#Version 3.20

1 messages · Page 1 of 1 (latest)

potent geyser
#

If you haven't played the latest version and haven't read the news releases then there are some major changes you need to pay attention to:

  1. You don't need to slider anymore
  2. Cap rolls on larger caps (50+ defenders) will always be punitive. No more positive cap rolls on large caps. Plan accordingly.
  3. bot behavior can be fully discerned from the avatars on the main screen. no need to click on the player to see if they are offline.
    a) MIA means that the player is not connected at the moment. This could be a mobile player that has switched temporarily to another app. It doesn't mean they are a bot or that they have quit. They are simply away and may come back at any time.
    b) if a bot icon appears below the avatar then the ai has taken over for the player. However they can still come back for 2 full turns.
    c) if a white flag shows up then the player had either surrendered or they have exceeded the time they can come back. They will permanently be a bot.
    d) if the bot has zzz next to it then it is s neutral bot and will not attack. Without the zzz it is an active bot and will take is turn immediately, even on the first turn of becoming a bot.
#

I put all this out here because I am absolutely loving these changes. I hope everyone else is too

south kernel
#

Does the bot still count for placement?

potent geyser
#

Yes. For active bots the bot placement is the players placement. For neutral bots their placement is locked in once they have perma-botted.
Bot placement may be updated in v3.21

glad night
#

Slider should still be beneficial, only less so. There were 2 reasons to slider:

  1. Slidering large stacks, to limit losses or even prevent the wonky approximation algorithm to replace the exact calculations.
  2. (Perfect) slider to lose slightly less troops on average.

Reason 1 was most obvious and that is fixed now.
Reason 2 still exists (if the implementation is as announced, didn't actually do tests yet). But the effect of reason 2 is fairly limited and only really makes a difference at all if you slider perfectly to the min 100% roll.

#

Would be interested in screenshots of rolls over 2,000 troops (showing it's balanced blitz), since those were still pretty wonky (in a way they shouldn't be) in the plots in the announcement.

potent geyser
#

#2 really isn't needed until total troops (attacker+defender)) totals more than 2,000. Of course you will also get ??? gor the percentages. At 2,000 troops the game enters an estimation phase. At 50,000 it reverts to true random.

#

I would suggest that there are only 2 reasons to use the slider today
1l) to partially attack a stack. You and an ally each put 50 troops into an opponent stack.
2) 0% rolling a stack to maintain a block

It will take a while to get used to not slidering but I think you'll find you don't need to in most other situations.

glad night
potent geyser
#

Using your example a 10v1 in prior version you could lose more than 2 troops, whereas a 4v1 could not.
That is not true with v3 20. You will never lose more than 2 troops regardless of the attacking size.
You really don't need to slider.

pine jasper
#

Good stuff here. I've been trying to educate people on the changes as well.

glad night
#

To repeat the bottom line, using the slider perfectly to go to the min 100% roll (in general, there are a few rare exceptions) will save you some troops. If you don't do it perfectly, it's a lot less effective.

That was true before and still is. Cause reason 2, is completely unrelated to reason 1.

potent geyser
#

During the closed beta, @thestaticbrock complied a lot of data on the the new rolls. Here is a summary of what he found when rolling capitals.
You'll notice that you can occasionally lose more than what a perfect slider would allow but not by much and not often.

image0.png
#

I have "slider failed" way too often, as most people who side regularly do.
Not slidering will save much more troops than trying to preserve troops with the slider.
That's why I'm definitively advocating to not use the slider anymore.
The edge cases are not worth it.
And it's a very hard habit to breaklol

glad night
glad night
#

Ran some calculations. Highlighted the results that don't match the data gathered. So we see some deviation at the larger rolls already. And I assume the 100% roll for a 10 cap is an user error (didn't confirm in-game)?

