Fantasy Strike is an interesting case study in game balance for an asymmetric game. I did the entire balancing process without any data at all, and now, after it’s all done, we have the data. Meaning, we were collecting data, I just didn’t have a way to actually look at it until now, after the whole balancing process is basically complete.

Relying on Experts

I find it much easier to balance a competitive game by going off the opinions of experts than looking at data anyway. Experts can get to the bottom of issues more quickly, and with a much smaller sample size than a data-driven approach needs in order to be good enough quality to rise above statistical noise.

In some games I’ve worked on, I’m one of the experts (Fantasy Strike, Street Fighter HD Remix, Puzzle Fighter HD Remix). In other games, I’m not at all (Yomi, Puzzle Strike, Kongai) so I rely on players who have expert knowledge about balance. I’ve found both ways can work just fine, it’s all a matter of knowing which experts to listen to. On any given issue, you can always find someone who says X and someone who says the opposite, so it’s pretty critical to be able to evaluate which arguments actually make sense, and factor in whose opinions on such complex things have turned out to be right a lot before.

Looking at the Data

Anyway, now that we have the data, how did it all turn out?

First, a reminder about what success even looks like. Like in any such game, it’s a struggle to keep every character within a reasonable power range versus every other. It’s normal to have 8-2 and 7-3 matchups (meaning, if two experts played 10 games in a specific character matchup, we expect character X to win 8 and character Y to win 2, etc). The question is how many of those matchups there are. I tried to give some perspective about what percentage of 7-3 and 8-2 or worse matchups we expect to find in these sorts of games in this post about Game Balance and Yomi.

In Yomi, the matchup chart at the time showed literally zero matchups of 7-3 or worse, which was incredible, and I was unable to even name another game with 10+ sides that has done that. It’s unreasonable to expect that of ANY asymmetric competitive game, and would be an impossible standard to hold any fighting game to.

But…we did it. Here’s the empirical matchup chart based on hundreds of thousands of games played over the last year:

I’ve highlighted the matchups of 7-3 or worse in red. Just kidding, there aren’t any.

You might think that an empirical matchup chart like that can hide balance problems because it’s including a bunch of low-skill players, and maybe when the top players play there’s really unfair stuff that no one can deal with. That’s a good point. It’s worth noting that an empirical matchup chart that includes ALL players is still worth looking at because that’s, you know, the experience of the playerbase overall. So it’s a relief to find out just how tightly balanced that chart is. But yes, a chart that’s just experts might look wildly different, so let’s look at just the “master league” players (as well as the top 20% of elo players in casual matches).

There still aren’t any 7-3 or worse matchups. It’s pretty similar to the other chart and a shockingly narrow range of balance to anyone who is familiar with these types of charts in other games.

If anything, I actually think some of these matchups are truly closer than the data shows. Two of the most lopsided matchups in the game according to that chart (reminder: none actually show as lopsided there since nothing is even 7-3) are Jaina vs Geiger and Lum vs Valerie.

Geiger is probably the #1 most losingest character against me personally, so I was curious what my own stats are. When filtering the above table of expert player results for just games that I personally played, I found that my Jaina vs expert Geigers is 6.7 - 3.3 in Jaina’s favor (compared to 3.4 - 6.6 the other way around). Admittedly a small sample size there, but I feel like it’s a case where more Jaina players just need to step it up.

I have more to say about Lum though. He is probably the most difficult character to play well. The largest number of different situations happen when Lum is on the playfield because his random sets of items create all sorts of opportunities to take advantage of things if you know how, improv, and act fast. It’s no surprise that even amongst the top 20% of players, they still don’t really play him well enough.

The chart above lists Lum vs Valerie as 3.7 - 6.3, yet if we only consider the matches I personally played as Lum against those same top players, the number becomes 6.75 - 3.25 in Lum’s favor. And unlike the previous example, it’s not a small sample size. I guess that would actually make it the most lopsided matchup in the entire game in the opposite direction that the chart shows, so it’s hard to know the real truth. It’s also interesting that players routinely call out that matchup as very very bad for Lum, yet I have often accepted challenges in it, win them, yet I haven’t changed anyone’s mind apparently.

The bottom line is that the above data shows 89% of matchups are in the 6-4 to 4-6 range, with ALL the remaining matchups just barely worse, and 0% in the 7-3 or worse range. And some of the ones that are barely worse seem likely to be within that 6-4 to 4-6 range too, ultimately. (Based on my play at least! Or wait, they’d still be least close matchups, just in the opposite direction? Well, who knows.)

Popularity

In Yomi, we found an amusing inverse correlation with character popularity and character win rate. Midori was the least popular, yet most winning character while Setsuki was the most popular, yet least winning character. And in Fantasy Strike…the same thing happened again, almost.

Here’s the Fantasy Strike results from highest win rate to lowest:

```
Midori   59%
```
```
Rook     54%
```
```
Lum      53%
```
```
Valerie  51%
```
```
Argagarg 50%
```
```
Grave    49%
```
```
DeGrey   49%
```
```
Geiger   47%
```
```
Setsuki  46%
```
```
Jaina    45%
```

Now here’s the chart by character popularity:

```
DeGrey     
```
```
Setsuki      
```
```
Valerie
```
```
Grave 
```
```
Jaina    
```
```
Rook   
```
```
Geiger   
```
```
Lum  
```
```
Argagarg    
```
```
Midori    
```

In the Fantasy Strike story world, Midori is a martial arts teacher, while Setsuki is a ninja student. It seems that in the games, masters play the master while students play the student.

If you’re interested in playing Fantasy Strike, it’s available on Switch, PlayStation 4, and Steam.
Join the Discord chat to find out about weekly tournaments or to ask for matches and strategy advice.