For fun I found the last "big" game I was involved in that didn't have dice mods.

http://www.wargear.net/games/view/574339

I pulled the log and put it into Excel. I found there were 838 times there was a 3v1 attack. 

Of those the attacker lost 277 times and the defender lost 561 times.

So the attacker won 66.95% of the time. Pure odds say it should be 65.97%. 

Seems legit -

Here's a game that we just completed: http://www.wargear.net/games/view/606270

My buddy "Mattyb" absolutely had the upper hand. He had stashed away 30+ men on several territories, his bonuses were substantial, and my own defenses were severely weakened from several previous rounds against "Epicsmacktalk"

The dice became the biggest factor in the game once we were 1v1, especially for him. I ended up winning, but only because of how consistently terrible my buddy's rolls were.

This happens ALL the time.

dcolum23 wrote:

Here's a game that we just completed: http://www.wargear.net/games/view/606270

My buddy "Mattyb" absolutely had the upper hand. He had stashed away 30+ men on several territories, his bonuses were substantial, and my own defenses were severely weakened from several previous rounds against "Epicsmacktalk"

Yeah, interestingly both you and your buddy were recipients of bad luck through most of the game - looks like your opponents beat up on each other enough to enable the two of you to survive.

The dice became the biggest factor in the game once we were 1v1, especially for him. I ended up winning, but only because of how consistently terrible my buddy's rolls were.

This happens ALL the time.

Yes, it does.. A Luck stat in the range of -30 is not un-common in games where each player is rolling close to 1000 dice.

BorisTheFrugal wrote:

I did a search because I remembered an old forum post.  Tom would have to confirm if that RNG has changed since....

http://www.wargear.net/forum/showthread/167p1/Ive_seen_enoughthe_RandomLuck_Talk

PHP 7 has random_int which is more random than mt_rand but not by much regarding the application in question, according to this article:

https://www.sitepoint.com/randomness-php-feel-lucky/

 

If it happens all the time then it comes around?....defense has the fire sword once and a while. I personally love losing 3 or 4 men several times in a row to singletons. ;)

Forgot that rand didn't use the same function as mt_rand before 7.1 and the article was written before the release of 7.1.

I could test it myself, though. Will provide results later.

Mad Bomber wrote: If it happens all the time then it comes around?....defense has the fire sword once and a while. I personally love losing 3 or 4 men several times in a row to singletons. ;)

love? You certainly can be vocal about it when it happens :)

 

 

random_int(1,6) vs mt_rand(1,6)*

4.8e6 rolls

8e5 expected rolls for each side 

note: array index = side rolled - 1

*used in WarGear

 

Trial 1:

Count

Array ( [0] => 800210 [1] => 801372 [2] => 799132 [3] => 799218 [4] => 799406 [5] => 800662 )
vs
Array ( [0] => 800176 [1] => 798878 [2] => 799797 [3] => 800732 [4] => 800740 [5] => 799677 )

Deviation from expected rolls in percentage

D1: 0.02625% vs 0.022%
D2: 0.1715% vs 0.14025%
D3: 0.1085% vs 0.025375%
D4: 0.09775% vs 0.0915%
D5: 0.07425% vs 0.0925%
D6: 0.08275% vs 0.040375%

Error Sum: 0.561% vs 0.412%

Max Streaks (max number of times the same side was rolled in succession)

Array ( [0] => 8 [1] => 7 [2] => 9 [3] => 7 [4] => 9 [5] => 8 )

vs

Array ( [0] => 7 [1] => 7 [2] => 7 [3] => 7 [4] => 8 [5] => 6 )

 

Trial 2:

Count

Array ( [0] => 799579 [1] => 800103 [2] => 799885 [3] => 799697 [4] => 799497 [5] => 801239 )
vs
Array ( [0] => 800176 [1] => 798878 [2] => 799797 [3] => 800732 [4] => 800740 [5] => 799677 )

Deviation from expected rolls in percentage

D1: 0.052625% vs 0.022%
D2: 0.012875% vs 0.14025%
D3: 0.014375% vs 0.025375%
D4: 0.037875% vs 0.0915%
D5: 0.062875% vs 0.0925%
D6: 0.154875% vs 0.040375%

Error Sum: 0.3355% vs 0.412%

Max Streaks (max number of times the same side was rolled in succession)

Array ( [0] => 9 [1] => 7 [2] => 7 [3] => 7 [4] => 7 [5] => 7 )

vs

Array ( [0] => 7 [1] => 7 [2] => 7 [3] => 7 [4] => 8 [5] => 6 )

 

Trial 3:

Count

Array ( [0] => 801146 [1] => 798364 [2] => 799252 [3] => 799382 [4] => 801128 [5] => 800728 )
vs
Array ( [0] => 800176 [1] => 798878 [2] => 799797 [3] => 800732 [4] => 800740 [5] => 799677 )

Deviation from expected rolls in percentage

D1: 0.14325% vs 0.022%
D2: 0.2045% vs 0.14025%
D3: 0.0935% vs 0.025375%
D4: 0.07725% vs 0.0915%
D5: 0.141% vs 0.0925%
D6: 0.091% vs 0.040375%

Error Sum: 0.7505% vs 0.412%

Max Streaks (max number of times the same side was rolled in succession)

Array ( [0] => 9 [1] => 7 [2] => 8 [3] => 7 [4] => 8 [5] => 7 )

vs

Array ( [0] => 7 [1] => 7 [2] => 7 [3] => 7 [4] => 8 [5] => 6 )

 

Conclusion:

Pretty similar results, mt_rand did better but I haven't done enough trials to say that it's the most random one. On of top of that, mt_rand is almost twice as fast but random_int uses a cryptographically secure generator while the former does not (we're talking about few microseconds per roll anyway).

