Social Security is designed to be actuarily neutral. Actuarily speaking the increase in monthly benefits you get from waiting to start your benefit is expected to equal the amount you'd gain from getting more checks as an early benefit recipient. So the real question is whether you expect to outlive your actuarily expected lifetime. If you're retiring in reasonably good health and your parents lived into their 90s, you're better off waiting. On the other hand, if you expect to die early, you'd want to take the benefit as early as you can.
If you're going to plug numbers into a calculator, using the expected returns of an equity portfolio is problematic because you're comparing an essentially zero investment risk option against a risky option. The fair comparison would be something like treasury bonds that are also backed by the government (and which will yield a lot less than 8%).
It's also problematic to take into account the annual COLA adjustment like the calculator you linked to does. If you're comparing two different cash flows over time, you'd want to compare their present values which means that you have to discount cash received in the future to account for inflation-- $1 in your hand today is more valuable than $1 in your hand in 10 years. Since the COLA adjustment is designed to track inflation, if you inflate future payouts using the COLA, you'd have to discount them back with the inflation rate. Since COLA is supposed to equal inflation, though, the two terms cancel out and you should just consider the un-adjusted cash flows for a fair comparison.
That being said, while basing the decision on whether you expect to outlive actuarial projections does create some risks that aren't symmetrical. If you think you're going to die relatively young but you get lucky and live into your 90s, you'll lose out on a lot of Social Security cash. If that also means you outlive your other investments, that could be very problematic. On the other hand, if you think you're going to make it to 100 but get hit by a bus the day after you claim, you're not going to be around to mourn the loss of a few years of checks and your heirs probably won't care too much that you didn't have a pile of Social Security cash to pass on. Personally, that would make me lean toward assuming you'll outlive the projections and taking Social Security later unless you're really sure that you won't live to the breakeven point (which should be somewhere in your late 70's or early 80's). You could mitigate some of those asymmetric risks by using your other investments to buy an immediate annuity that would pay out however long you live in order to ensure that you weren't going to outlive your investments if you wanted to take Social Security somewhat early.
And, of course, all this assumes that you're single or that your partner would be claiming on their record rather than on your record. If your partner is going to use your record to claim her Social Security, then you'd need to consider the joint probability of dying to figure out the breakeven point. And it assumes that Social Security continues to pay out at the full rate even after the trust fund is exhausted which is projected to happen in 2032-2034 depending on the assumptions used.
Here's a good article on calculating the break-even point (though I do find it annoying that they say not to include COLA adjustments but don't attempt to explain why).