Tile drop predictions

dfields3710

Hi, the reason I created this thread is to ask anyone were they able to predict any of the enemy cascades. I know I did a number of times.

For example, Tank Characters

Have anyone noticed that your tanks take most of the enemy cascades, just to get them down quicker. Like the other day I was using Ares to help my climb in PvP. I noticed when I used a character that was used straight for spammable damage(Ares, johnny storm, wolverine) they took the most damage. They were the ones that fell victim to massive cascades leading to a third of their health gone in one turn.

Also I noticed in certain situations that I was able to predict what color tile(s) would fall to help the ai out in gaining more ap, dealing more damage, and making you use more health packs.

I'm not ranting, its just crazy how random drop tiles are supposed to be random, but they're more predictable than random.

Feel the same way. Comment

Thx for reading and commenting.

Unknown

This effect is called Confirmation Bias.

In any given match, you're going to see a very large number of tiles generated. Some of them will fit your prediction ("hits"), some of them won't ("misses"). The human brain discards the misses and focuses on the hits, confirming the prediction even though it's happening at a rate equal to chance.

Unknown

You can't predict tile drops but with experience you can often predict which kind of drop is likely favorable or not. For example:






Making the yellow vertical match 3 is a very bad idea. It is not Confirmation Bias as if you constantly make this move it will end up as a match 5 for your opponent a surprising number of times. So some of what seems like bias is actually just very bad/good moves that may not be obvious to you.

scottee

Phantron wrote:

You can't predict tile drops but with experience you can often predict which kind of drop is likely favorable or not. For example:






Making the yellow vertical match 3 is a very bad idea. It is not Confirmation Bias as if you constantly make this move it will end up as a match 5 for your opponent a surprising number of times. So some of what seems like bias is actually just very bad/good moves that may not be obvious to you.

Yeah, people need to calculate odds of stuff like this more often. It gives you a match-5 1/7 of the time, but sets up the AI for a match-5 almost 2/7 of the time.

Most of people complaining at cascades is them not being able to look ahead and see cascades themselves. A lot of these times, the AI is making a cascade you could have made but just didn't see. Once you get good at reducing this, it's actually a lot easier to control what matches are available to the AI and prevent all cascades on the board, leaving the only possibility being those that drop from the ceiling. And as Phantron noted, you can reduce the likelihood on even those.

Unknown

Based on what D3 said I don't think the AI particularly tries to make good cascades but it never misses a match 4/5 if one is available, while it is quite possible to miss one as a player if you're in a hurry. I also had a lot of time where I saw say a green match 4 and didn't even notice there was another red match 5 elsewhere which should've been a no brainer to take first because I was tunnel visioned. From what I can tell, certain dangerous positions (like the one I have above) seems quite attractive to a human, probably because they're thinking 'oh if a red falls in the middle I get a match 5!' without realizing it's twice as likely for the red to fall above or below. The AI, of course, is not tempted by such things because it doesn't care.

In general, if you see a position like above that looks dangerous, you really should defuse it because that move is bad for you whether you made it or the AI made it. If you made it, you're giving the AI 2 to 1 odds compared to you of making a match 5. But if the AI takes it? If you get the red match 5 it's probably another 3000 damage for you, which is helpful but hardly game deciding against a tough opponent. If the AI gets a red match 5, given their 5/5/5 covers in PvE that usually means someone is going to get downed. The tradeoff here is greatly in favor of the AI that even at 1 to 2 odds, they'd likely come out ahead if they made the same move simply because with their inflated levels you still come out behind, so you want to break that up and not give the AI a chance to hit a red match 5 that will likely immediately result someone being downed. Even in PvP, it's not a scenario you want to tempt very often because your opponent at least always starts with 100% health compared to you, which can be quite a significant advantage now due to the tightened MMR, and of course losing a fight can easily kill your entire playing session so the risk vs reward can still make that move favorable for the AI even at 2 to 1 odds.

Unknown

The odds are actually much less than 2 to 1, but still in favor of the AI more than 3 to 2.
Rough Odds of getting the red crit: 1/7 direct red match
1/49 vertical black match, 5/343 other vertical match. These can't happen in conjunction with the red match. Total 12/343. Ignoring the extra effects of a column clear and other horizontal matches, we use geometric series to give a rough total of (1/7)/(1-12/343) = (1/7)*(343/331) = 49/331
~ 14.80% for you

Rough Odds of AI getting the match: of the 3 vertical tiles, center is not red. There are 48 possibilities for the tiles above and below; they both can't match the center tile. 13/48. In 7/48 of all possibilities, the 3rd tile in the column matches the 4th, and 1/6 of those times they'll match the 2nd. 7/(48*6) = 7/288. This includes (1/48) * (1/6) = 1/288 of the possibilities where red is on top.
[(13*6-1)/288]/[288-7/288] = 77/281

77/281 * (1- 49/331) = 77/281 * 282/331 =
~ 23.35% for the AI

1.58 to 1 in favor of the AI.