I have a friend that said we can not simulate HRL vs HMG. I told him my usual response, "you can not". I have here R code for that simulation with no packages. For ARO I may have to use a package but that was not the question. I tried to keep things rather straight forward for naming. If you have questions please let me know. HRL tended to do a good bit more damage the the HMG (I think its highest was 10 in one run). Simple assumption: The same model is shooting both the hrl and hmg at the same target. That range modifiers cancel out other modifiers This is in the models active turn That the model is in a haris (though you can easily tweak it) Each run represents an order spent shooting Though our fireteam go to is nyokas there +1 damage is not considered. That the dice being used are fare dice and can be represented by a uniform distribution HRL is firing in blast mode vs the target in cover. #code starts here #Setting base values bs_haris<-12 armor <-4 cover <-3 hrl_burst_ft <- 3 hmg_burst_ft <-5 hrl_dg <-14 hmg_dg <-15 #Firing up HRL hrl_haris<-runif(hrl_burst_ft, min=1, max=20) hrl_haris_rd<-round(hrl_haris, digits = 0) hrl_hr_hit <-hrl_haris_rd <= bs_haris hrl_hit <- sum(hrl_haris_rd <= bs_haris) hrl_crit <- sum(hrl_haris_rd == bs_haris) hrl_wd = hrl_crit+hrl_hit hrl_wd #damaging with hrl blast mode save_roll <- runif(hrl_wd, min=1, max=20) save_roll save_roll_rd <-round(save_roll, ,digits = 0) save_roll_rd save_arm_roll <- save_roll_rd+armor save_arm_roll sum(save_arm_roll < hrl_dg) hrl_wd_cause <- sum(save_arm_roll < hrl_dg) hrl_wd_cause #modeling continuous damage hrl_wd_init<-hrl_wd_cause hrl_wd_tr<-hrl_wd_cause hrl_wd_init repeat { # statements save_roll_rep <- runif(hrl_wd_cause, min=1, max=20) save_roll_rd_rep <-round(save_roll_rep, ,digits = 0) save_arm_roll_rep <- save_roll_rd_rep+armor print(save_arm_roll_rep) hrl_wd_cause<- sum(save_arm_roll_rep < hrl_dg) hrl_wd_tr<-hrl_wd_tr+hrl_wd_cause # checking stop condition if(hrl_wd_cause < 1) { print(hrl_wd_tr) break } } #Firing up HMG hmg_haris<-runif(hmg_burst_ft, min=1, max=20) hmg_haris_rd<-round(hmg_haris, digits = 0) hmg_hr_hit <-hmg_haris_rd <= bs_haris hmg_hit <- sum(hmg_haris_rd <= bs_haris) hmg_crit <- sum(hmg_haris_rd == bs_haris) hmg_wd = hmg_crit+hmg_hit hmg_wd #damaging with hmg blast mode save_roll <- runif(hmg_wd, min=1, max=20) save_roll save_roll_rd <-round(save_roll, ,digits = 0) save_roll_rd save_arm_roll <- save_roll_rd+armor+cover save_arm_roll sum(save_arm_roll < hmg_dg) hmg_wd_cause <- sum(save_arm_roll < hmg_dg) hrl_hit_success<-hrl_wd hmg_hit_success<-hmg_wd hrl_hit_success hrl_wd_tr hmg_hit_success hmg_wd_cause
What exactly is the question you’re trying to solve? I’m the author of the dice calc linked above, please feel free to shoot me a PM if you need assistance setting up a scenario that you want to check the probability of, or if you want to chat about how to do face to face roll math.
Infinity dice calculator is a dice calculator that gives probability for a normal distribution; it calculates probability by taking a large sample. You can actually look at total rolls and actual rolls made and see this. What it does not do is simulates a single roll for active comparison. For instance to get rolls to come up with a comparison it needed 12651264, so let’s go crazy and say you in the active turn have 48 orders in a game (the 48 is 16 times 3) and that they can all go into shooting, that is 263568 games. May be I am wrong and each die represented is actual part of the actual roll made and it is not aggregated by order. That would mean the 48 is actually 384 (240 for the HMG and 144 for the HRL). This works out to 32946 games. This is in essence mathhammer, now it is very good mathhammer. How to interpret infinity dice calculator is given enough orders, by law of large numbers, the probability of an event happening is some percent. The law of large numbers will let poisson distributions (the distribution used to represent count data) approximate or become a normal distribution. It shows and tells you nothing about each individual order you spend nor does it calculate the number of orders need to achieve a goal. To see individual order use you would want to set up a simulation of each individual order. What I did is a simulation of an order spent for the point of a comparison. If you do it a few thousand times, it can be used to approximate the dice calculator but that is not its intended goal for a 1 sided set of roll, as said ARO can be done but are not currently in it. If I were to set it in a loop vs an enemy I could also get number of order spent to remove the given enemy, I did something similar to this when I tested N3 crit vs N4 crit mechanics on a Jotem. Now for the power in this. Because the code is set up for a comparison, I do not need to worry about overall probabilities. Hence the only probability in the code is the dice which is a uniform distribution so a mostly equal chance of any number appearing (there is a small error there with rounding and the 1 to 20, should be a .6 to 20.4). Furthermore, this allows for the setting of shared characteristics in 1 spot. TLDR version: one calculates probability in a normal curve while the other show possible event output per an occurrence for comparison that in multiple runs forms a poisson distribution. They are not the same. I hope this makes sense.
