Forecast rounds: one replayable artifact per forecast

A forecast round binds everything that produced a forecast – the model and its calibration, the data vintage, the filtration, the forecast with its conditions and judgment – into one object that can be saved, reloaded, re-run, and compared with later rounds via compare_rounds(). This is the object an institution archives: next quarter, "why did the forecast move?" is answered from the rounds, not from memory.

Usage

qpm_round(
  name,
  model,
  data,
  observables = NULL,
  horizon = 12,
  bands = c(0.5, 0.7, 0.9),
  measurement_error = 0,
  kappa = 1e+06
)

Arguments

name: Round name, e.g. "2026-Q3 September".
model: A qpm_model.
data: Data frame in levels with a period column; passed to qpm_filter(). Quarter labels ("2026-Q3" style) are required for cross-round comparison.
observables: Columns treated as observed (default: all columns matching declared variables).
horizon: Forecast horizon in quarters.
bands: Fan coverage levels.
measurement_error, kappa: Passed to qpm_filter().

Value

An object of class qpm_round with elements model, data, solution, fit, and forecast.

Details

qpm_round() runs the standard pipeline (solve, filter, baseline forecast). Conditions, scenarios, and judgment are then applied to the round directly: qpm_condition(), qpm_scenario(), and add_judgment() all accept a round and update its forecast.

Examples

m <- qpm_template("bkl")
sol <- qpm_solve(m)
obs <- simulate(sol, nsim = 40, seed = 1, burn = 20)
obs$period <- next_quarters("2016-Q1", 40)
r <- qpm_round("test round", m, obs[, c("period", "pi", "i", "q")],
               horizon = 8)
r
#> <qpm_round> test round
#>   created 2026-09-19 07:26 - qpmR 1.1.0
#>   model: Canonical small open economy QPM (BKL, stationary trends) - 25 parameters
#>   data: 2016-Q2 ... 2026-Q1 (40 quarters) - observables: pi, i, q
#>   filter: log-likelihood -195.15
#>   forecast: 8 quarters (2026-Q2 ... 2028-Q1) - 0 conditions, 0 judgment entries