44 replies to this topic

[ItsAPotato]

I've felt lately that I've been dropping on the 2 River City maps rather more often than I should be, and now with stats, I can confirm that I am not [totally] crazy.



That's a ~37% frequency of River City (both variants). Should obviously be closer to 25%.

There are a few other anomalies, but nothing too major. Alpine occurs 5%, should be 12.5%, and Caustic occurs 17%, should be 12.5%

Unrelated: I apparently suck horribly at Forest Colony, but am rather good on Forest Colony Snow

Anyway, 69 total matches is not a great sample size, so I thought I'd see if anyone else is seeing this, as I've talked to several folks who have seen similar results with River City and I'm too impatient to wait another week or two

[Lefty Lucy]

It's random, it's not evenly distributed.

[Cebi]

MOAR STATS

[Kunae]

ItsAPotato, on 08 March 2013 - 07:43 AM, said:

I've felt lately that I've been dropping on the 2 River City maps rather more often than I should be, and now with stats, I can confirm that I am not [totally] crazy.

<snip>

That's a ~37% frequency of River City (both variants). Should obviously be closer to 25%.

There are a few other anomalies, but nothing too major. Alpine occurs 5%, should be 12.5%, and Caustic occurs 17%, should be 12.5%

Unrelated: I apparently suck horribly at Forest Colony, but am rather good on Forest Colony Snow

Anyway, 69 total matches is not a great sample size, so I thought I'd see if anyone else is seeing this, as I've talked to several folks who have seen similar results with River City and I'm too impatient to wait another week or two

I see much the same, if not more skewed than yours. RCN 12, next closest 7, of which there are a few.

[Agent of Change]

To be fair i'm noticing a statistical spike specifically with river city night, how ever neither my own nor Potato's sample set is really statistically relevant at this point.

If I hit 200 games and i'm seeing the exact same proportional spike on the same map I think that would be a buit more firm but i haven't seen anything at all to indicate it isn't entirely random with any give map showing up 12.5% and any given map with a variant showing up 25% (ignoring original or variant)

[Th0rsten]

River city is 10 of 43 drops for me so pretty close to the 25%

But my 43 drops are even more insufficient to be taken seriously for statistical analysis than your 69 :/

I guess it is better to be patient and see how it will look like with maybe ~500+ drops.

[ItsAPotato]

I agree, the sample size is too small at this point to make anything statistically significant.

But I was hoping to merge my data with everyone else's to get a larger set, rather than wait another 2 weeks (and I'm bored at work)

[TexAce]

Damn, i wanted to make a similar thread where I wanted to ask everybody to post their screenshot so we can add numbers.

I do believe something is off with the RNG (since I know that you can influence the RNG by variables).

so please everybody post yours!

If we get numbers from like 2000 matches its enough to see where the numbers are going. I didnt played enough in the last days to post mine but will do after the weekend.

Lefty Lucy, on 08 March 2013 - 07:43 AM, said:

It's random, it's not evenly distributed.

A pure random number generator (in theory) should distribute it evenly, we try to prove that it doesn't so the devs can fix it.

Edited by TexAss, 08 March 2013 - 07:56 AM.

[Kunae]

ItsAPotato, on 08 March 2013 - 07:53 AM, said:

I agree, the sample size is too small at this point to make anything statistically significant.

But I was hoping to merge my data with everyone else's to get a larger set, rather than wait another 2 weeks (and I'm bored at work)

Mine: 45 games : 12 of which were RCN.

[Kekkone]

77 games of which:

Forest colony + variant ~38%

River city + variant ~20%

Caustic valley ~6%

Frozen city + variant ~22%

Alpine peaks ~13%

Edit: Added the rest of the data

Forest Colony 16
Frozen City 9
Caustic Valley 5
River City 8
Forest Colony Snow 13
Frozen City Night 8
River City Night 8
Alpine Peaks 10

Edited by Kekkone, 08 March 2013 - 08:23 AM.

[Lyrik]

A data sample of 69 drops to complain about a random factor? Wow ...

