Triple A vs Frood

[Jennifer]

Okay, so here’s the situation:

W. Russia:

Russia attacks with 5 Infantry, Artillery, 2 Fighters
Germany defends with 3 Infantry, Artillery, Armor

Now, I’ve run this 12 times with Triple A and Russia gets creamed everytime.  By creamed, I mean reduced to nothing but fighters and sometimes with only one fighter out of the two left.

Frood, on the other hand, has the following:

Attacker: 5 Inf, 1 Art, 2 Fig. v. Defender: 3 Inf, 1 Art, 1 Arm.

Average battle duration: 2.8 rounds of combat
  avg. # units left IPC value Punch
Attacker: 4.53.5 27.311.7 9.84.2
Defender: 0.14.9 0.317.7 0.210.8

Surviving Attackers

Surviving Defenders #Casualties

Overall %*: A. survives: 94.9% D. survives: 4.1% No one survives: 1%

• percentages may not total 100 due to rounding. The average results from above are highlighted in charts below, while the median result (equal odds of getting a worse or better result) is written in red.
  Attacker results:
  Probability % # units / losses
    2.11% 8: 5 Inf, 1 Art, 2 Fig. no units. : 0 IPCs
    9.91% 7: 4 Inf, 1 Art, 2 Fig. 1 Inf. : 3 IPCs
    20.16% 6: 3 Inf, 1 Art, 2 Fig. 2 Inf. : 6 IPCs
    22.96% 5: 2 Inf, 1 Art, 2 Fig. 3 Inf. : 9 IPCs
    18.45% 4: 1 Inf, 1 Art, 2 Fig. 4 Inf. : 12 IPCs
    11.72% 3: 1 Art, 2 Fig. 5 Inf. : 15 IPCs
    6.5% 2: 2 Fig. 5 Inf, 1 Art. : 19 IPCs
    3.05% 1: 1 Fig. 5 Inf, 1 Art, 1 Fig. : 29 IPCs
    5.14% 0: no units. 5 Inf, 1 Art, 2 Fig. : 39 IPCs
  Defender results:
  Probability % # units / losses
    0.01% 5: 3 Inf, 1 Art, 1 Arm. no units. : 0 IPCs
    0.23% 4: 2 Inf, 1 Art, 1 Arm. 1 Inf. : 3 IPCs
    0.7% 3: 1 Inf, 1 Art, 1 Arm. 2 Inf. : 6 IPCs
    1.39% 2: 1 Art, 1 Arm. 3 Inf. : 9 IPCs
    1.79% 1: 1 Arm. 3 Inf, 1 Art. : 13 IPCs
    95.88% 0: no units. 3 Inf, 1 Art, 1 Arm. : 18 IPCs

So which is it, is Dan’s math way off, or is TripleA completely insane with it’s dice!?!?

[U-505]

The battle sim I ran supports Frood for the most part. After 10000 rolls the only difference is that it says the Russians are more likely to end up with 1 inf, 1 art, 2 fig as opposed to 2 inf, 1 art, 2 fig. The percentage difference is less than a 2% shift though so it’s not significant.

[Jennifer]

Think I’m more bytchin about how lousy the dice are in TripleA then anything else.  Just frustrating when you KNOW what the result should be, give or take, and cannot duplicate it.

I did manage to win W. Russia on the 23rd attempt.  22 Losses of everything but fighters followed by a Russian win without loss.  Honestly, how can you play test anything with TripleA!?!?!?

[Weekend Gamer]

The other weird thing with Triple A is that once or twice a game, someone will roll a big battle and take the territory WITHOUT LOSS and against large odds, but then the dice roll emails don’t go out.  I’ve had that happen 3 or 4 times now.  Anyone else notice that?

[frood]

I’ve had a number of attacks in TripleA lately where I attack with 10 or so units and don’t get more than a single hit in a round of combat. Maybe there’s something wrong with the randomizer?

However, any fault in tripleA’s dice would not necessarily apply when doing PBEM, because the dice are rolled by a dice server rather than TripleA itself.

Based on my limited experience I can’t say it’s anything more than bad luck, but I’ll have to try this battle in TripleA myself. Just looking at it though it looks like it SHOULD be a win for Russia.

[frood]

I just ran that battle in TripleA’s Battle Simulator and it has the attacker win 95% of the time, just like my results. That’s based on 5000 battles.

Now, I just want to make one observation: 12 attempts is not a lot, certainly not enough to base any conclusions on. Even with running 10,000 times, my sim usually shows a range of at least +/- 1% in the calculated percentages.

10,000 is a lot more than 12. But if people are comfortable with predicting future probabilities based on 12 runs, I could make my sim run a lot faster by reducing it to 12 iterations…

[Guest]

Don;t be a smart ass Dan :-P

But if you want to add a “Jen” setting for rolling a battle 12 times, go for it :-D

I can see the radio buttons now…

1,000
5,000
10,000
Jen

:-D

[newpaintbrush]

@Gamer:

The other weird thing with Triple A is that once or twice a game, someone will roll a big battle and take the territory WITHOUT LOSS and against large odds, but then the dice roll emails don’t go out.  I’ve had that happen 3 or 4 times now.  Anyone else notice that?

Ya it happen to me too.

[newpaintbrush]

Fact is, when you have a single outcome with 8% probability that is the single most likely outcome, you still get something else 92% of the time.

@Frood:

Now, I just want to make one observation: 12 attempts is not a lot, certainly not enough to base any conclusions on. Even with running 10,000 times, my sim usually shows a range of at least +/- 1% in the calculated percentages.

(nods head)

. . . I wasn’t sleeping!

[frood]

@ShadowHAwk:

Well it aint a simulator it just rolls the dice.
If you roll the dice 100 times you can have a strange result on the average every time
Using Low luck to calculate the result manualy ( without a calculator ) i get to 1 inf 1 art 2 figs surviving.
Then again i saw 6 inf hit 3 times and 20 inf hit nothing in real games so it really does not say a lot.
BTW isnt tripleA open source anyway ? so if you think it is flawed just read the sourcecode :D

I call it a simulator because it simulates rolling actual dice, as opposed to a calculator which would work out the exact odds mathematically according to the probabilities.

[Jennifer]

@Frood:

10,000 is a lot more than 12. But if people are comfortable with predicting future probabilities based on 12 runs, I could make my sim run a lot faster by reducing it to 12 iterations…

I don’t use it to predict, I just kept doing it trying to get TripleA to give me a realistic response so I could play with alternatives and realized, after 12 attempts, that the dicey in TripleA is FUBAR.

Anyway, didn’t know that TripleA will farm the dice out to a different server if you tell it too.  That might redeem the program for me.

Now, if they would redesign MapView so it’s as flexible with stacking units as Battlemap, we’d have a serious upgrade for our games!

[tekkyy]

Thats sound pretty bad.
Have you notified TripleA of your results?

TripleA has a history dice statistic thing. What does it say on your computer?

There is a calculator in TripleA too. Its set to do more than 12 runs by default.