Ludia Forums

Critical Strikes - are the stats correct?

I have noticed I don’t get many critical strikes for dinosaurs that are supposed to have high critical strike percentages. I did a study (each dinosaur recorded at least 40 strikes) and here are my results for my highest critical strike creatures:

Erlidominus: crit stat 40%
in game: 32%

Ardentismaxima: crit stat 30%
in game: 23%

Thoradolosaur: crit stat 40%
in game: 22%

I’m pretty upset at the Thoradolosaur, especially after it got nerfed. Every dinosaur at the 20% stat and lower all were right around their respective stats, but these three were pretty far removed. A critical strike can make a break a game, especially in the higher trophy levels.

Anyone else having an issue getting enough critical strikes from your highest critical strike creatures?

just gotta say, 40 as a sample size is not enough to accurately gauge if the crit chance is hitting the mark.

5 Likes

It takes a lot of battles to get up to 40, trust me :wink:

The best number would be 100. but the more the merrier, @OrigamiRobot would be the best person to take that up with

50 works too.

The only good thing about 100 is it’s easy math. Think about it, if your sample size is only 100, the absolute minimum error you can get is 1%. That’s not great.

This claim interests me. The problem with any of these self studies is that you have to be extremely diligent in recording your findings. If you are lax at all, it introduces the possibility of memory biases. But if you were diligent in your data gathering and the sample sizes are close to the same, it is odd that only the creatures with the highest crit chances would be the ones with the highest margin of error.

A pro tip for anyone attempting to perform an experiment like this: record everything in great detail and provide those records when you present your findings.

1 Like

40 is a relatively small sample size. The smallest sample size is usually considered to be 30, so 40 is still on the smaller end. It should approximate towards 40, 30, and 40 respectively after hundreds of moves

Not really easy so much as it is efficient, and efficiency in math tends to cut down on errors, yielding more accurate data. Maybe they 1000 would be better, but you’d need a lot of time to get those numbers

It only cuts down on errors when you’re doing things in your head. No part of this should be done in one’s head.

40 is a decent sample size. Not the most, but decent enough.
And I agree, I constantly am getting irritated cuz my Magnapyritor doesn’t seem to critical ever. I went 13 strikes with no crit! Ouch. Yet Tenonto, who is also 20% seems to critical every third hit. And DioRaja at 30%, also crits less than some 20%.
I guess a bigger concern is, some lower percents seem to critical more than higher.
I have noticed this for awhile now. Glad someone else is noticing it also.
I havent done any research, so my observations mean nothing haha. Just observations. Stats might prove me wrong

tl;dr Observing a 22% crit rate from Thor is still pretty likely given a sample size of 40.

What is or isn’t a good sample size depends on the exact experiment. In this case, 40 is not. Here is a table showing the effects of different sample sizes. This table shows the Probability that the Experimental Crit Chance will be shown for the given Sample Size. Using the OP’s Thor example:

image

So what this is telling us is that if the sample size were 10, there is an 89.21% likelihood that we would observe a crit rate of 22% when the true crit rate is 40%. When we raise the sample size to 40, we reduce the likelihood of observing 22% to 9.28%. 9.28% is still pretty likely, all things considered. When we increase the Sample Size to 100, there’s only a 0.01% chance that we would observe a 22% crit rate.

Note that this table IS NOT showing us what a good sample size would be, it’s merely showing us the likelihood of a particular experimental result. To figure out what a good sample size would be, I’d have to rearrange these formulae or let someone more familiar with that particular question chime in.

Source: https://www.youtube.com/watch?v=oYeJBdCGwxk

1 Like

Doubt there is any good sample size at all in this case. Crits are working on long run, not on short runs. Basically, on long run it evens out at 40% or extremely close for Thor. Though it counts crits since they were added. So if your Thor critted with 90% rate two months ago, this crit rate needs to fall closer to 40%. So thats why sometimes there are no crits for longer period from high crit % creature.

Basically it takes a few days of playing the game to get 40 strikes from each one of your 8 creatures.

While I agree, 100 is a better sample size, the amount of time it would take to get there is just not something I’m interested in doing. Think about it, if you get 2 strikes in a game from a creature that’s good, 3 or 4 is rare at the higher levels. Plus, with randomized team selection across 8 creatures it takes a loooooong time to get enough strikes from every creature on your team.

Just saying, this isn’t easy. I kept track of everything with tally marks on some paper, so it’s not super advanced. But it’s accurate.