[Agent of Change]

well here then, why didn't you say so:
mine + yours = new total

Individual maps:

ForC 11 + 6 = 17 (-2)
FroC 11 + 5 = 16 (-3)
CauV 10 + 12 = 22 (+3)
RivC 8 + 11 = 19 (+0)
FoCS 10 + 9 = 19 (+0)
FrCN 16 + 7 = 23 (+4)
RvCN 7 + 15 = 22 (+3)
Alpline 9 + 4 = 13 (-6)

Total games 66 + 82 = 151

Map probability: 12.5% = 18.87 (19)
Set Probability: 25% = 37.75 (38)
Rounded for quick math

Sets:

ForC (Combined) 36 (-2)
FroC (combined) 39 (+1)
RivC (combined) 41 (+3)
Caustic/alpine 35 (-3)


Some error for rounding but all individual maps are with +/- 6 of individual probability and the "sets" are all within +/-3 so given the limited sample set it's looking pretty spot on to random.

Edited by Agent of Change, 08 March 2013 - 08:11 AM.

[ItsAPotato]

Lyrik, on 08 March 2013 - 07:59 AM, said:

A data sample of 69 drops to complain about a random factor? Wow ...

Oh! I totally didn't realize that 69 was a small sample size!

ItsAPotato, on 08 March 2013 - 07:53 AM, said:

I agree, the sample size is too small at this point to make anything statistically significant.

Oh, except I totally did...

ItsAPotato, on 08 March 2013 - 07:43 AM, said:

...Anyway, 69 total matches is not a great sample size, so I thought I'd see if anyone else is seeing this, as I've talked to several folks who have seen similar results with River City and I'm too impatient to wait another week or two

...twice...

Asking the community to compile their data!?! What is this sorcery?!?

[Kunae]

Not really looking at, or caring about the other maps, in this post. The only stats that matter are:

Total games played:
Occurrences of RCN:

If we limit to those two data-bits, this post becomes cleaner and alot easier to collate.

[Lefty Lucy]

TexAss, on 08 March 2013 - 07:54 AM, said:

A pure random number generator (in theory) should distribute it evenly, we try to prove that it doesn't so the devs can fix it.

That is absolutely true... over a large enough sample size. True randomness is not evenly distributed, but has streaks and "patterns". Over small sample sizes true randomness does not create an even distribution, in fact quite the opposite.

[Kunae]

Kunae, on 08 March 2013 - 08:13 AM, said:

Not really looking at, or caring about the other maps, in this post. The only stats that matter are:

Total games played:
Occurrences of RCN:

If we limit to those two data-bits, this post becomes cleaner and alot easier to collate.

For example:

Total games played: 45
Occurrences of RCN: 12

[Agent of Change]

people can post just the list of number of games for each map in the proper order as they appear top to bottom (a total would also be nice but not neccesary)

no need to label them if they are in order and i will compile them.

[Agent 0 Fortune]

This is why we can't have "through armor crits" there would be one player who kills everything with 1 hit, while half of us get "deflected" every shot.

I wonder if they used the same RNG in the ELO system or if it is all calculated?

[Agent of Change]

Kunae, on 08 March 2013 - 08:14 AM, said:

For example:

Total games played: 45
Occurrences of RCN: 12

adding you rcn data:

total games: 151+45=196

Expected Percentage: 12.5% of 196 = 24.5

Total RCN: 22 + 12 = 34 (+9.5 games or 17.3%)

but contextually i'd like to see all the numbers.

Edited by Agent of Change, 08 March 2013 - 08:21 AM.

[Lyrik]

Lefty Lucy, on 08 March 2013 - 08:13 AM, said:

That is absolutely true... over a large enough sample size. True randomness is not evenly distributed, but has streaks and "patterns". Over small sample sizes true randomness does not create an even distribution, in fact quite the opposite.

Excactly. Until we get several 1000 drops samples this is just a waste of time ;-)
I really don't complain that I didn't got Caustic in 28 matches but 6 Frozen City and only 2x Alpine. I wait until I get several hundred matches then I will fill a bug report.