Why cycle predictions are estimates, and what makes them better
· 2 min read
No app can know when your next period will start. Here is what the number is actually based on.
Every cycle app shows a predicted date. None of them knows it. Understanding what the prediction is built from tells you how much weight to put on it.
What a prediction is made of
Almost all of them work from your recent cycle lengths — the gaps between the last few start dates — and project forward from your most recent period. Some weight recent cycles more heavily. That is the whole of it.
There is no test happening, no measurement of hormones, nothing being sensed. It is arithmetic on the dates you entered.
What makes it better
More cycles. One logged cycle gives an app nothing but a population average. Three or more give it *your* average, which is a different and much better thing.
Consistency. Logging the actual first day of bleeding, not the day you remembered to open the app.
Honesty about variability. If your cycles range from 25 to 34 days, no method produces a reliable single date. What a tracker should do is widen the window and say the confidence is lower — rather than print a date and let you assume it means something.
What makes it worse
Perimenopause, most of all: the cycle is genuinely changing, so recent history is a weaker guide to next month than it used to be. Also hormonal contraception, significant stress or weight change, illness, and long gaps in logging.
What to expect from a good app
It should tell you how confident it is, widen the range when your cycles are irregular rather than hiding it, and never present the fertile window as contraception.
Sathi predicts from your last two or three cycle gaps, states its confidence, and widens the band when you have told it your cycles are irregular. With one cycle logged it assumes an average and says so.