zeeke wrote: Wow Northern, just wow. On Loki's power its doesn't change place on 32 tiles, it swaps up to 32 pairs of colored tiles. Maybe that is what's bugging because those numbers actually look quite high as you say. /z
vudu3 wrote: What's the definition of a cascade? Does a single match-3 count as a cascade?
CrookedKnight wrote: The part where the odds of a cascade actually drop slightly when you level up Magneto's blue from 1 to 3 (and then go back up from 3 to 5) kind of blows my mind.
Red Panda wrote: Do you have any results on devil Dino that would lead to recommending 3, 4 or 5 purple covers? Mine is currently 3/5/5 but I have some covers expiring in a couple days
daibar wrote: Can you do a simulation of one coloured tile being destroyed? This would fit in with goons and many countdown abilities. I'm sure after the last heroic, people would be interested in Unstoppable Crash results.
NorthernPolarity wrote: Red Panda wrote: Do you have any results on devil Dino that would lead to recommending 3, 4 or 5 purple covers? Mine is currently 3/5/5 but I have some covers expiring in a couple days Done.
vudu3 wrote: 3 purple - 22.3% chance of cascade 4 purple - 27.8% chance of cascade 5 purple - 34.8% chance of cascade As expected, doesn't seem like going with 5 in purple is worth nerfing one--or both--of his other abilities.
mischiefmaker wrote: vudu3 wrote: 3 purple - 22.3% chance of cascade 4 purple - 27.8% chance of cascade 5 purple - 34.8% chance of cascade As expected, doesn't seem like going with 5 in purple is worth nerfing one--or both--of his other abilities. Level 3 average tiles destroyed: 1.237676 Level 4 average tiles destroyed: 1.598566 Level 5 average tiles destroyed: 2.121538 I'm not so sure I agree. L3 to L5 purple makes getting any match at all over 50% more likely, and it gives you nearly 1 more AP on average, which is a pretty big deal for an ability you'll cast 2-4 times on offense and maybe twice that on defense. Also consider opportunity cost -- purple is getting better, with Mystique and Blade in the mix, but in general you'd rather cast almost any green, and usually you'll have a better red, too. Sure, the damage difference for even L4 to L5 green/red is big (about 1k), so it's definitely not cut-and-dried, but I'm running mine 4/4/5 and have no plans to respec him (already tossed a green cover, in fact). I feel like Dino's main utility is defensive scarecrow, and if you're looking to get defensive wins, I think your best bet is to increase the chance the AI gets a lucky cascade or four.
Disembiggen wrote: Since a cascade has to destroy at least 3 tiles, I'm assuming that the "average tiles destroyed" is the average number of tiles destroyed in all simulations. However, I think it might also be helpful for people to see the average number of tiles destroyed in only the situations where a cascade occurs. Based on the simulation results you posted, I believe these are those numbers. Please correct me if I'm wrong.
ark123 wrote: Would it be possible to do this like: % chance of destroying 3 tiles % chance of destroying 6 tiles etc? Because the magic behind cmags red and xforce's black is that they cause multi-level cascades. I mean obviously the first match after the skill is used is already very useful but I'd love to see something like that. Maybe a google docs spreadsheet if it becomes too long?
NorthernPolarity wrote: Disembiggen wrote: Since a cascade has to destroy at least 3 tiles, I'm assuming that the "average tiles destroyed" is the average number of tiles destroyed in all simulations. However, I think it might also be helpful for people to see the average number of tiles destroyed in only the situations where a cascade occurs. Based on the simulation results you posted, I believe these are those numbers. Please correct me if I'm wrong. Your assumption is correct, but your numbers make me think that something is wrong with the simulation... (mjolnirs destroying 6 tiles on average when it actually does things? whaaa?) After programming in the rest of the current batch of characters I'm working on, I think I'll stop programming new abilities and start writing tests to make sure that what I'm doing is correct.