Yomi Season 2 Ends, Season 3 Begins
Thursday, December 8, 2011 at 2:05PM I posted here a while ago about the Season 1 results for Yomi, and how the ranking system works. While Puzzle Strike is also available on fantasystrike.com and now has all TWENTY characters with ranked matches, this post is about Yomi's ranking stats.
Season 2 just ended, and these results are in ranked matches from June 20th to December 6th, 2011.
DeGrey 786 / 1430 55.0% Midori 712 / 1329 53.5% Argagarg651 / 1220 53.4% Rook 836 / 1613 51.8% Geiger 702 / 1405 50.0% Grave 833 / 1672 49.8% Lum 482 / 992 48.6% Setsuki 1066 / 2253 47.3% Valerie 706 / 1483 47.6% Jaina 920 / 1986 46.3%
Again, the balance has held up remarkably well, and is closer than any known fighting game. In season 1, the surprise was that Midori was claimed to be the worst, yet he had the highest win rate in the end. DeGrey is the other character clamed to be worst, and this time he posted the highest win rate, with Midori at #2. It's important to keep some perspective though, as these are razor thin differences anyway.
Now let's look at those stats ranked by popularity instead by win-rate:
Setsuki 1066 / 2253 47.3%
Jaina 920 / 1986 46.3%
Grave 833 / 1672 49.8%
Rook 836 / 1613 51.8%
Valerie 706 / 1483 47.6%
DeGrey 786 / 1430 55.0%
Geiger 702 / 1405 50.0%
Midori 712 / 1329 53.5%
Argagarg 651 / 1220 53.4%
Lum 482 / 992 48.6%
Just like in season 1, Setsuki is the most popular and has the worst win-rate. All those noobs dragging down the stats! Jaina has the same effect. She's believed to be very strong, though she's showing the same signs, given that she's #2 in popularity yet last in win-rate. I think her last place here will surprise to a lot of people. Lum is still hard and still less popular, Midori is still strong but less popular. Characters shuffled around the win-rate list a fair bit, which is what we'd expect for a game where characters are very close in power level.
Some people correctly pointed out last time that the matchmaking system would make even really imbalanced characters appear 50-50. While that's true, it's only true for a perfect matchmaking system. Ours is so far from perfect right now, that it hardly matters. You're basically matched with whoever most of the time. Speaking of that, how about we make it a whole lot easier to get a match? Read on....
Best of 3 Matches vs. Best of 1 Matches
Season 2 was a best of 3 format, meaning when you quckmatch, you play more than one game until someone has two wins total. Season 3 will revert back to best of 1 format (just one game), because the players are asking for that change, and the data supports it.
One thing in favor of best of 3 matches is that the more games you play in a match, the more fair the result is (less swings from randomness.) During the Dreamhack Starcraft 2 tournament, they mentioned that 85% of best of 5 matches were won by the player who won the first game. I thought that was pretty interesting. Here is Yomi, we'd expect the best of 3 format to magnify any character imbalances...but it didn't. It's a tight spread either way.
Another effect we'd expect is that characters who have higher variance in their gameplay (like Makoto in Street Fighter 3s) would do "unfairly" good in a best of 1 format. Best of 3 would "fix" that by making them get lucky twice to win a match. DeGrey is such a character, and yet he managed to top the win rates in best of 3 anyway while not topping them in the best of 1 season!
Really all that stuff about fairness of best of 1 vs best of 3 doesn't add up to much. In both formats, the best players are rising to the top and the character balance is holding. But there is something that matters A LOT. And that's the length of time that a match takes. There were fewer quickmatches this season, even correcting for it being shorter than season1. Many players complained that they would use quickmatch more if only it was a shorter time commitment. Well now it is. Season 3 started last night, and it's best of 1 format. Matches are fast and quickmatch will be more popular.
Gold Testing
Quickmatches also give you "gold" now. The exact amounts are being adjusted and the gain rate is way high right now just for testing purposes. All this gold is just fake anyway and it will be completely reset when this (short) season ends when we go to an actual launch. At that point, the gold will be an option for you to buy decks without using real money. You will also be able to buy decks directly, you'll be able to subscribe to unlock all characters in all games, and there will always be two different free characters available to everyone each week.
Fantasystrike.com
The site itself has going through a lot of upgrades recently. It now has much more "real" art for headers and interface. The forums have been completely upgraded recently as well. We've added the ability to record and playback matches. There will be another round of major graphical upgrades to the "play" page--the place where you quickmatch, see lists of available matches and so on. This will also include graphical upgrades to the pre-match "versus" screen, profile pages, and various popup windows. I'm really looking forward to this, and to your continued support after our launch that will allow us to improve the site more and more, keep the games updated, and to add new games like Flash Duel and our customizable card game someday. Oh by the way, Flash Duel is now in stock and ships starting tomorrow and monday, depending on when you ordered!
