But it is happening more like 10 games out of 40, that is why I am wondering if there is a way to improve my mulligan rng if skill is already capped and deck is already crafted with hand/draw/mulligan deeply in mind?
Without providing a large data sample, I can only guess that you have confirmation bias.
1 - (25/40 * 24/39 * … * 20/35 * 22/37)
= 1- [(25! ÷ 19!) ÷ (40! ÷ 34!) * 22/37]
= 1- [34! * 25! ÷ (40! * 19!) * 22/37]
≈ 2.74%, or about 1 in 36.45 games. Roughly 2 in 73.
Those are kinda two separate issues.
On the one hand, any sample where you aren’t diligent enough to use software to record or keep very detailed notes is prone to you remembering things wrong and making incorrect estimates. Confirmation bias definitely can play a role there. But that’s not particularly a “sample size issue,” sample could be small or big, it doesn’t really matter.
Now if you are taking diligent care to make sure that your data is accurate, then a sample size of 40 in a CCG has a standard error of approximately 8%. So if something happens 10 out of 40 times, then that’d be 25% ±8%, so you should be confident that reality is somewhere between 17% and 33%.
However, in this particular case Cuddles has a history of trolling with lies to create engagement. He’s created a thread saying that he just came back from an 8 year break a single day after making a thread about some other issue. So when he says that 15 out of 40 one-drops are evading his mulligan 25% of the time, it’s not confirmation bias, it’s not sample size, it’s outright lying for attention. You were fished in.