Yikes!! Interesting, the map I won is when luck was on my side on the last link. And if we look into the other games... Whoever had the LUCK side to them are most of the time the winners of the board.
Yeah luck isn't everything obviously...that's why we have people at the top stay there. They have experience and know how to play better than others. But sometimes...even they can't beat luck :P
I look at it this way if there are 5 people standing in the street and 15 people attack them how often do you think the 5 people would win if everyone is evenly trained once in a blue moon, so how can the 5 people win on 3 or 4 different occasions during the same game. Maybe there should be some type of limit to how bad or good your luck can get. Cheers!
Angusjustice wrote:I look at it this way if there are 5 people standing in the street and 15 people attack them how often do you think the 5 people would win if everyone is evenly trained once in a blue moon, so how can the 5 people win on 3 or 4 different occasions during the same game. Maybe there should be some type of limit to how bad or good your luck can get. Cheers!
A, What you are asking for is to alter the rules of the game. The dice are the dice. They are subject to the laws of probability, which means 5 armies will defeat 15 armies about 0.3% of the time, and 5 armies will withstand an attack from 15 armies about 1.5% of the time.
What would you propose? That if 15 armies attack 5 armies and it gets down to 1 army attacking 1 defending army, then that attacking army should automatically win, regardless of the dice thrown?
M57 wrote:Angusjustice wrote:I look at it this way if there are 5 people standing in the street and 15 people attack them how often do you think the 5 people would win if everyone is evenly trained once in a blue moon, so how can the 5 people win on 3 or 4 different occasions during the same game. Maybe there should be some type of limit to how bad or good your luck can get. Cheers!
A, What you are asking for is to alter the rules of the game. The dice are the dice. They are subject to the laws of probability, which means 5 armies will defeat 15 armies about 0.3% of the time, and 5 armies will withstand an attack from 15 armies about 1.5% of the time.
What would you propose? That if 15 armies attack 5 armies and it gets down to 1 army attacking 1 defending army, then that attacking army should automatically win, regardless of the dice thrown?
I agree. Although it's true that you can lose a game because of bad rolls, dice are dice and are even, and if there are strong players, it's not just because of the luck. Other things matter too in this game!
=)
After creating the luck stat and thinking about some of the implications of the analysis, Mongrel and I came to the conclusion that:
1) arbitrarily long streaks of good luck and bad luck will occur.
2) some players will have bad luck and some players will have good luck (across all their games).
These have nothing to do with the random generator of the site or computers in general, but the vary nature of the dice.
In remembering the days of sit down games with live people and real dice, I recall losing massive attacks/defenses and also beating the odds to win the game.
Alpha wrote: vary nature
what a wit!
Alpha wrote:2) some players will have bad luck and some players will have good luck (across all their games).
It would be cool to find such a player. My thinking is that they are a rare breed.
Amidon37 wrote:Alpha wrote:2) some players will have bad luck and some players will have good luck (across all their games).
It would be cool to find such a player. My thinking is that they are a rare breed.
Agreed, a rare breed indeed, but they are predicted with enough players playing games.
In fact every player will be this rare player, eventually. Moreover, every player will attain every overall luck score (an infinite number of times!). It might take 100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
Mongrel wrote:In fact every player will be this rare player, eventually. Moreover, every player will attain every overall luck score (an infinite number of times!). It might take 100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
00000000000000000000000000000000000000000000000000000000209357 rolls for your site-wide luck to be -1000, but it will happen.
You theoretical mathematicians are so amusing.
Bad dice are just like bad beats in poker... people tend to only remember the times that they get the worst of it and (myself included) forget the times (usually on defense because it's harder to see) they get the better of it. Over time it all evens out.
But I'm human - I still curse at the computer when bad rolls happen to me though! They suck.
My worst in recent memory was 3 against 1 in Rockem Sockem witrh damage dice (75%) and I wiffed on all three - which is a 1 in 64 chance. Then a few turns later I went for the head and had 9 against 7 - he got 8 (I had to attack from two different spots) and I got only 3! The chances of me getting 3 or less by itself are under 1%. I still won the game, but was very nervous about it! Argh!
Mongrel wrote:Mongrel wrote:In fact every player will be this rare player, eventually. Moreover, every player will attain every overall luck score (an infinite number of times!). It might take 100000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000
00000000000000000000000000000000000000000000000000000000209357 rolls for your site-wide luck to be -1000, but it will happen.
No doubt it would take less rolls than this for most of us, which brings me back to my thinking about the relevancy of the LStat numbers. I still contend that a LS number has little value without context, but I'm realizing there's more to it than just # of total rolls. There is a good amount context built in to a given LStat when it is viewed against the LStats of others. Certainly, one of the great strengths of the LStat is its zero-sum nature, which states your luck of one in the context of the luck of the other players.