Oh, and congratulations to friiik, the Yomi Season 2 winner. Note that he didn't *just* place #1 on the leaderboard, but also dominated in tournaments. Top skills!
Reader Comments (13)
Nice post, as usual. I wonder how you would respond to a claim that the reason the game is so well-balanced is that for a one-on-one competitive game, there is a very high amount of luck or variance involved.
It doesn't surprise me at all that Jaina is bottom. She's strong, but she's *hard* to play well, which is a bad combination with "popular with newbies".
I'm still surprised Grave is so middling, again. Deg being high doesn't surprise me, since he's sometimes considered a "noob friendly" character.. No one plays Midori or Arg, so it's no surprise that they're high, haha. But Grave is damn easy (provided you don't fall into the "OMFG, TPOS FOREVAR!!!!" trap mindset), yet he does worse than a more popular character that's generally considered harder (Rook). Odd.
I really love fantasystrike.com by the way. I've purchased the real version of Puzzle Strike but I've probably played the online version way more by now (just because it shuffles for you and enforces rules automatically).
For me personally, if fantasystrike had some some sound effects (like shuffling...crash sounds maybe) would make it THAT much better.
I know that requires more work without pay (fantasystrike being free and all) and this isn't exactly related to the article per se. Just wanted to give my input since I've followed your blog for many years now and enjoy and appreciate the effort you've put.
- Jeff
Jeff, it does need sound effects. Puzzle Strike in particular needs a lot of UI enhancements too. I can't even follow what's going on as a spectator sometimes. We'll try to make it a lot more "readable" what's happening without you having to read the game log. That will be before sound effects, but after the next round of graphics updates for the "play"/matchmaking page. I guess we are pretty slow here, but things will keep getting better.
Keith I'm not sure where to even start with the whole "it's all just random" claim. The whole competitive scene seems to be a counter to that. Or all the strategy resources here: http://www.fantasystrike.com/forums/index.php?threads/helpful-posts-for-yomi-players.5328/
Here's another strategy post: http://boardgamegeek.com/article/7989656
GREAT analysis identifying the dominant tradeoffs of match length then using data to decide.
Dave, I did not mean to say that it was "all random", or completely random. However, a game with a higher level of randomness would bring the win%s all closer to 50%, wouldn't it? I'm not trying to downplay the intense amount of work that went into balancing the game, nor the amount of strategy that exists in playing the game - those are both obvious. I'm merely saying that because the game does have a high amount of variance, doesn't that help a lot?
Are there any 1on1 competitive games that have more randomness than Yomi? Outside of CCGs, I can't think of any.
It actually seems like about the least random competitive card game that I can think of. Hence the ability of top players to have such high win-rates.
Would it be possible to post statistics on character-specific match-ups?
It'd be really interesting to see if certain match-ups favor a certain character heavily or if it still evens out to about 50%.
I would be most interested in Lum vs Rook and DeGrey vs Jaina.
Keith Burgun, you are confusing two things here. One thing is character balance. A high level of character balance will have win rates close to 50%.
Another thing is randomness. Randomness has NOTHING to do with character balance. Randomness is related to the chance that a skilled player will win over an unskilled player. Because these are aggregated matchup figures, we can draw zero information about randomness from this data.
"Are there any 1on1 competitive games that have more randomness than Yomi? Outside of CCGs, I can't think of any." - You should try looking at the win rates of the top players on leaderboards. This is actually a really insulting thing you said, but I forgive you since it was based on the bad logic thing I explained above, haha.
Garcia1000: Hmm. It seems you are correct.
My apologies, I really didn't mean to be insulting. I'm a huge fan of Yomi, I've done a lot to promote it (created the Facebook group, wrote a Destructoid article that got it a lot of attention, as well as having local Yomi tournaments) and I enjoy the game (just got done playing 4 games of it).
The fact is I've been struggling with the level of randomness that is there for a long time, despite enjoying the game immensely. In general I have a somewhat anti-random philosophy about games, and so I'm sensitive to things that are totally out of the player's control. However I do totally get the idea behind Yomi and I generally think it overrides the randomness.
Keep up the good work.
Hi Keith Burgun,
That's okay! I didn't think it was insulting but it was possible for the words to be misinterpreted by others, that's all.
I also agree that sometimes there is a bit too much randomness in Yomi. These occur generally when both players are using high-variance characters. For example Degrey vs Setsuki is like a bunch of laughs or something. On the other hand, something like Grave vs Geiger has very low randomness. I guess it's acceptable since if you are playing those high variance characters you would be accepting of the randomness.
You might enjoy Yomi a lot more if you focus on the low variance characters. I also focus on the low variance guys and it has increased my enjoyment tremendously!
Im just posting to tell you that Tom Vasel did a review on the Puzzle Strike Upgrade Pack, at http://youtu.be/iUKtXIPSqgI
Wow! This is fantastic. I haven't been as active as I used to be but all the changes and everything are great! This is phenomenal. Great job Sirlin.