But it makes sense, someone who is pretty healthy and active at an age will statistically live longer than someone the same age who had serious health problems. So average life expectancies that don't take into account the actual health state of an individual could be really off.

Exactly.

Northwest Mutual actually has a lifespan calculator that looks at risk factors like family history, smoking, exercise, recreational drugs, use of seatbelts, medical checkups, etc., and they give Buffett an expected age of death of 103, towards the top of my 5-10 year guesstimate. media.nmfn.com: Lifespan Calculator – Test Your Life Expectancy || NM

However, they don't don't explain how they do the calculation,, and most importantly, they don't ask what appears to me to be the most important question: what current medical conditions do you have, such as cancer, heart failure, Alzheimer's/Parkinson/ALS, etc. A fairly high percentage of people at 94 have known health conditions that mean they will die within the next year or two. For instance, if you take ten 94 year-olds, with 3 of them like Buffett (no known medical conditions), but 2 with Alzheimer's, 2 with metastatic cancer, 2 with severe COPD and another with congestive heart failure, you might have the first 3 with 10 year life expectancy, and the other 7 with 2 year life expectancy, and if that is typical for 94 year-old males, then your population mortality rates and life table will tell you that the average life expectancy is (3*10+7*2)/10 = 4.4 years. The key point is that the average includes people that are much sicker than Buffett is today.

dtb