overwatch Overwatch Blue Posts: Win-Loss Ratios and Support Medals

[Valks 2,373]

Jeff Kaplan has been setting a fantastic example for others to follow in his attempt to keep on top of all the feedback the community provides. Check out some of his latest posts on the Overwatch forums, where he discusses the feedback he has received concerning win-loss ratios, as well as the introduction of a new medal for supports.

The first post is a pretty huge one, in which Kaplan goes through some pretty amazingly well thought-out points. The initial OP posted concerning his win-loss ratio of 18-32 since the stats on Beta were reset. He stated that it was beginning to get demoralising, since he simply couldn't win the majority of his games. Kaplan responded really well, walking the readers through his thought processes and what was being done, as well as why it was being done in that way. Check out the post below:

Jeff Kaplan on Win-Loss Ratios

+ Show- Hide

Hi Koot,

We really appreciate the feedback. For what it's worth, joining a match late should not result in a loss. If it does, this is a bug and will be fixed.

There are a lot of converging factors that here at work. It's a very complicated problem. Most people believe that putting people in matches together is a very scientific process (and parts of it are). But there is so much going on psychologically and emotionally that at times, matchmaking has more to do with perception than anything else.

Ideally, all players would have a 50/50 win loss ratio. I think intellectually everyone gets that. It sounds very reasonable. Or if I said, we're going to have two teams play 2 games. Team 1 will win the first game and Team 2 will win the second game. That sounds pretty balanced. I don't think there is disagreement on this part of the discussion.

Where it gets more tricky is when you as an individual decide to play 500 matches over the course of many weeks. If we were sitting over coffee and I said to you, hey you're going lose 250 matches of Overwatch in the next few weeks, suddenly that sounds pretty lousy. Losing half of your games is not what most players want. Players want to win more often than they lose (this is why we love making PvE stuff so much -- plus we don't let our bots post on the forums...)

Another factor that comes into play here is streaks. If we assume that 50/50 is a good win/loss ratio and you happen to have some good winning streaks it inevitably means that some loss streaks are going to happen. Psychologically, we tend to take note and remember the loss streaks more than the win streaks. Yesterday, there was a beta tester who started a thread citing that they were upset because they had lost 4 games in a row. It was a good thread and I appreciated the player's point of view (hearing their perception is important to us). But it also made me think about the psychology of winning vs. losing. I haven't ever seen someone post in these forums that the matchmaker is broken because they just won 4 games in a row...
 
Now, to provide a little more substance (other than my subjective rambling) we do actively break up matches when we detect players have been losing too much. We don't do anything to prevent a group or team from winning too much, however. Losing streaks are bad and ideally we could minimize these. We try to put a person who has lost many times in a row against players he/she is more likely to beat. We can't always guarantee that win, however.
 
Another compounding problem we have is what heroes players decide to play. All of us are better and worse at some heroes. It's impossible for the game to know that this is the round that you decided to start learning D.Va on. The match is already in progress by the time you might switch to her. Many times we analyze what players have perceived to be an "unfair" match and realize that the skill levels were actually very close but one team played their "best" heroes while the other team played their "worst" heroes.
 
Overwatch is a very team oriented game. There are not a lot of shooters that demand as much team coordination as Overwatch does. This is what makes Overwatch special but also can cause frustration for players queuing by themselves. Obviously, we want to support solo queuing. And we have a lot of ideas how to make it better. But it's never going to be as consistently fun as playing with others who you know and trust. Think about going out on a Friday night. Imagine going out with your 5 closest friends and doing whatever it is you think is fun. Now imagine you asked me to find 5 people for you to go out with tonight. I could be very systematic, strategic and scientific in finding those people for you. Maybe I look for people of similar age, demographic... people with similar interests. What are the odds that you're going to have fun with the 5 strangers I just picked for you? Personally, I have no clue. It could be an amazing time or it could be a disaster. If it's a disaster, you're going to blame me and my methodology for picking. That's the situation we're in here.
 
Another factor weighing in (and one that will cause controversy and ire when I mention it) is that we are in a low population scenario right now. This is by design. We're not looking to support a large population until Open Beta. Lower population situations mean that we're often "looser" in the skill banding when it comes to matchmaking. We have a protected "newbie" pool but it becomes less protected with fewer people (not to mention when people decide to form groups of varying skill and experience levels). Things will smooth out quite a bit once we have more players to match from.
 
I was a very active player in the Hearthstone Alpha as well as Closed Beta. Early on, I was having fun and loving the game. But later, I could no longer win anymore. Hearthstone was my first CCG so I was still learning. I stopped having fun in the small pool because it was full of sharks. But fast forward to today and I am usually somewhere between rank 7-12 at the end of as season (my highest is 4). So as an "above average" player the alpha was rough for me but the live game, with lots of players playing is extremely fun for me. Looking at your stats, Koot, you're a very good player. You're above average and in about the top 25% of the beta player base. So it's not you and some lack of skill.
 
We're going to make some tweaks soon to some things. We're going to look again at Groups and how we match them up. We want to be careful to NOT discourage grouping (we've seen this happen in all of our multiplayer games -- if you're too strict, players feel like grouping is bad and they prefer to play solo). Earlier in the beta we were too strict on grouping and players stopped grouping. We loosened up but we think by too much. We're also going to look at newbies (players with less than 20 matches) and player levels and see if there's something we can do there. If nothing else than to help with the perception of "why did I get matched with this guy who doesn't know what he is doing". Also, we're doing a big pass of re-flowing the new user experience. Right now, the game dumps new players into Quick Play too quickly. They should be doing the tutorial, practicing in the range and playing vs. AI before coming into quick play. We think this will help some too.
 
I know players want more information than this. But like I said, matchmaking can be extremely complicated. I hope you understand how seriously we take our matchmaking ability. We are constantly working on it. It's also really rough because we do experiments while we're in beta that we would not do in a live environment. I know this can be unpleasant for you and cause some strife for that, I am sorry. But we do appreciate you helping to make the game better.
 
--jeffrey

(source)

In the same thread as above, Kaplan noted that the team has plans to introduce a different system for "backfills" than is currently in place. In the beta currently, if you join a game late and lose, you will be given a loss, regardless of if you join with 5 seconds to spare. The team instead hopes to prevent this from happening, while still granting you a win if you join late and end up winning the game.

Kaplan on Backfills

+ Show- Hide

Just as a quick update, we have not yet fixed the issue that's causing "backfills" to get losses. In an upcoming patch, if you are backfilled into a losing match, you will not take a loss for it. On the flip side, if you backfill and win, we will give you the win.

(source)

The next blue post from Kaplan gave us new insight into the statistics surrounding the ever-complained about Temple of Anubis. The OP posted concerning the choke point on the map in which the defense team can often hold indefinitely, unless you employ a solid strategy and efficient teamwork. A patch was implemented recently that changed elements of the map, so Kaplan posted statistics relating to games played since Tuesday.

Statistics on Temple of Anubis

+ Show- Hide

For perspective...

As of Tuesday's patch Temple of Anubis is won 49.2% of the time by attackers and 50.8% of the time by defenders.

The first point falls to attackers 76.9% of the time.

(source)

The last post coming from Kaplan was a short, but extremely welcome one. It looks like the final blow medal will be replaced with a support-based medal, in order to provide some more recognition for the non-damage dealers.

New Medal

+ Show- Hide

We just replaced the Final Blow medal with Healing Done. Coming soon(ish)

(source)