Slitherine

marioslaz wrote:Why do you want to lost so much time with statistic calculation? I made too first times I want to explore the chances of a match up. Then I found it's very less time consuming to make a simple PC program to simulate a melee round (and also CT and death roll subsequent) and repeat this one million of times. After one million of such simulations, the average is very close to the chance calculated with mathematics formulas (error is under 0,1%). To repeat one million of times requires on my PC (Athlon 2600 2Ghz, a PC 3 years old) a few seconds. To write a such program it's very easy also for who isn't a great informatics expert and you can do it with freeware tools (I use Netbeans, but many other exist).

I modelled it in Excel this way - I only did it 65000 time because I am lazy

It recalculates in something under a second - never timed it as it is not relevant. I can then change a variable and it simulates another combat. Took me less than half an hour to create. Perhaps it was a waste of time, but I was not doing anything else.

Wargamers with mad Excel and stats skillz.

You guys must get all the chicks.

Spike

Biometric mathematician - limited to integers of 0 - 20, 21 on a good night.

spikemesq wrote:Wargamers with mad Excel and stats skillz.

You guys must get all the chicks.

Spike

Biometric mathematician - limited to integers of 0 - 20, 21 on a good night.

You've got it the wrong way round.

It's because we don't get all the chicks that we have time to develop the skills.

spikemesq wrote:Wargamers with mad Excel and stats skillz.

You guys must get all the chicks.

Spike

Biometric mathematician - limited to integers of 0 - 20, 21 on a good night.

What are 'chicks'?

timurilenk wrote:

spikemesq wrote:Wargamers with mad Excel and stats skillz.

You guys must get all the chicks.

Spike

Biometric mathematician - limited to integers of 0 - 20, 21 on a good night.

What are 'chicks'?

They are the ones who smell nice but flee shrieking when you approach.

HTH

Spike

timurilenk wrote:

marioslaz wrote:Why do you want to lost so much time with statistic calculation? I made too first times I want to explore the chances of a match up. Then I found it's very less time consuming to make a simple PC program to simulate a melee round (and also CT and death roll subsequent) and repeat this one million of times. After one million of such simulations, the average is very close to the chance calculated with mathematics formulas (error is under 0,1%). To repeat one million of times requires on my PC (Athlon 2600 2Ghz, a PC 3 years old) a few seconds. To write a such program it's very easy also for who isn't a great informatics expert and you can do it with freeware tools (I use Netbeans, but many other exist).

I modelled it in Excel this way - I only did it 65000 time because I am lazy

It recalculates in something under a second - never timed it as it is not relevant. I can then change a variable and it simulates another combat. Took me less than half an hour to create. Perhaps it was a waste of time, but I was not doing anything else.

I supposed you used statistical formulas to calculate chances like I did first times. It has some sense if you want to calculate a simple event (like who will score great number of hits) but it becomes a panic if you want to calculate chance of more complicated events (like chance to pass a CT, where you must consider how much hits you suffer, difference between hits made and hits got, and so on). With java I made a program with which I can simulate a full melee (not just a round, but the melee until one BG breaks). One million of simulations in just 4 seconds (this according to my program output... it seems a little more to me, but not so much I can blame it to be wrong). Then a lot of data (chances to win/lose, average length in turns of melee, average number of casualties, and so on).

The tecnique described is known as 'Monte Carlo' simulation. This is widely used in all sorts of areas of science - especially in Nuclear Physics. So it is a highly respected method which is often run on seriously large supercomputers at some of the most interesting laboratories in the world (when you want to do several million particles/photons interacting with a complex mesh of materials then the run time on a many teraflop supercomputer runs into days or even weeks.

Not much to do with wargaming but hopefully of interest. The data are interesting - has anybody managed to come up with a program to take these factors and turn them into precisely calibrated points values to get a perfectly even matchup?

paulburton wrote:The tecnique described is known as 'Monte Carlo' simulation. This is widely used in all sorts of areas of science - especially in Nuclear Physics. So it is a highly respected method which is often run on seriously large supercomputers at some of the most interesting laboratories in the world (when you want to do several million particles/photons interacting with a complex mesh of materials then the run time on a many teraflop supercomputer runs into days or even weeks.

