A Decimal Is a Promise. Don’t Make One You Can’t Keep.
The number and your confidence are two different things.
Here are two ways to tell you the same forecast. One: “81.4 wins.” Two: “somewhere around 76 to 86, best guess 81.” The first sounds sharper, more expert, more like the person has done the work. The second is the one telling the truth.
A number carries a hidden promise about how sure you are — its digits are the confidence claim (that idea has its own primer, Precision vs. Accuracy). “81.4” promises you can pin the season to a tenth of a win. Nobody can. So the whole skill of communicating uncertainty comes down to one habit: match the digits to your confidence. If you only know the win total within five games either way, you say “mid-80s,” not 81.4. The range is not you being wishy-washy. The range is the actual answer.
Two ways to lie about what you know.
There are two opposite failures, and honest forecasting threads between them.
The first is false precision — over-claiming. The championship odds reported as 23.6%. The live win-probability that ticks to the tenth of a percent. The projection of 81.4 wins. Each is a quantity that is genuinely uncertain, dressed up in the costume of a quantity that is exactly known. It feels rigorous. It is the opposite: it is hiding the uncertainty instead of reporting it.
The second failure is the one people reach for when they get scolded about the first: useless hedging — under-claiming. “Anything can happen.” “It’s a coin flip.” “On any given Sunday.” Refusing to commit sounds humble, but when a game is not a coin flip, calling it one throws away real information. A forecast that says nothing cannot be wrong, which is exactly why it is worthless.
The honest middle is a best guess plus an honest range: “76 to 86, most likely 81.” “About a one-in-four chance.” “A slight favorite, maybe 60-40.” Each says two things at once — what you believe, and how firmly. That second half is not an apology attached to the forecast. It is the forecast. A projection reported without its uncertainty is not a brave, clean number; it is an unfinished one.
The full treatment: where the range comes from.
How uncertainty is actually measured — confidence and credible intervals, standard errors, the width of what you can honestly claim — is developed in The Sports Page’s companion statistics textbook, a free, open, graduate-level text, with the R code to compute a range for yourself. Read it free here — the same “read, play, learn” idea, one rung deeper.
Where this concept shows up in The Sports Page
- Precision vs. Accuracy (Concept No. 7) — the sibling idea: the digits you show are a claim about your confidence.
- Sunday Editions — predictions are graded against ranges where possible, not single point estimates.
- The “acceleration” issue — the whole warning was that a metric’s value came with a wide, window-dependent uncertainty.
- Any Bayesian recovery piece — the answer is a spread of outcomes, reported as percentiles, not a single date.