions with that calculator. :)

    One issue with that last one that I haven't fully decided how to handle is how to value infantry in terms of resource cost and how to value transport ships in general. I am giving transport ships some arbitrary and pretty small resource costs, assuming they are mostly infantry, but the calc doesn't currently allow the user to specify what the transport is composed of (may change this later, but may not be worth the extra complexity). Also, it seems like infantry in general should perhaps have some additional cost, but currently they only get a time cost. Btw, the HP lost value in army B above is mostly from transport ships which have hp:10. But in reality the hp loss would be at least 20 per unit (infantry). So in terms of HP loss it would be a much bigger loss for side B. On the other hand the time cost is probably inflated for the transport ships. I had their time cost at 12 hours which is what I had for infantry for minimum time, but I think infantry can be produced in double resource province with level 2 barracks in about 6.2 hours. Is that right? I use the minimum production time for all units so they are comparable, even though they may not be produced at minimum time in reality.

    Edited 2 times, last by DxC ().

    I actually like the way time cost is currently implemented. True, time cost can be lower if provinces have double provinces or barracks, but they can be higher if provinces have low morale (for all unit types, as you mention). But that's not an attribute of the battle, where we want to use standardised metrics. So 12 hours per infantry.


    Implementing transport ships other than infantry would improve the usability of the calculator only in niche applications – like deciding whether or not to sacrifice a force in an attempt to destroy a transported railgun is economically beneficial. But in most cases other factors will be more important when making that decision.


    On a side note, how is damage distributed among transport ships of different unit types?

    Quote from Norton

    how is damage distributed among transport ships of different unit types

    As far as I know transport ships are treated as a single unit type having 10 HP and in the naval class. Thus, they are treated just as any other unit type given their class. I suppose in-game if a heavy tank transport ship at full health loses 5 HP it would have 260 / 2 = 130 HP when it got back on land. Is that right?

    By default it will run a max of 100 rounds. When the battle finishes it will tell you how many rounds it took for one side to win if it's less than "Max Rounds". This can be set to 1 or more at the bottom:



    Or are you asking how a round is defined? If so a round corresponds to one round of attack tics for one or both sides. If only one side is attacking it's the number of attack tics for that side. If both sides are attacking both will attack in round 1, both will attack in round 2, etc.

    DxC, I just used this tool to predict a sea outcome. I'm surprised to see the impact of transport ships as they should join the battle. How can 40 submarines easily destroy 19 cruisers, 25 ships and 7 submarines but lose if there is a stack of 1000 transport units added to the stack? I understand that damage is evenly distributed on the stack, but that's still strange because the game suggests the battle advantage is on the submarine's stack side by over 25%.

    Quote from CandideV007

    I understand that damage is evenly distributed on the stack, but that's still strange because the game suggests the battle advantage is on the submarine's stack side by over 25%.

    I'm not quite following your meaning. Are you saying that the in-game result differed from the calculator result?

    Edited once, last by DxC ().

    Hello there,


    Thank you for creating this tool!! I have checked it a little bit and tried it on actual battles and so far it has been pretty accurate.


    I have a fairly new Youtube channel (started about a month ago) about Supremacy with tips and strategies along with daily video logg of my current campaign. Check this link for the channel: https://www.youtube.com/channel/UC16GAEnIu-abF_ZARadleUw


    What about I made a video of me using this tool and I spread the word on youtube also? Of course with all credit to you and a link to the site in the video info. Let me know if you like that idea and we can plan the setup together. Send me a private message or reply and we can talk about the details.