When You Will Die

With absolute certainty, you will die. When will it happen? That is a trickier question. But we can use what we do know and run simulations to get a more accurate and more meaningful answer than average life expectancy.

In the chart below, each ball   represents a simulated life. It moves along the curve, which represents the chances that you live to the next year. The curve changes by age and sex, always falling off later in life. A ball drops when the end of life arrives.

Adjust the settings to match your age and sex.

Running Simulations

If you are -- and -- years old in 2025, then there is a -- chance that you will die within a year. More optimistically, in a glass-half-full kind of way, that is a -- chance you will live to the next year (100% - chances of dying in a year). If you live, then you turn -- years old and have a -- chance that you live through the next year, and so on. We repeat until the simulation does not make it. The age at that time is a simulated age of death.

Restart at your current age and repeat.

This is a Markov Chain Monte Carlo method, also known as MCMC. Each step in a chain of events is determined by the previous step.

In this example, we step through your age and each year after based on the outcome of the previous year. You either live or die. The chain ends when you die.

Running simulations with this data lets you:

So far, we have run 0 simulations.

When you are younger, chances are high that you live to the next year. Probabilities stay above 99%, which is pretty good. Then we age into our 60s and things change.

Where the Data Comes From

When you read about life expectancy, it usually refers to an average age of death during a given year, starting from birth.

For example, the Centers for Disease Control and Prevention estimates life expectancy at birth, based on 2022 mortality, is 77.5 years for the total U.S. population. To get this number, they counted deaths at each age in 2022 and calculated the average.

This is called period life expectancy. It is based on a fixed range of time, which in this case is mortality during one year. It is useful for comparing conditions in a year against another but is less useful for estimating how long you will live in future years. Health and medicine tend to improve.

Cohort life expectancy is a more accurate estimate for how long you will live. Count the age of death for everyone born in a given year. Then calculate the average.

For example, you could look at everyone who was born in 1920, tabulate the age of when each person died, and you would have an accurate picture of mortality for those born in 1920.

The challenge is that you have to wait for everyone in a cohort to pass before you can get the full counts. Since we cannot wait around for a century to run these simulations, the Social Security Administration projects future years based on past and current data.

Look closely at the 1910 curve versus the 2025 curve. The former is jagged, because it is based on actual deaths, whereas the latter is smooth and projected for later years.

Now we can get into the heart of the question. When will you die?

With cohort life expectancy, we gain better estimates for when you might die. The probabilities are different for male and female, and the curves shift based on when you were born, so set your information above if you have not already.

Below are the results of running 0 simulations so far for a who is years old. Choose to “live fast” to speed up the process. Notice the distribution shifts and how the median, or the halfway point, eventually converges to a certain age in the following chart.

After 0 simulations...

The median age of death, currently at 77 years old, marks when half of the simulations ended at this age or later, and the remainder occurred before.

These are your chances for how much time you have left:

The oldest known American lived to 119 years old in 1999. The longest running simulation so far went to years old.

Coming from the other side, the youngest age of death so far is years old, or years from now.

As you can see, there are many possibilities in your future and nothing is guaranteed, except that you will die eventually. This makes it difficult to say when you will die, exactly.

If I had to guess a specific age — based on limited data and without knowing your family history or daily habits — I would go with the most common age of death, the mode, in these simulations.

Ready? You will die at --.

Yep.

But who really knows? I certainly don’t.

Life is random, the human body is fragile, and a single event can change everything.

We can only say what is most likely to happen rather than what will happen. Catastrophe might strike tomorrow. You might die at 120 years old and live in the record books for all time. Most likely, reality falls somewhere in between.

These estimates are useful for planning and making long-term life decisions, such as when you can retire or when you should modify your health plan.

As for the everyday, it is impossible to know the future with absolute certainty.

Also, when our last day comes, we do not get to throw another ball into the life simulator, despite what movies tell you.

Today is a day. Let’s make it a good one.

Check the following resources for more on this finite life.

Questions or comments? Email me (Nathan Yau) or find me on the socials @flowingdata. Sign up for the FlowingData newsletter for updates.