MischiefSC, on 09 February 2014 - 02:19 PM, said:
In reading here I'm curious to see a few things -
When did the MM stop taking tonnage into account? It's always attempted to match tonnage.
To the best of my knowledge, tonnage was no longer taken into account when they implemented the Elo system. Regardless, we have been informed several times in the "Patch Feedback" forum that it is not taken into account at this time but they plan to implement it in April...a couple of months away now.
MischiefSC, on 09 February 2014 - 02:19 PM, said:
When did Elo get reset? My understanding is Elo did NOT get reset (yet) but that when no match is available that you'll fit into within 180 seconds it broadens the target in order to find a match and do so more quickly.
Officially? It probably hasn't. A lot of us have noticed that our stats have been borked, however. If you check your "mode" stats, you'll see that you've played an equal amount of Assault, Conquest and Skirmish matches (which is odd because since Skirmish came out, I haven't played anything else....except 1 round of Conquest when I forgot to pick Skirmish on accident). Assuming that our 'Elo' is based on previous performance, this could be considered a "reset," I guess. It all depends on whether they actually screwed up the information they have stored or if it's just a glitch in the website reporting stats.
MischiefSC, on 09 February 2014 - 02:19 PM, said:
Currently, at least as of Jan 23, threshold is 1400 for Elo - as in it'll grab people within a max range of 1400 Elo of each other. IMO way, way too wide but I'm not holding the data showing populations and average time to find a match. Currently the average difference (as of Pauls post from 1/21/2014) was ~175 points of Elo difference between teams. Average tonnage mismatch is ~100 tons or less, most within 50 tons. There are certainly outliers.
Ok, assuming this is the case, then there is a very obvious problem here. According to what we've been told, a "new player" starts off with an Elo of 1300 and it is modified +/- for every fight. So, it's entirely possible that those of us that have been playing a while and have had our score shift up and down are ending up with a 1400-1600 while the guy that started 10 minutes ago is within our range for no effort on his part, whatsoever. Perhaps they're starting new players with too high a score.
But, then again, that brings up yet another problem. If they were to start the new players at a lower score, they'd NEVER find a match because of the small sampling of the player base. People are leaving this game in droves due to (fill in the blank...the reasons are too numerous to mention here). PGI doesn't seem to mind that they're losing players daily or, rather, they're not communicating any concern to the rest of the community...not as long as the cash flow continues, I guess.
As for the tonnage part, again, PGI has stated that it's not even considered part of the matchmaker now. Nor is any form of equipment such as ECM.
MischiefSC, on 09 February 2014 - 02:19 PM, said:
Weight limits are going to change that metric dramatically. I'm curious to know how that's going to be implemented - your Elo score is tied to the weight (mech class) of the mech you're bringing so you can't match teams for Elo then let them pick mechs.
It's possible, but I doubt it. More on this at the bottom.
MischiefSC, on 09 February 2014 - 02:19 PM, said:
Next, the game does not try and force you into a 50/50 win/loss. This has been explained a lot of times but I'll do so again just because it comes up, gets answered, then doesn't show up for a couple of these same threads, then comes up again.
Elo does NOT try to force anyone to a 50/50 win/loss. The matchmaker uses Elo and weight matching to try and put players in as even a match as possible in any given match but that can vary - sometimes there just aren't 23 other players in your same Elo range and in the right balance of mechs to make two split teams with balanced Elo and tonnage who hit 'launch' within 180 seconds of each other. So it builds as balanced a match of teams as it can, then based on the variance predicts who it expects to win. Then it alters the awards for winning and losing accordingly. If you're predicted to win because you had the advantage in that match you gain little if any Elo. The loser loses little if any. If it's an upset the matchmaker recognizes that the people in that match were either better or worse than predicted and gives them an adjustment up or down of between 0 and 50 points based on the difference in their scores.
If you win some matches the matchmaker does not put you in a difficult match to make you lose. That's absolutely false. It does try to consistently make challenging matches for you based on your performance in previous matches - what skill level of people do you beat consistently, who do you have trouble beating.
Again, that may be your opinion, but there are many of us that feel differently. Perhaps you have a higher Elo than us and are ending up in better matches. Perhaps you play at a specific time of day where some of us play at all hours of the day. Who knows? I could just as easily say "It seems to me that the Matchmaker is purposely putting together teams designed to be a Roflstomp every single match" and I'd get just as many people to agree with that statement as yours. It's all a matter of perspective and unless either one of us are actually developers working on the project, all we can do is conjecture.
MischiefSC, on 09 February 2014 - 02:19 PM, said:
Battlevalue is skewed and unreliable. Tonnage is important but secondary to skill. Ideally the matchmaker would match players to as close a relative skill range as possible in the mechs they're in. I'm the first to say there's room for improvement but that without question the introduction of weight limits is going to utterly change the matchmakers required tools and weighting and trying to alter anything in a big way before then would be precipitous.
Again, we will have to agree to disagree. The Battlevalue system not only works, it works well. Granted, however, it is a system used to keep teams somewhat even for Tabletop Battletech.....but, Elo is for Chess so...whatever.
The BV point value is based on a total of your Defensive and Offensive totals. I'm not going through all the math here, but I'm going to give you a couple of "for instances."
One guy has an Atlas D-DC with 2x LL, 2x UAC5, AMS, ECM, STD325
The other guy has an Atlas D with 2x LRM20+A, 1xAC20, 4xML, STD 285
Which one is going to have the higher BV? I don't even need to do the math to tell you. The D-DC will. It gets BIG Defensive points for having AMS and ECM. The STD325 makes it faster and speed is a multiplier.
Even though it doesn't really factor (although it could), the D having 2 LRM20s and only 10 tubes to support it takes away from it's offensive value. Lack of anything defensive (AMS, BAP) kills it's rating as well.
So, from a player's perspective...they both weigh 100 tons but are they actually "equal?" From a player's perspective, absolutely not. I'd MUCH rather have a guy with ECM and close brawling weapons than an Atlas LRM boat wannabe.
From a BV standpoint, are they equal? Not even close.
Keep in mind that this is a "bare bones" approach to Battlevalue. Certain eccentricities of MW:O do not necessarily apply as they stand (example: TAG, Narc, BAP shut down ECM but do not in BT and should be considered both offensive and defensive, modules do not exist in BT, etc). But, mathematically, it wouldn't be too difficult to assign them a value and put them into the system.
Don't misunderstand, tonnage balancing will go a LONG way toward getting more even teams but, Battlevalue DOES work.
7
This topic is locked