But it still doesn't answer the question "How lucky were you?" For example, how unlucky do you have to be to achieve an Lstat of -1000. As you pointed out, the answer is "not very" if you have 10^200 rolls to do it in. Statements like, "Well of course I lost; I had a luck stat of -10" have no meaning in the context of a 5000 roll game, but in a 100 roll game, it becomes much more meaningful.
I have been proposing the need for an additional, different type of statistic/score that answers this question with the StDev idea in mind. Not unlike the H-score, which puts your winning percentage in the context of how many players you play in your games, the stat I would like to see would to express an LStat in the context of how many times the dice have been rolled. The problem is that simply dividing your LStat by the number of rolls doesn't accomplish this goal, and Alpha has pointed out that even using a backdoor StDev approach, which I'm pretty sure wouldn't be all that processor hungry, would still require a LOT of tables.
So I've been fooling around with some more very simple artificial equations that simulate such a number. Let's call it a Luck Score, and it's based on:
x/sqrt(n) where n = Lstat; x = number of rolls:
Variants like 100x/(sqrt(n)+log(100x)) produce numbers that range from 0 - 100+ where 1-25 looks to be marginally lucky and 100+ looks to be very lucky. Of course, this can't technically be called a stat. It would necessarily need to be called something like an (artificially) Roll-Adjusted Luck Score.
Using this system, that -10 Lstat turns into a -77 LScore where 100 rolls are used, and -30 in a 1000 roll game, and -14 in a 5000 roll game.
A -20 LStat would turn into a score of -162 in a 100 die roll context. E.g., scores over 100 fall into the extreme luck ranges.
As I mentioned before, such a system is arbitrary and not a stat per se. It is meant to be an interpretation of a given LStat over a range of rolls.
M57 wrote:So I've been fooling around with some more very simple artificial equations that simulate such a number. Let's call it a Luck Score, and it's based on:
x/sqrt(n) where n = Lstat; x = number of rolls:
Variants like 100x/(sqrt(n)+log(100x)) produce numbers that range from 0 - 100+ where 1-25 looks to be marginally lucky and 100+ looks to be very lucky. Of course, this can't technically be called a stat. It would necessarily need to be called something like an (artificially) Roll-Adjusted Luck Score.
typo:
x = Lstat and n = number of rolls.
A look at the suggestions from M57:
The is the suggestion to normalize luck stats for more meaning.
Currently, the luck stat has a very concrete meaning. Every time you take an attack with 3 attacker vs. 2 defenders, the attack is expected (average outcome of all possible) to kill 1.08 of the defenders. Likewise, the defender is expected to kill 0.92 of the attackers. The current luck stat keeps track of the difference: actual kills minus expected kills. Thus, a LS of 4.92 means that over the duration of all attacks (both on defense and on offense), you have killed 4.92 more units than is expected (You were lucky). A LS of -30.29 means that you have lost 30.29 units more than is expected (You were unlucky).
To me a LS of 100 or -100 has a meaning regardless of the number of attacks which have occurred, but there is merit to look at something different.
First suggestion, let's look at the Dice Rolled Normalized Luck Stat (someone come up with a good acronym):
DaRN Luck = LS / sqrt (n),
where n is the number of defender dice rolled (number of potential death/kills) for the calculation of the current luck stat.
For example in a game with two players might give (all 6 sided dice):
A attacks B: A luck: 0.00 B luck: 0.00
(6,6,4) to (5,4) A luck: 0.92 B luck: -0.92
(5,2,1) to (5,1) A luck: 0.84 B luck: -0.84
(4,4,2) to (3) A luck: 1.18 B luck: -1.18
B attacks Neutral:
(6,3,2) to (6,4) B luck: -2.26
(5,3,2) to (5,5) B luck: -3.34
(4,4,1) to (6,5) B luck: -4.42
B attacks A:
(4,2,1) to (5) A luck: 1.52 B luck -4.76
End of Round 1:
A luck: 1.52
B luck: -4.76
A DaRN luck: 1.52/sqrt(6) = .62
B DaRN luck: -4.76/sqrt(12) = -1.37
The issue with DaRN Luck is that I think it will go to zero rather quickly, but I do not know this for sure so if I find some time I will look at some lucky/unlucky games and see.
luck of 50 would be:
DaRN Luck: 5 (100 potential deaths), 2.23 (500 potential deaths), 1.58 (1000), 0.5 (10,000)
Second suggestion: I have no idea what to call this nor do I have idea what it calculates.