Not much to do with wargaming but hopefully of interest. The data are interesting - has anybody managed to come up with a program to take these factors and turn them into precisely calibrated points values to get a perfectly even matchup?

I really know very few about Nuclear Physics, I'm only a modest Engineer
Anyway, my simulation pointed out a problem with the mechanic of combat: less dice more turns the melee lasts. Near 20% of melee with 4 dice and even PoA last over 6 turns with a max always over 25 turns!. Near 40% of melee with those conditions end with a No Contest (a tie or looser pass CT and death roll). With 8 dice those percentages become 10% and 15% respectively. This is the opposite I want to simulate an historical battle. I'm working with friends to look for a solution, but it's hard.

Fewer dice - longer melees is what I would expect (total hits per unit will be lower so Death rolls are less likely to result in casualties - especially for fights which are drawn or narrowly won. This is why armies which hae 4 base BGs are popular. one 8 pack vs two 4 packs is at a disadvantage as it is likely to be 1 or 2 casualties to nil on bound 1.

The list writers have been getting wise to this and 4 base foot BGs are getting rare (none for ther Romano British when they are essentially the same units as the Dominate Roman Auxilia).

If you are playing within a group of friends then use the larger battle groups in your armies which give the results you would like to see. Change the lists to force the larger groups if needed. Obviously this won't help if you want to play in competition.

paulburton wrote:Fewer dice - longer melees is what I would expect (total hits per unit will be lower so Death rolls are less likely to result in casualties - especially for fights which are drawn or narrowly won. This is why armies which hae 4 base BGs are popular. one 8 pack vs two 4 packs is at a disadvantage as it is likely to be 1 or 2 casualties to nil on bound 1.

The list writers have been getting wise to this and 4 base foot BGs are getting rare (none for ther Romano British when they are essentially the same units as the Dominate Roman Auxilia).

If you are playing within a group of friends then use the larger battle groups in your armies which give the results you would like to see. Change the lists to force the larger groups if needed. Obviously this won't help if you want to play in competition.

I respect your opinion, but it differs totally from mine. I don't understand your opinion in a point, because Death rolls are not less likely in draw or narrowly won melee, but only in lower dice melee. If you get 4 hits, for example, you have a good chance to make a kill even in a draw, and to get 4 hits is more likely obviously when you roll a lot of dice.
Apart from that, the size of BG we use is 4 for mounted and 6 for foot (with exception for Romans and Pikes) because we found are the best to represent the units we use. With such dimensions, melee of mounted last longer than those of foot, and this is the opposite we want.

marioslaz wrote: I respect your opinion, but it differs totally from mine. I don't understand your opinion in a point, because Death rolls are not less likely in draw or narrowly won melee, but only in lower dice melee. If you get 4 hits, for example, you have a good chance to make a kill even in a draw, and to get 4 hits is more likely obviously when you roll a lot of dice.
Apart from that, the size of BG we use is 4 for mounted and 6 for foot (with exception for Romans and Pikes) because we found are the best to represent the units we use. With such dimensions, melee of mounted last longer than those of foot, and this is the opposite we want.

In which case give mounted double dice against other mounted.

hammy wrote:

marioslaz wrote: I respect your opinion, but it differs totally from mine. I don't understand your opinion in a point, because Death rolls are not less likely in draw or narrowly won melee, but only in lower dice melee. If you get 4 hits, for example, you have a good chance to make a kill even in a draw, and to get 4 hits is more likely obviously when you roll a lot of dice.
Apart from that, the size of BG we use is 4 for mounted and 6 for foot (with exception for Romans and Pikes) because we found are the best to represent the units we use. With such dimensions, melee of mounted last longer than those of foot, and this is the opposite we want.

In which case give mounted double dice against other mounted.

I reached the same solution! In my simulation seems good, but we need to see how this fit in the overall mechanism.