Random Weather GODS are Germanic!!!!!!

[HardLuckYetAgain]

Nada

[56ajax]

Random Weather is pro AXIS probably for the reason that historically the weather that year, 1941, was excessively harsh ie abnormal, so when you go for random it should be milder...

[Crackaces]

ORIGINAL: johntoml56

Random Weather is pro AXIS probably for the reason that historically the weather that year, 1941, was excessively harsh ie abnormal, so when you go for random it should be milder...

Are you suggesting central limit theorem ? That is there really is no mud season, and 1941 was an exception?
The harshness was not so much a change in the mud season but some of the coldest weather in recorded history. From a cut off high over Europe and Artic air blasts ..the question is how much variation over time on the arrival of the mud season ...

[Nix77]

ORIGINAL: HardLuckYetAgain

I know it is supposed to be random but I like how my last 3 random weather games as Soviets I get no mud at all up to turn 17. Clear weather as far as the eye can see turns 1-17. Then BAM!!!! turn 18 there is snow, not one, not two, but my last 3 games with random weather had snow in one of the zones. So I will just chalk that one up as a coincidence. Turn 19 roles around in these games and yet again snow in 3 of the zones. So I am 100% convinced that the Germans have a weather control machine of some sorts. Turn 20 I had mud in all the zones in 2 of the games and yet to be seen in the current one. Darn Germans ;-P

Here's a small chart I've made (interpreted from the manual, not sure if it's 100% accurate), you can roll your 11-sided die for the Gods of Fortune and Weather!

In my game as Soviets, I also had 17 turns straight clear weather. If you actually calculate the chances for that, it's something under 1%... I think there might be some other weightings for clear weather during summer turns than only a pure Die(11).





I'm also interested in how the game calculates the "75% reliability" on random weather? Is there a 12,5% chance to jump up and 12,5% to jump down on the chart, or how does the forecast uncertainty work? Basically that would mean that on a turn where you have a 1/11 chance for mud but the forecast was clear, there would be two possible "Clear" forecasts from where you could jump up or down to hit the mud. Chance would be 1/11*1/8+1/11*1/8=1/44 to hit the mud with a clear forecast. When you have more chances for different kind of weather, it gets more complicated (if it's calculated like this)

[morvael]

This is current random weather table:

[morvael]

Non-random weather, where alternating mud and clear turns in May-early June are expressed as 50%/50% chance.

[morvael]

Old random weather table:

[morvael]

Non-Random Weather GODS are even more Germanic, from June 20 to October 9 all clear weather...

[Nix77]

Seems like my table is on the right track, just made a mistake with one snow chance in February

Morvael, do you have any clue how the random weather forecast 75% certainty calculation works? Was I on the right track with my previous calculation (12,5% to jump up on the chart, 12,5% to jump down)?

Sometimes it's crucially important to know how much you can trust the forecast. Clear forecast turning to Mud in July is for sure more rare than Clear turning to Mud in October?

[morvael]

When a weather is randomly determined, there is a 75% chance that forecast will show that weather. If not, a separate "forecast" weather will be determined randomly again, using the same rules as for the "true" weather. It can actually show the same weather, even if the forecast was not accurate, which is common especially in the months when one weather type is dominant. Therefore the chance for the weather forecast to lie is pretty small.

example: 95% chance for clear, 5% chance for mud
weather clear, forecast accurate: 0.95*0.75 = 0.7125
weather clear, forecast not accurate, forecast clear: 0.95*0.25*0.95 = 0.225625
weather clear, forecast not accurate, forecast mud: 0.95*0.25*0.05 = 0.011875
weather mud, forecast accurate: 0.05*0.75 = 0.0375
weather mud, forecast not accurate, forecast clear: 0.05*0.25*0.95 = 0.011875
weather mud, forecast not accurate, forecast mud: 0.05*0.25*0.05 = 0.000625

check to see if all combinations are covered:
0.7125 + 0.225625 + 0.011875 + 0.0375 + 0.011875 + 0.000625 = 1, which is correct

Chance of clear forecast turning to mud is the fifth option with 1.1875% chance.

Only option 3 and 5 are where forecast weather is different from true weather, which means forecast has actually 97,625% chance to be "accurate" (defined as forecast weather the same as true weather) in this case. It will be less reliable in periods like May - early June.

[56ajax]

Clear as mud.[:)]

[Nix77]

ORIGINAL: johntoml56

Clear as mud.[:)]

Indeed!

Thanks Morvael, now it's definitely all clear as mud to me, except South Soviet that has 2.27% chance of being clear

[HardLuckYetAgain]

Nada

[Nix77]

ORIGINAL: HardLuckYetAgain

I think Mike29 had a similar situation but not exactly like mine. I'm not complaining at all since I have had games where there were no problems at all. I just thought it was weird is all. Not exactly something you see everyday but interesting all the same

As always thank you for all the very detailed information received everyone

The chance to get all clear for 17 turns straight in North, Central and South Soviet is..... 8,63%! So it's like once every twelve games or so, not that rare! I'm not counting Europe zone here since it's not consequential in '41.


Explanation on the calculation:

June is guaranteed clear, so we're talking about 15 random turns, out of which over three zones 39 are 10/11 clear and six turns are 8/11 clear (September turns in North Soviet).

10/11:
Clear/fc correct = 10/11*3/4 = 330/451
Clear/fc wrong => clear = 10/11*1/4*10/11= 100/451
Mud/fc wrong => clear = 1/11*1/4*10/11 = 10/451
= 440/451

8/11:
Clear/fc correct = 8/11*3/4 = 264/451
Clear/fc wrong => clear = 8/11*1/4*8/11= 64/451
Mud/fc wrong => clear = 3/11*1/4*8/11 = 24/451
= 352/451

(440/451)^39 + (352/451)^6 = 0,0863!

[morvael]

Your calculations are for old weather model (/11 not /20). And I think it's not important to include whether forecast is accurate or not.

[morvael]

In 1941:
June is 3 turns
July-August is 9 turns
September-October 10th is 6 turns

In three zones, except North Soviet Zone, there is 95% chance for clear weather on every single turn out of 15, which gives 46.33% chance for all 15 turns of clear weather (in each zone).
In North Soviet Zone this is reduced to 33.49%, because on the last 6 turns the chance for mud is 10%.

edit: this means that the chance to have 15 turns of clear weather in all the important zones (except Europe) is 7.19%.
This is all still much worse than non-random weather with 100% chance for 15 turns of clear weather.

[morvael]

Unfortunately, I think non-random weather is the way to go in WitE. The four crude types of weather applied to huge areas at once represent only the major weather condition changes, and should not be used to simulate less significant weather effects. Luckily WitW/WitE2 has new, much better, dynamic weather model.

[HardLuckYetAgain]

Nada

[Nix77]

ORIGINAL: morvael

Unfortunately, I think non-random weather is the way to go in WitE. The four crude types of weather applied to huge areas at once represent only the major weather condition changes, and should not be used to simulate less significant weather effects. Luckily WitW/WitE2 has new, much better, dynamic weather model.

Oh I completely missed the new model was RND(20) based! Well, that explains the greater chance for the dominant weather type to appear.

As for WitW/WitE2, I can imagine myself staring at the battlefield twice longer than I now am, just because of the new realistic weather model

[HardLuckYetAgain]

Nada