Does anyone else ever feel that on rare occasions, the dice calculator gets stuck and the defender churns off 6 to 8 max numbers in a row?  That's the paranoid element that remains for me.

Thingol wrote:

Does anyone else ever feel that on rare occasions, the dice calculator gets stuck and the defender churns off 6 to 8 max numbers in a row?  That's the paranoid element that remains for me.

that could be where all these 'bad luck' things come from. one of the reasons i went to rarely using T/A and instead roll most of mine individually. may not have been a 'thing' but makes me feel better.

weathertop wrote:

Thingol wrote:

Does anyone else ever feel that on rare occasions, the dice calculator gets stuck and the defender churns off 6 to 8 max numbers in a row?  That's the paranoid element that remains for me.

that could be where all these 'bad luck' things come from. one of the reasons i went to rarely using T/A and instead roll most of mine individually. may not have been a 'thing' but makes me feel better.

It does seem that T/A are more likely to get huge lucky or unlucky streaks. Rolling manually, when I'm on a losing streak I try to break out of it by doing longer pauses between rolls, or just mix the intervals.

redshift wrote:

Conclusion:

Pretty similar results, mt_rand did better but I haven't done enough trials to say that it's the most random one. On of top of that, mt_rand is almost twice as fast but random_int uses a cryptographically secure generator while the former does not (we're talking about few microseconds per roll anyway).

Cool --

I was reading about confirmation bias and came across this - 

Zeigarnik effect: uncompleted or interrupted tasks are remembered better than completed ones.

Which goes along with us remembering when attacks fail more than when they succeed. Usually we are attacking as part of a plan - and then some bad dice interrupts that plan - and we remember it

 

Also - it just makes sense that we see a lot of times when something that should have succeeded fails but rarely see something that should have failed succeed.  A 5v1 has a 97% chance of success while a 4v9 has a 97% chance of failure. However I have done 100's (1000's?) of 5v1's and probably have never done a 4v9.

 

 

redshift wrote:

It does seem that T/A are more likely to get huge lucky or unlucky streaks. Rolling manually, when I'm on a losing streak I try to break out of it by doing longer pauses between rolls, or just mix the intervals.

I do this too

redshift wrote:

It does seem that T/A are more likely to get huge lucky or unlucky streaks. Rolling manually, when I'm on a losing streak I try to break out of it by doing longer pauses between rolls, or just mix the intervals.

Well then, if you're a streak conspiracy theory proponent ;) , you should consider the corollary tactic:  When your situation demands a streak, e.g, 9v21 (with a 97% chance of failure), you're better off hitting T or A

Amidon37 wrote:

I was reading about confirmation bias and came across this - 

Zeigarnik effect: uncompleted or interrupted tasks are remembered better than completed ones.

Which goes along with us remembering when attacks fail more than when they succeed. Usually we are attacking as part of a plan - and then some bad dice interrupts that plan - and we remember it

 

Also - it just makes sense that we see a lot of times when something that should have succeeded fails but rarely see something that should have failed succeed.  A 5v1 has a 97% chance of success while a 4v9 has a 97% chance of failure. However I have done 100's (1000's?) of 5v1's and probably have never done a 4v9.

 

 

I did a 4 vs 9 in a team game once, with the aim of weakening the 9. I managed to conquer it for the loss on 0 armies :) Just sayin'. I totally agree with your post. 

I just keep another window open showing the current milliseconds on the server clock so i can time the dice rolls right.

Korrun wrote:

I just keep another window open showing the current milliseconds on the server clock so i can time the dice rolls right.

I use cheat codes that can be created by a board's designer, which explains why I pretty much only play my own boards.

Litotes wrote:

I did a 4 vs 9 in a team game once, with the aim of weakening the 9. I managed to conquer it for the loss on 0 armies :) Just sayin'. I totally agree with your post. 

Right - I have done things like this where my main stack is on one side but I have a small stack I want to use to weaken before bringing my big one to finish - but then i wouldn't go all the way with the small stack - 

M57 wrote:

redshift wrote:

It does seem that T/A are more likely to get huge lucky or unlucky streaks. Rolling manually, when I'm on a losing streak I try to break out of it by doing longer pauses between rolls, or just mix the intervals.

Well then, if you're a streak conspiracy theory proponent ;) , you should consider the corollary tactic:  When your situation demands a streak, e.g, 9v21 (with a 97% chance of failure), you're better off hitting T or A

Well pointed.

weathertop wrote:

Thingol wrote:

Does anyone else ever feel that on rare occasions, the dice calculator gets stuck and the defender churns off 6 to 8 max numbers in a row?  That's the paranoid element that remains for me.

that could be where all these 'bad luck' things come from. one of the reasons i went to rarely using T/A and instead roll most of mine individually. may not have been a 'thing' but makes me feel better.

I should mention that I almost always use A. It should be just as random as rolling with 1 dice. The only time I attack with 1 at a time is if I am suspicious of the game results already. I just rolled 11 v 1 and lost all of my men using A. I then rolled 13 v 1 and lost all of my men. That is mathematically impossible, guys. And it is extremely infuriating. I swear odds are determined at the start of the game, not truly random each roll. Certain players tend to dominate all game regardless of rolls.