

Elo Question
#21
Posted 26 May 2014 - 10:59 AM
#23
Posted 26 May 2014 - 11:10 AM
Say in a hypothetical universe we have four identical 'mechs facing off against four other 'mechs. If each 'mech shoots at the one in front of it, they'll all take about the same damge, and be destroyed (or nearly/partially destroyed) at roughly the same time. However, if our side shoots at just one of the enemy, they'll kill that enemy four times faster - each of them would only be down 25% when the first enemy was destroyed. Now our team hasn't lost any firepower, but the enemy is down 25% of theirs; by the time we've killed their next guy, the're only gonna take out 20% more armor off the three of us they're still shooting at. The process will continue until eventually the entire enemy lance is destroyed without quite killing a single one of us. This is the standard basic illustration of how focus fire works. It's simplified - but it's accurate, assuming equal skill and loadouts.
In real matches, this snowball effect often works out even better for the team who can best focus their firepower - because they can isolate enemy players, or focus the fire of many snipers on just one guy without taking return damage. As a practical application, then, focus fire means that the first team to lose two 'mechs usually has a significant disadvantage - and that this disadvantage tends to snowball. Especially as players are being trained to respond to danger by seeking cover in stead of doing something about it, you just end up with match scores of 5v12, 4v12, etc. It's not a matchmaker failure - that's mythology. It's just the nature of this kind of game.
#24
Posted 26 May 2014 - 11:15 AM
LordKnightFandragon, on 26 May 2014 - 10:48 AM, said:
Just wanna play, on 26 May 2014 - 10:59 AM, said:
PGI has declined to release Elo rating in order to prevent gaming of the system - but Elo (which is not an acronym! /pedantry) is not based on your stats. It's based on your performance in matches relative to the comparison of your team's and the enemy's Elo ratings. So it's related to your stats, but not based on them - you can theorize pretty soundly that if your win/loss is high, you have a fairly high Elo - but you can't accurately calculate your Elo from your stats.
#25
Posted 26 May 2014 - 11:20 AM

You can assume low ELO.......
Edited by Just wanna play, 26 May 2014 - 11:20 AM.
#26
Posted 26 May 2014 - 11:25 AM
Void Angel, on 26 May 2014 - 11:15 AM, said:
I keep accidentally holding the shift button down when I type it out.
Could be he is having a similar problem.
Just wanna play, on 26 May 2014 - 11:20 AM, said:

You can assume low ELO.......
See now, that is what I would have thought from my W/L - but then I started running into players like Eglar fairly regularly.
Remember - it doesn't necessarily drop your number if you lose.
Only if it predicted you would win, and you lost.
#28
Posted 26 May 2014 - 10:19 PM
Shar Wolf, on 26 May 2014 - 11:25 AM, said:
Could be he is having a similar problem.
See now, that is what I would have thought from my W/L - but then I started running into players like Eglar fairly regularly.
Remember - it doesn't necessarily drop your number if you lose.
Only if it predicted you would win, and you lost.
not having an issue holding shift for elo, it just makes sense that it would be an acronym of sorts, but judging from what ive seen so far of PGI and this game, they have as much sense as SoE....and they are pretty lost..
I mean, ELO? How can we just arbitrarily name something elo rating and have it not mean something?
#30
Posted 26 May 2014 - 10:59 PM
LordKnightFandragon, on 26 May 2014 - 10:19 PM, said:
not having an issue holding shift for elo, it just makes sense that it would be an acronym of sorts, but judging from what ive seen so far of PGI and this game, they have as much sense as SoE....and they are pretty lost..
I mean, ELO? How can we just arbitrarily name something elo rating and have it not mean something?
Actually Elo is named after Arpad Elo who was a master-level chess player.
#32
Posted 28 May 2014 - 01:14 AM
#33
Posted 28 May 2014 - 01:20 AM
Gaussfodder, on 28 May 2014 - 01:14 AM, said:
No. Elo moves are based strictly on your wins and losses as compared to the result predictions (which are based on your Elo going into a match).
#34
Posted 28 May 2014 - 10:34 AM
#35
Posted 28 May 2014 - 10:38 AM
Void Angel, on 28 May 2014 - 10:34 AM, said:
Algebra is also actually pretty simple.
People still struggle with it though.

#38
Posted 28 May 2014 - 12:06 PM
Redshift2k5, on 23 May 2014 - 02:38 PM, said:
This doesn't lead to stagnation tough, because in general players will gradually improve their own skills, so your Elo is constantly being challenged and if you win, your elo rises, setting you against tougher opponents.
It does not intentionally put you against higher opponents because your wins are too high, but it does gradually give you a higher Elo matching you with equal opponents.
In general your wins:losses should be fairly even over time, but your Elo will gradually increase as you improve putting you against progressively tougher opponents.
Maybe this was what their MM is supposed to do, but I can say that after over 10,000 matches and over 60 mechs grandmastered, that MM does NOT do this. I have a positive W/L ratio, and it is improving with my skills and premade teamwork, but it started way below 1 when I was a noob at MWO.
Edited by 7ynx, 28 May 2014 - 12:07 PM.
#40
Posted 28 May 2014 - 02:18 PM
Basically you can't win no matter where you go.

But, you have become famous!
Anyway. What Redshift said is the principle behind ELO. But since ELO is a system based on two players and MWO has 24 players, some really wonky things come along.
7ynx, on 28 May 2014 - 12:06 PM, said:
I purposely tank my kdr to try and keep at 1.0 and I gotta tell you it just doesn't work.


1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users