In season six, our team's 5v5 matchmaking rating was around ~3020 while our team rating was about 2815. Even though we were playing teams that were only 2200-2400 rated, we would still gain significant points because of the large gap (200ish points) between MMR and TR. However, having a very high MMR is not always favorable -- upon our first loss (disconnects are so unfortunate sometimes), we dropped 47 team rating... and a lot of matchmaking rating.
The most one can lose in a single match governed by an Elo-based rating system (like the one used in arena seasons one through four) is ~30 team rating. Dropping from 2815 to 2785 might not seem that much better than dropping from 2815 to 2768 for the same type of loss -- until you factor in MMR.
By losing to a 2300 team, our 2800 team dropped 177 matchmaking rating. Again, MMR determines how quickly your TR jumps or falls. By losing 177 MMR, we had lost 47 TR in addition to an incredible amount of TR generating potential.
In an Elo system, even if you lose the most amount of team rating possible (~30), you only lose that much matchmaking rating because your team rating is your matchmaking rating. By adding in a second system, potential wins and losses are enormous and streaks become much more important. Winning multiple games in a row gives a team an inflated MMR which allows TR to boost up. Losing many games in a row does the same thing in the opposite direction.
Think about the New York Yankees winning a baseball game against the Philadelphia Phillies. The Yankees get their first win of the season and are now 1-0. After their win, they play against the Chicago White Sox, which are also 1-0 on the season, but the Yankees lose. The Yanks are now 1-1. However, because the White Sox won two games in a row, their record is now 3-0, even though they've only played two games of baseball on the season.
While this is a vast oversimplification of what's going on, MMR and TR piggyback one another pretty hard which is why we've seen some pretty ridiculous ratings inflation even though arena representation is much lower than it was in The Burning Crusade.
The Good, the Bad, and the Ugly
The goal of MMR is for gladiators to fight gladiators, duelists vs. duelists, challengers vs. challengers, etc. The MMR system actually does this fairly well. While some gladiators will have a 3v3 MMR of 2950 and other gladiators will have 2600 3v3 MMRs, they still get paired up even though the disparity of points is great. People with 1500 MMRs generally get paired up against other people with similar MMRs from 1400-1600.
I've been playing a lot of chess lately. Don't worry, this is going somewhere.
I find when I play against people with a similar skill level to myself, we trade games back and forth. This is, after all, how a balanced game works when you play multiple games against someone with basically the same skill level. However, I've also noticed something else: my own skill does not dramatically improve. I get much better at chess when I play people who have more experience at the game than I do. It makes me a better player to face different successful strategies.
Arena is a lot like chess. If you don't play opponents who are much more skilled than yourself, you can't see new and interesting strategies to combat or mimic. This is why "top arena players" sometimes aggregate to one battlegroup, or why they want to get sponsored and do LAN events with other top players. It makes them better players.
In this vein, matchmaking rating has robbed mid-level arena teams from trying to combat the top players on the battlegroup. I remember my first time facing the #1 2v2 team on our battlegroup. It was season two; we were priest-warlock vs. hunter-druid. The hunter-druid team was incredibly good and overall did very well against us on the day. We were able to sneak a few wins in, although they gained about 30 rating off of us overall. I think we went 3-6 against them, maybe 2-5. The fact that we were able to beat them at all gave us confidence to continue on. Without those games against them, I might have given up and followed other endeavors in WoW.
Team Rating: 0
I'm not sure how to flow into a negative statement, so...
Negative statement incoming! Thou hast been warned.
Starting at a team rating of zero is possibly the worst thing that has ever happened to arena. I'm really not a big doom-and-gloom guy. I think most of my work praises Blizzard for trying to make arena fun and friendly. Unfortunately, this is not one of those times.
The idea behind starting arena teams at zero was to encourage new arena players. Starting at 1500 is awesome, but when you're on the same team for a few days and suddenly you're down to 1100 rating, it can get a bit depressing.
Blizzard developers decided that starting at zero would be much more casual-friendly. If you only play 10 or 20 games a week, you make a substantial amount of progress -- especially because when you lose, you lose 0 points, but when you win, you win anywhere from 1-47 points! Sounds great so far, right?
Well, not really. You see, that doesn't really fix the old problem. If a team is a 1100 rated team, they will naturally gravitate to that rating. Matchmaking rating (as we've mentioned before) actually bogs newer players down because they can't make substantial TR gains, and because they can't play much better players to evolve superior strategies. Moreover, if a team with the new rating decides they don't want to be at 1100 rating anymore they have to do one of the following...
- Team hop, which the new system heavily discourages.
- Get new players on the team. (how many players want to leave their team to join a 1100 rated team?)
- Start a new team with new players. (and make the 50+ game grind again)
A team in the old system had to simply disband the team, create a new one, and boom -- they were instantly up 400 points from 1100 to 1500. You might think that this would cause ratings inflation because those 400 points were essentially created out of nowhere. Yep, no doubt ratings inflation did start to creep up with the end of season four. I think my Brutal Gladiator 5v5 team was just over 2300. A number which is obviously pale in comparison to many of the ratings out today. I was going to make this an individual argument, but I think it's sufficient to say that ratings inflation has gotten out of hand with the Gaussian Density Filter
You might be keen to the clause that states if you join a team that has a team rating of 1000 or more, you will instantly get 1000 personal rating when you join.
Starting at 1000 personal rating if someone grinds a new team up for you isn't much better. That's 500 points lower than 1500 which is still 11+ wins you have to sit through until you can break the magic barrier of 1500. 11+ wins takes a good amount of time, especially if you have lots of losses in between. You'll have to play far more games than that if you have a matchmaking rating lower than 1500 (which remember, are the people we're trying to penalize the least).
Sadly, a measure established to prevent new players from growing bored of arena or advancing is doing exactly that. Teams need to start back at 1500 so new players aren't penalized for losing or having a low MMR.Next Week
Okay, so I'm pretty sure I'm done talking about rating systems -- like, forever. Next week we'll talk about some interesting compositions in arena and why they succeed.
Want to ascend the arena ladders faster than a fireman playing Donkey Kong? Check out WoW.com's articles on arena, successful arena PvPers, PvP, and our arena column, Blood Sport.