I had no question to solve, just wanted to show some code and put another tool out there. For those that do not know R is free and open source. Story: I was talking to a friend and we where discussing HRL vs HMG in the new meta as crits has changed and armor is more important now. They made the statement that you can not simulate continuous damage. My response was "you can not" with the implication that I could. The tone may not come out in text. This is sort of like people who say you can not compare models cross factions, well if you believe they balance values with math and then add some fluctuation then you can. The how can be explained another time, as that can start a debate. I tend to do a number of simulations to compare performance rather then look at probability in a normal distribution. I think this gives a better idea of economy of orders. I feel simulations comparison and probability are equally valid approaches to evaluating actions in the game. PS: I was in favor of the HRL. It just happens to be a bigger risk reward weapon but the continuous damage and the blast mode I thinks tips it in its favor vs the HMG for hard targets.
Feels like there's still a whole conversation missing for context, but... what units are you looking at? I usually find that practice and normal distribution aligns quite well for these two guns, with external factors making a difference like proximity to other units and the presence of a Fireteam, and more importantly whether the unit needs to do ARO duty or not.
No particular unit. Though I could move from a generalized form to units easily. Units tend to have a lot of additional not often need information for shooting comparison. For instance for a Syurat vs and Ork all I have to do is increase the bs by 1 or Nyoka vs and Orc I increase the damage by 1. Other things to add depending on unit: MSV, additional negative modifiers, burst and damaging that effects that are independent from the general model rules are all important. If you are comparing to non similar range band guns you also have to factor that in as a modifier to there BS. When ARO is added then the dodge would also have to be add. After that you could add a wip roll to see if they seek more cover. Once that is done you are simulating the entire event chain which is not the initial purpose and I try to avoid scope creep. The purpose was to see a comparison of performance, the more you add the more convoluted the comparison gets. I am not looking at ARO duty from these guns, I may add ARO from the enemy but for now I am looking at active turn. People could claim that your missing the opponent ARO and that is a major part of the game, which I would agree with. The main power of the HMG is the amount dice you have to overcome ARO. Your luck sounds more reliable then mine (my luck tends to be crazy, like killing a Jotem with HRL crazy). As for a conversation of missing context maybe but not sure what all you want to know.
It's more about tempering expectations and handling risks than having average luck, though it does tend to leave me at a loss for decent ARO weapons due to me looking at the active turn and relying on hiding during reactive turn a bit too much. I find that without looking at specific units it's pointless to look at the individual guns. The unit carrying them makes a world of difference, particularly if they can get Haris or Core bonuses which I don't think makes e.g. a Missile Launcher good enough to use actively, but pushes the HRL high enough compared to potential AROs that it matters a lot.
I tend to choose my guns, then role, and then put guns them on good units for that role. Working in reverse to fill in gaps on how I want to fight more then say I need X unit and then slap a gun on them, a hard lessen I learned from when I went to an open board at a tournament and had no guns for that range. Units can effect the gun performance but understanding the guns natural match up has worked well for me. For instance an Omega HMG is great but I would not reach for him if I want to fire fight armor 5 plus cover for a total of 8. Yes the Omega will probably win the face to face but his shot are more likely to kill the enemy efficiently where a Nyoka HRL may has a worse chance at winning but has a better chance of doing damage but if I am fighting light or medium infantry the Omega HMG works well. This is because HMG and Spitfires are good at shredding lighter targets, HRL and ML are better at getting around cover, and AP is better vs armor 6+, Snipers are good at killing HMGs while they are in their deployment zone. A model is more then the sum of its parts and there are multiple ways of looking at them, braking them down and approaching who or what you want.
So it seems like people like comparing models vs models. I have an issue in the fact that choosing the right target drastically effects the results and cause favoritism, so what target do you all choose when comparing guns?
The one I'm shooting at. If I want to see how various models/profiles will work against, say, a Tanko ML, then I put a Tanko ML on the right side. Or if I want to see how various models will work vs. a Taagma Sniper, I put a Taagma Sniper on the left. I must admit I'm still confused as to what you're trying to model. For any given model, the B4 of the HMG will outperform the B2 of the HRL in active turn; in ARO they're pretty equal though the HRL hits harder due to ignoring the ARM boost from Cover + Continuous Damage.
How do you even begin to compare how a gun shoots when you have no stats to compare to? I usually compare best case, worst case, and expected case. I also usually look at rangebands that I can expect to use the guns in, which for e.g. a combi is several different and for a sniper tends to be exclusively +3 due to how generous it is, and HMG either +3 or -3 because of where I expect to deal with snipers or manipulate TR REM odds (the +0 band isn't very interesting unless it's a TR REM). Results in a lot of comparisons, but I there are just so many data points.
