Sled Team uses a method called Monte Carlo simulation to forecast when your work will be done. For each task on your list, we take your past work data--everything you finished super quickly, everything that was a tedious slog. Then, we essentially roll a bunch of dice against that data, to see how long you'd take to get all your work done in one specific timeline.

Rolling those dice one time wouldn't give you a very useful forecast, though--that would just be one timeline, where maybe you had a bunch of really great days in a row, or maybe a bunch of really terrible ones. We can't trust that forecast to be representative of reality.

We deal with that by rolling all those dice once, writing down the number we get, then rolling them again, and again, and again. That gives us a date forecast that's much better than a simple average, because we can adjust it by the confidence level you need, and because we can account for the fact that a delayed project might not just be a little delayed. In the real world, when projects get stuck, they might not just be five or ten percent late, they might be 300% late, or even worse.

So the more simulations we run, the better we can feel about our forecasts. However. The more times we run them, the slower your computer is going to run. There's a balance to be struck here between precision and performance, and eventually you reach a point of diminishing returns.

All of this is a very nerdy way to say that we've recently upgraded our code to run ten thousand simulations at a time, covering every task in your list. We think that gives you really good precision, without making Sled Team sluggish. As computers get faster, we'll periodically reevaluate and see if it's helpful to to raise or lower that number.

Thanks for coming with me on a look into the guts of how we do forecasting! And may the odds be ever in your favor.

—Evan