LS2 = 100LS / (sqrt(n) + log(100x))
luck of 50 would be:
LS2 = 364.99 (100 potential deaths), 191.87 (500), 144.56 (1000), 48.21 (10,000)
LS2 = 270.02 (100 using ln instead of log), 161.93 (500), 124.56 (1000), 46.08 (10,000)
The log was an attempt to reign in perceived premature high luck ratings. I.e., as the number of rolls gets high, the addition of the log has practically no effect on the score. I think LS/sqrt(n) might turn out to be pretty reasonable [actually, 100(LS)/sqrt(n)] and the jury is out regarding it's approaching 0.
Take 1 million rolls with a LS of -150 (this doesn't sound unreasonable to me). Using the latter of the two above equations, you would end up with a DaRN luck stat of 15 = Marginal luck.
In some ways, I think this suggests you are right. Over large samples, luck should all but dissappear. Even a result of 2+ standard deviations is probably only a few hundred expected wins or losses from 0 given a 1m roll sample. But in the context of 1m rolls, this kind of "supposedly" outstanding luck will probably have a minimal/marginal impact on the results of the 100s of games that were played to produce it. That's just the way it is. It's not a fault of the stat. In fact, it supports the argument that over time, luck just doesn't have an impact on your rating. Does this make sense?
Your post makes sense to me and I agree that after hundreds of games, the effect of luck should have little impact on overall standing. However, with that said, luck is being calculated all of the time and what is really important is whether or not you get lucky when you need to, something that would be near impossible to measure for a player. These moments have an extreme impact on standings (luck in the first few turns of setup, luck when taking a continent, luck of breaking a bonus, luck going for an elimination). I think that the great players of the site intuitively react and plan in such a way that the effect of luck is minimize, that is they have good strategy.
DaRN Luck is probably something worth looking at, at least at a game level as I think the current luck stat is useful at a turn level.
OKAYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY im SICK OF IT NOW.
LET ME LINK YOU THIS GAME AND SEE IF THIS IS NOT RIDICULOUS.
Start from turn 674 and ENJOY watching till the end.
http://www.wargear.net/games/player/45081
So... I don't even have enough room to talk about all the terrible luck and amazing luck of opponent so I'll just state the luck that happened within 1 round.
He breaks my 26 units with 25 and has 8 UNITS LEFT over.
I cant break his europe with 6 vs 2. I lose.
I cant break his 3 units with my 10 in south america.
See I'd have to agree that it's not about general luck... it's about when it happens. THIS WAS A CRUCIAL point of the game where Its a 1v1. this is like my 10th game of this kind of luck streak. Im so annoyed.
A number. I never use attack/transfer.
kinetix wrote:OKAYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY im SICK OF IT NOW.
LET ME LINK YOU THIS GAME AND SEE IF THIS IS NOT RIDICULOUS.
Start from turn 674 and ENJOY watching till the end.
Yes you had bad luck that game:
# | Player | Kills | Deaths | Luck |
---|---|---|---|---|
1 | Lantern | 156 (91 + 65) | 123 (62 + 61) | 8.84 (1.51 + 7.33) |
2 | Paulville | 105 (64 + 41) | 110 (52 + 58) | -1.49 (0.11 - 1.60) |
3 | samshbros13 | 45 (26 + 19) | 59 (23 + 36) | -4.04 (0.63 - 4.67) |
4 | kinetix | 143 (81 + 62) | 162 (64 + 98) | -7.08 (0.30 - 7.39) |
N | Neutral | 14 (0 + 14) | 9 (0 + 9) | 3.77 (0.00 + 3.77) |
Total | 463 (262 + 201) | 463 (201 + 262) | 0 (2.55 - 2.55) |
Of course you are on the reverse side in this game and probably didn't blink an eye: http://www.wargear.net/games/view/18508
# | Player | Kills | Deaths | Luck |
---|---|---|---|---|
1 | Amal02 | 27 (14 + 13) | 21 (12 + 9) | 2.37 (-0.51 + 2.88) |
2 | kinetix | 82 (69 + 13) | 56 (42 + 14) | 7.15 (5.11 + 2.04) |
3 | wurzel133 | 55 (23 + 32) | 72 (26 + 46) | -7.24 (-5.05 - 2.19) |
4 | Bask5 | 5 ( + 5) | 4 ( + 4) | 0.03 (0.00 + 0.03) |
5 | luvero | 10 (6 + 4) | 36 (20 + 16) | -12.20 (-7.71 - 4.49) |
6 | francis23 | 17 (5 + 12) | 19 (13 + 6) | 1.70 (-2.51 + 4.21) |
N | Neutral | 34 (0 + 34) | 22 (0 + 22) | 8.20 (0.00 + 8.20) |
Total | 230 (117 + 113) | 230 (113 + 117) | 0 (-10.67 + 10.67) |
"But many who are first will be last, and many who are last will be first." Matthew 19:30 - Good strategy for life and WarGear!