8 replies to this topic

[occusoj]

To keep track of the new mode-vote MM and stomps I took screenshots of every match I played today. Entering the data in a table showed one map completely missing from the rotation.

So, my question is: How many consecutive games in the same mech would it take for you to
a.) raise some suspicion that map selection maybe skips one?
b.) provide reasonable evidence that makes it worth discussing if it really does so?

Thanks in advance for your answers.

Edited by occusoj, 08 October 2014 - 03:41 PM.

[LauLiao]

If you're asking when statistically there is enough evidence to say conclusively that a map is being purposely left out and not due to random chance then never.

If we have 10 maps, then every game in theory there is a 1 in 10 chance for each map. Statistics don't have memory, so it's entirely possible for you to NEVER get a certain map and it still be just random chance.

That's how they get suckers at the roulette table in Vegas. They show the last winning numbers/colors so someone walking by will say, "Oh hey, the last 10 wins were all on black, so red MUST be coming up." Nope, you've still got a slightly below 50% chance of landing on black, and an equal chance of landing on red.

Edit: Also, for the record, I've played on EVERY map within the last month or so.

Edited by LauLiao, 08 October 2014 - 03:52 PM.

[Monkey Lover]

I did 50 to show hot mechs get hot maps ,also lrm boats get canyon, river and forest. For what I can tell the new map is set on the lrm boat rotation

No one cares most said I was wrong. So maybe 500 ?

Edited by Monkey Lover, 08 October 2014 - 03:53 PM.

[occusoj]

No, Im not looking for a bulletproof 100% undeniable evidence. As you states - thats not possible.
I could play 1000 matches, never get a map and than get it 100 times in a row. Theroretically not broken at all.

Its like rolling a dice for a whole day and never getting a 6. While still perfectly fine theoretically it would still make me suspicious. And if that happened on a second day it would be at least worth discussing to me.

I liked analysis and solving all kinds of differential equations but I admit beeing repelled by statistics .

[Rebas Kradd]

LauLiao, on 08 October 2014 - 03:51 PM, said:

That's how they get suckers at the roulette table in Vegas. They show the last winning numbers/colors so someone walking by will say, "Oh hey, the last 10 wins were all on black, so red MUST be coming up." Nope, you've still got a slightly below 50% chance of landing on black, and an equal chance of landing on red.

Gambler's fallacy.

I had to defeat this in football arguments, too. Back in 2011, fellow Seahawks fans of mine were insistent on trying to take a quarterback in the later rounds and develop him into a starting QB. They pointed to Tom Brady and Tony Romo as examples that it was possible, then, when confronted with the reality that the overwhelming majority of late-round picks were total busts, delivered this gem: "Sooner or later, a late-round pick HAS to work out!"

Nope. Gambler's fallacy. The chances of the next late-round flier working out are exactly the same as that of the last one. (The reality was that fans were just paralyzed by the idea of a first-round pick, which carries great risk but also the majority of the success stories.)

Of course, then the Seahawks had to go draft Russell Wilson in the third round and prove those dimwits right. When will I ever win an argument?

Edited by Rebas Kradd, 08 October 2014 - 04:08 PM.

[Monkey Lover]

occusoj, on 08 October 2014 - 04:00 PM, said:

No, Im not looking for a bulletproof 100% undeniable evidence. As you states - thats not possible.
I could play 1000 matches, never get a map and than get it 100 times in a row. Theroretically not broken at all.

Its like rolling a dice for a whole day and never getting a 6. While still perfectly fine theoretically it would still make me suspicious. And if that happened on a second day it would be at least worth discussing to me.

I liked analysis and solving all kinds of differential equations but I admit beeing repelled by statistics .

If all you care about is how often maps are played and not by what layout you have look at your profile stats.

Edited by Monkey Lover, 08 October 2014 - 04:03 PM.

[occusoj]

Quote

If all you care about is how often maps are played and not by what layout you have look at your profile stats

Im interested in both. How often are maps played (expect it to be very even overall) and how is the distribution for each layout. What Ive seen today just sparked my interest on that.
But in order to get any somewhat useable information out of it I have to know what kind of sample size is needed to get near a representative sample.

And with the current build Im at a whole evening of playing with one map totally missing. So Im dropping only in this mech/build and waiting to get that one map.

You stated 50 as a sample size, well that seems manageabe. 500 consecutive drops in the same mech, even though I really love that build - dont know if I can take it.

[Monkey Lover]

The odds are 1 in 144 you will play the same map two times in a row. Its 1 in 20736 you will play it 4 times. I have played maps 4 times in a row a couple times. Its not random its more "average" if anything.

Again what ever sample size you do no one will believe you and will just say it's in your head.

Edited by Monkey Lover, 08 October 2014 - 05:07 PM.

[Monkey Lover]

ok well as for your sample size. I have played 4,000 matches or so. My stats show its average to around 10% error.

http://www.surveysystem.com/sscalc.htm

4000 games
with 10% error in the sample
would be 97 matches.