This post is written by Mat, our gameplay designer.
Hello again!
As Mia is preparing the modifiers for our extended Hardmode, I had a day or two to focus on some minor balancing. So I wanted to share with you a bit of the telemetry data we’ve been gathering and how we go about noticing what needs a bump or a nerf.
So, one thing I was a bit concerned with was how powerful each enemy was in comparison to each other. Is a Knave as dangerous as a Pillager? Is a Thorntoad comparable to a Basilisk? In the search for that answer, we could look at how much damage each enemy deals overall, but that can be skewed by how many of each unit there is. So, knowing it isn’t a perfect data set either, I wanted to ask "how many kills does each enemy have?", starting with the bandits.
In the latest 720 kills performed by a human bandit, these were the top 5:
Raider: 187
Rascal: 70
Looter: 68
Outlaw: 58
Brigand: 56
Now, besides the subtle prevalence of ranged bandits, I was very surprised to see that the Raider had SO many more kills. Surely, it meant that the Raider enemy was the most powerful of all the bandits!
Well, not quite. After noticing nothing immediately wrong, I saw that he’s a user of the Prepare Bomb skill, one of those delayed AoE effects. And, as intended, this sometimes ends with a friendly-fire kill, especially if the player has some knockback effects, and those are counted in the telemetry! Following that thread, I eventually discovered that the Raider made no attempts at avoiding his friends. Oh boy. A small tweak to the AI, and we’ll see the results in a few weeks!
In the Abomination department, the story was a bit different.
In the latest 3490 kills performed by an Abomination, these were the top 5:
Mandrake: 1308
Thorntoad: 494
Behemoth: 425
Redcap: 384
Wyvern: 325
A similarly scary jump. Now, in this case, it's not so clear cut - Mandrake’s delayed AoE, Barrage, does not do friendly fire, and Mandrake is pretty much as common as a Redcap or Wyvern, even though it has pretty much double their killshots combined. What gives?
Well, this is where you gotta do a bit of a leap of faith. You see, It’s normal to expect a ranged abomination with a decent power stat to be the top killer, but I felt the numbers to be a bit high compared to its allies. Then I remembered that we did a recent change where its Quicken ability enabled Mandrake to shoot twice in a single turn. A bit of a needed power boost, sure, but I don’t think it allowed for much counterplay.
So I took the opportunity to remove that capacity and instead add evasion to the same ability! Still a buff from the previous version, but now less murderous. Again, we’ll see the results in a few weeks, eh?
That sort of stuff happened all over the place. Charge, Parry and Take Aim were hardly picked (especially in builds that ended up winning the game), so they got a bit of a buff. Sieges in zone 1 are pretty lethal, but I didn’t touch them until we have more information on the Mandrake change. And, finally, take a look at the hero picks:
In the latest 5325 Hero picks, these are their rankings, rounded:
Breunor: 1506 (28% of all parties had him)
Merewen: 1427 (26%)
Morgan: 1334 (25%)
Rana: 241 (5%)
Salim: 205 (4%)
Isabel: 147 (3%)
Wilfred: 114 (2%)
Zahra: 94 (2%)
Safir: 86 (2%)
Tariq: 83 (2%)
Rumayla: 70 (1%)
Alain: 18 (<1%)
Now, the biggest input point is, of course, availability. Breunor, Merewen and Morgan are the default heroes, so it’s safe to assume most games feature at least one of them, perhaps with a small indication that Breunor is a bit ahead of the rest of the Warriors. Rumayla and Tariq are a bit harder to gauge, but they are also a bit unorthodox, so that (maybe) explains it. Nothing too scary there, but worth keeping an eye out.
Now, Alain? I know, he’s the hardest character to unlock, requiring you to reach party rank 10. But the size of the jump surprised me a bit. Here’s where internal feedback comes in: when we first submitted Alain, even our dev team felt he was cool, but maybe a bit weak, with an odd triggering mechanism for his passive. With that in mind, I decided to buff his passive a bit, and see what happens with those numbers.
Not an exact science, and you gotta be careful to keep them feeling different, you see.
So, yeah, that’s pretty much the process. A lot of parsing numbers, coming up with hypotheses, and testing them. Sometimes it's as easy as reducing the cooldown of a supporting skill or tweaking the AI of a unit, other times you just gotta go with your gut. Hopefully, the more iteration cycles, the better the result!
My next stop is spells. Did you know that 47% of all spells cast in the game are Restoration? Oof.
Any thoughts on that or any of these changes? Think you noticed something different? Drop us a line on our Discord server and join us in making The Hand of Merlin as good as it can be!
Mat