Victor the Vicous has now been fully updated,
with troops needed and lost for eight different setups, including all kinds of battle hardened generals and also cannoneers.
It very much looks like a better setup for those needing the 220 seconds...Originally Posted by XiozZe
Although 2200 simulations are really too few to tell exactly how much better your setup is...
Let's compare your graph and mine:
Looking at those graphs it's strange to see how 10-round losses are so much more common than 9- and 11-round losses.
And likewise how 11-round victories are so much more common than 10-round victories. Something I cannot explain...
In reality these distributions should be more like bell curves (Gaussian normal distribution),
but I cannot say if 2,200 is simply too small a sample to show the real shape of the distribution.
EDIT: The more I'm looking into these (been simulating quite a number of different setups tonight to see if I could find something better than yours, but I couldn't), the more obvious it becomes that there must be something tipping over the outcome of the fight in the first round. Probably just the difference of whether one more or one fewer of some unit dies or survives. The destiny for this one unit is probably enough to lead to either a 9-round victory or to a 12-round defeat (on average).
The best tool I've got for these kinds of comparisons is Andelar's DOS-based simulator.
It always makes 10,000 simulations, and by running it for the same setup tens of times you get a pretty good distribution.
Unfortunately it doesn't give the whole distribution though, it only gives the minimum duration and the percentual chance for it and the percentual chance that it will be one round longer.
By running your setup of 1R 7M 190C 150,000 times, I get these chances for 160 and 180 seconds of fighting (but no estimate of 200 seconds, as there are always a number of rounds lasting only 160 seconds:
160 seconds: 131 times out of 150,000 = 0.09% (varying between 0.06% and 0.14%)
180 seconds: 486 times out of 150,000 = 0.32% (varying between 0.15% and 0.40%)
By running my own setup of 1R 6M 1S 190C 150,000 times, I get these chances for 160, 180 and 200 seconds of fighting (since there are 7 cases where the chance for 160 seconds is 0.00%, I do get a pretty good estimate for the chance that it will last 200 seconds):
160 seconds: 17 times out of 150,000 = 0.01% (varying between 0.00% and 0.04%)
180 seconds: 124 times out of 150,000 = 0.08% (varying between 0.04% and 0.15%)
200 seconds: 5936 times out of 70,000 = 8.48% (varying between 8.14% and 8.97%)
However, Andelar's simulator also gives intermediate results after 1,000 simulations, and by checking 15 of these rounds with no occurrance of 160 seconds, I can get a distribution for your setup for 200 seconds:
200 seconds: 793 times out of 15,000 = 5.3% (varying between 4.1% and 6.3%)
So yes, your setup is clearly better if you want a block that lasts 220 seconds or longer.
The chance that it will last less than 220 seconds is about 5.7%, compared to 8.6% for my setup.
Reply With Quote