Doing during the game vs the target is a given. As for the performance of the HMG vs HRL in the active turn you are incorrect. The HMG has more dice so it can hit more, not a defiantly hits more but a possibility of hitting more. Then the wounding performance varies and swings because of the armor and cover of the target causes the hits from the HMG to be less effective. Then the wounds caused by the HRL have continuous damage so they can cascade to more damage. This is why it is a simulation and not straight probability, as one can create a more realistic view from order to order.
If you look at the code and assumption you could notice that there a line that has bs <- 12, armor <- 4, cover <-3 and a line that assumes the range modifiers cancel out. This sets up one of the most common cases of shooting in infinity, good range bands but with cover, and buts it at bs 12 vs armor 4. You do not need models to have values, to specify a models move the simulation from a generalized form comparison to a specified comparison. Further more in this case the HMG and HRL have the same range bands, this to can be changed.
... but you know all those numbers. And can crunch those numbers using the dice calculator. Even with the damage difference, +2 Burst and +1 damage is going to outperform the HRL's continuous damage almost always. Even against a Jotum in Cover, the highest ARM target in the entire game, an arbitrary BS 13 model in cover shooting both will look like this: (HMG on top, HRL on bottom; I didn't bother setting up the Jotum's HMG so the exact numbers for the Reactive Player are off). Even in the absolute BEST CASE scenario for the HRL, the weapons are extremely close in terms of offensive output, but the HRL has a greatly increased chance to lose the FtF and take a hit (as you can see by the 15% vs. 22% chance). Hence why I'm confused as to what you're trying to model. The numbers are already known.
Like I wrote earlier, though; the situation on the table dictates which is better. HMGs almost always are better at forcing single pieces (such as HRLs), while the heavy rocket launcher (and missile launchers) is excellent at hindering fireteam movement. If I can afford it, I try to take both, but I'll try and get the HRL (or HRL equivalent disposable) on ARO pieces and HMGs on active pieces. CB does have a tendency to railroad you into those choices, though.
There is a button show raw stats. Under there you will see Actual Rolls Made. That is how many times the program rolled dice to get those numbers. is that number small enough to happen in a game. The dice calculator does takes a distribution from a large numbers simulation and shows that as the probabilities of something happening. This gives us a look at probabilities as a whole but nothing else. Now to use yours data: HRL has a .2% chance of killing it, well unconscious level 2 because of remote presence. A .89 % of doing 3 vs .07% of the hmg, 12.7 times increase in favor of the HRL. a 3.73% of doing to vs the hmg 1.39%, 2.68 times increase in favor of the HRL. The probability of doing 1 wound is close, all others have the HRL as a multiple of the HMG. So how many orders will it take to kill the Jotum? Answer: you do not know from this. The fact that it will happen eventually does not take into account the resource in the fuel that eventually. Where is the brake point for something being worth more because it could save you orders? Hence to see what could happen in an order/round a simulation is a better. I started by modeling a comparison between the two on an order basis because everything needs a start.
That was true in N3 where crit fishing could murder everything and as such armor was less important. I do not think that is the case anymore with how things have changed in the game. Essentially you are relegating HRL to being an ARO only weapon when it has more uses, especially in a link. I have found chewing my way through HI in cover with HMG or Spitfires to eat up my orders for little gain.
A HRL is only +2 DAM over an HMG, and it's still true that most ARO pieces are weaker units, so the two extra burst will usually have better yields. It's become more important to have an AP HMG or better, or otherwise have a plan for mad lads putting TAGs on ARO duty, such as Smoke/Eclipse, Pitchers, or otherwise dealing with the problem. So wrapping back around to actual cold hard facts, I do think the Gamma is a great tool to have in N4, but I'm thinking about the Feuerbach profile and not the HMG, with the probably biggest contender for the role being Knauf (I don't like the conceptualization of Starmada so I'm not considering the sectorial's options). For HMG I like the simplicity of the Omega, but for budget concerns Epsilon and a bunch of Varangians will typically do nicely. Then regardless of who's pointman and who's TAG teaming wingmate, a Dakini HMG is always good to have around as a solid plan B. More than three suppression pieces (you have to give your opponent a reason to cower behind walls) and I'll just jeopardise mission, though, and me being bad at completing objectives in the first place... well... that just won't do.
Personal opinion: Gama Feuerbach is great, I love the AP DA even at only two shots. It also have the wounds to take a bad round of aro while it marches forward if need. HMG I go Omega or Epsilon though I have liked the Nyoka HMG, it is damage 16, because my meta is very heavy into MSV ARO models. Dakini: My first army was SP and then Aleph as a whole so if I want to play them I just go back to them and would rather not have them in O 12. LRL Crushers are also useful. I have fun with a Beta trouper with spitfire and constantly debate changing it to a Bronze. I have played a little with the tag, 5 dice is fun to role. I feel O 12 is missing a missile equipped HI and a CC HI, then it would have all the HI I would want. Yes the Bronze can do CC but he is not to the level I would like.
