Countdown to Battle

[SimonC]

As we look at a new system for countdown to battle, I would like to get a baseline of what we currently have. So I created some test scripts and ran them 10K times to see what the outcome was.

There is some constant data, that being the event list was 1915 German (for no other reason it was the open page).  There are 2 modes to run in - that being 'safe mode' where you always rollover, and a more risky always roll method.

In 'safe mode' (using the 3,3,4,3,3 pattern) across 10,000 games you get



in 'always roll/risk mode' (using the 3,3,4,3,3 pattern) across 10,000 games you get



There is nothing too outlandish here, so you can see that the number of events doubles - but on 2-12 range opposed to a 8-18

If you think you have some clever strategies (for dice allocation) let me know and I can run them through the simulator

[SimonC]

As an example of trying to influence the outcome. In our example Army 1 switches to a 1,6,1,6,2 pattern and always roll over
Army 2 attacker breakdown (they stick on a 3,3,4,3,3 - always roll)

[SimonC]

If we keep the patterns the same and swap the rollover, (both never rollover)





actually illustrates the rollover has far more of an effect that the dice allocation. But it does show that you can swing the result by 60% at the cost of a 'lower level' battle (if you do attack)

[Colonel Kilgore]

Simon, I think your model assumes rational (or at least consistent) choices from both parties.

Can you readily Monte Carlo (?) completely random choices (e.g. either both players being random, or one being random and the other consistent according to one of these patterns) too?

Simon

[SimonC]

Random in what way ?  choosing to roll over , or random dice placement. Making it random will just add noise, and doesn't really add anything to the analysis. This way to just illustrates what influence you have. Random just adds 'random' influence.

[SimonC]

I think all we really to know from this is that the attack type is a bell curve. The attacker gets between 4-6 event and the defender gets 2-4

holding the rollovers gives a better change to attack.

[SimonC]

here is data based on random rollover selection.



doesn't make huge difference to the outcome

[martin goddard]

Marvellous work Simon.
Thank you for doing it.
Does it lead you to the conclusion that it is better to "carry over"?


martin

[SimonC]

Does it lead you to the conclusion that it is better to "carry over"?

If you want to attack - for sure. It influences the results more than the allocation of dice.

[SimonC]

I'll script the new method - maybe- tomorrow, then we can compare and contrast

[martin goddard]

Good research work Simon.

thanks

martin

[Colonel Kilgore]

Random in what way ?  choosing to roll over , or random dice placement. Making it random will just add noise, and doesn't really add anything to the analysis. This way to just illustrates what influence you have. Random just adds 'random' influence.

Thanks for this, Simon. I was actually thinking random dice placement. Maybe even combined with random rollover too. The interest for me would be to understand whether there were a clearly advantageous choice regardless of what my opponent (and wargamers can be an irrational lot) did.

Simon

[Sean Clark]

This makes my brain hurt but is interesting to read. 2 questions.

The 3343 method you talk of is only 4 allocations as opposed to 16162. I might be being dumb here, but can you explain it to me?

Do you have a hypothesis and what have the results shown you so far?

Also, if you wish to attack, carry over helps more than allocation. But you have to win the day to ensure you get the option to carry over, so are they not intrinsically linked?

I  tend to go with a chaos theory method when I play, but this deep dive into statistical analysis is fascinating.

[SimonC]

The 3343 method you talk of is only 4 allocations as opposed to 16162. I might be being dumb here, but can you explain it to me?

thats my lazy shorthand. it is 33433 , the simulator does the full 3 week , 5 day countdown. I've corrected the prior posts

The purpose of this was really to illustrate what we have got now - rather than come to any conclusions per se. If there is a new method it needs to be broadly in keeping with the existing system. 

However, I think you have hit the nail on the head with he chaos theory, and you should trust your brain. What it tells me is that this is a complex systems with a lot of external variables. It is very hard to influence the outcome in a desired way. The only real outcome is that both sides fiddling with the variables is the type of game gets 'squashed' in the middle. That said that looking at averages over 10 000 iterations. Playing a game you do it only once ... so it may get swingy - or it may not 

IMHO this is the main problem with this mechanism

[Colonel Kilgore]

I quite like the notion that it's hard to influence the outcome in a meaningful way - it keeps everyone on their toes and you never know what you're going to get

Simon