To be clear, I consider troops in battle (while your screenshot does troops on territory). And the for the min troops lost I considered the 100% roll, since those can/will be slightly higher for 2,000 troops, which makes the comparison harder.

new_BB_data_comp.png
#

The averages in my calculation should also be on the higher end, since those are based on the 2,000 troops attacking (and would be lower for a perfect 100% roll).

#

The only thing that I can think of that could explain it, is if the new 2,000 approximation limit is in total (while the previous 75 was per side). Maybe if in testing they attacked the 500 cap with over 1,500 troops (but under 2,000), that could explain the low result without inconsistencies.

potent geyser
#

The 2000 limit is total attackers + defenders. That's when it switches to an estimation model. All bets are off at that point

potent geyser
# glad night Nice, more data is always good. So if I understand correctly, they always rolled...

That logic did change. For example none of the dice calculators match the new balanced blitz algorithm.
It's a complete change.
I agree with collecting more data to see where things break down as I'm sure there are more edge cases than we currently know.
If you were doing 10v1 or 20v1 or 50v1 in the past and not using the slider you would be losing a lot more troops than you'd suspect.
You can now hit 1s all day long with 500 stacks and no slider and you will not see those bad rolls.

glad night
#

Can you give concrete examples of that, or in what way it has changed? Cause I don't see any difference in the published documentation so far.

#

If you were doing 10v1 or 20v1 or 50v1 in the past and not using the slider you would be losing a lot more troops than you'd suspect.

This is just false, I don't know why you keep repeating it.

#

May I also ask how do you know: "The 2000 limit is total attackers + defenders"? Cause that was not how it was calculated when the limit was 70, so it's a bit confusing if they did communicate it went from 70 to 2,000. And those numbers mean something completely different.

eager smelt
#

Yeah, I assumed it was on one side

potent geyser
#

if you absolutely want to mix/max, you can continue with the slider, but I really believe it is not necessary on normal rolls. Time may prove me wrong, but the best advice at this time is to not use the slider for normal attacks.

potent geyser
potent geyser
#

here's an example of a 39v1 losing 3. one point of interest is that it was a capital, so maybe that somehow plays into. most of the recorded rolls i have losing > 2 on a 1 are against capitals with 1 troop.

image.png
glad night
glad night
# potent geyser through exprimentation. the ??? show up when attackers+defenders > 2000

In the past the win%/slider calculation seemed to be done before the actual battle outcome calculation. For example, we saw very wonky troops remaining, while the actual winning was still normal (you don't expect that if the troops are determined first and the win% is derived from that). So idk if the ??? is a good measure for what's happening with troop calculations, but maybe it is for win% itself.

potent geyser
#

That is true. It is 2 steps i believe. Calculate the win%. Calculate troops lost.
Until we get official confirmation one way or the other I'm working on the assumption that if the win% is switching to estimation that the dice calculation is also.

glad night
#

Good to know at least what is confirmed info and how that is established 🙂 I think it makes sense anyway to do the slider calculation seperately. Since you need to be able to calculate more values, even while the player might never take that actual battle.

potent geyser
#

The wonkiest rolls were with > 2000 and < 50k. 50k reverts to true random is my understanding.
A request was made to get rid of estimation all together and switch to true random at 2k. I support that change but have no idea of it will happen

glad night
#

I don't even know what they do differently for TR (guess just simulation??), because doing the BB logic takes no computational effort, all the slow bits are in the TR part of the calculation.

#

Also posted before, but from the announcement the plots for large rolls seemed off, not just the estimations, but also the calculations #risk_strategy message

potent geyser
#

True random does the actual rolls. I'm not aware of any issues with that logic.
I know how the original BB win % was calculated but I'm not sure how the troops lost was, but somehow it doesn't scale well.

eager smelt
potent geyser
#

Did you file a bug report?

pine jasper
eager smelt
potent geyser
eager smelt
#

Done

potent geyser
#

Also, there seems to be a bug where a player turns to a white flag (permabot) but it's neutral bot keeps adding troops. If that happens the bot will receive placement like an active player.
This should be fixed in a patch, but just be aware if you care about rank points