If I understand the nature of the math correctly, altruism should be "beneficial" (from evolutionary perspective), if the number of people multiplied by closeness factor of their genes to yours ultimately translates to a larger portion of your genes propagating to later generations, than if you have expended those resources on yourself.

E.g. suppose I have a loaf of lembas. I can eat it myself, and it increases my chance of surviving and reproducing merely by 10%, because, while I'm starving, I'm healthy. But I have three siblings who share 50% of their genes with me, and who are starving and sick. If I give all my lembas to them, their chances of surviving and reproducing increase twofold, and become same as my current chance. In that case, it makes sense for me (or rather my genes) to evolve the behavior whereby I give away the lembas, because these three survivors will spread the genes that we share more efficiently than I alone.

Now, obviously, the math here has to add up in real-world scenarios - is that what you mean by "it wasn't trivial to stack the dice"? It just feels like it shouldn't be hard to come up with many cases where all the coefficients line up...