Runs Scored in World Baseball Classic

The piece explains a forecast of total runs in the 2026 World Baseball Classic knockout stage, where seven games were expected. A simple estimate from four previous tournaments gives 8.7 runs per game and about 61 runs overall. Because that estimate rests on very little data, the author widens the initial 90% range from 43-79 to 35-87 runs to account for uncertainty in the estimated variation. The community forecast was just under 70 runs; the author’s own forecast was 60, and the actual total was 59.

The more detailed approach uses the structure of the data: runs scored by each team in each game. Across the available tournaments, those team-game counts average 4.3 runs but have variance 10, more than twice the mean. That excess variation indicates runs cluster rather than occurring as independent Poisson events, so the author models them with a negative binomial distribution, fitted with mean 4.3 and parameter r = 3.4. The explanation shows how this distribution ranges from Poisson-like counts to more variable, heavy-tailed counts. Since seven games produce 14 team scores, the model’s summation property gives a distribution for the knockout-stage total by adding the means and r parameters. The discussion also offers an intuitive coin-flip analogy for overdispersion, while noting that its imagined baseball mechanism is not needed for the forecast.