Record checked October 21, 2026

Evaluation

We compare simulated answers with published human results and preserve the errors.

Current public measures

3.72 points MAE

500 historical questions, 50 states and 200 simulated respondents per question. September 1, 2026 run. Calibrated retrospective evaluation; the score is not held out or cross-validated.

Selection note and calculation

2.95% MAPE

As of October 19, 2026, three graded spending outcomes and four forecasts. Both NRF forecasts are averaged within their outcome; the three outcomes receive equal weight. The published persistence reference is 2.09% MAPE on the same weights.

Included outcomes and calculation

Not yet shown

These headline scores do not establish subgroup validity, future behavior, open-ended text quality or performance on custom customer audiences. We evaluate the question, audience and instrument before interpreting a result.

Figure 1 · Public forecast record

Published forecasts and outcomes

Download CSV

Measure	Forecast	Published outcome	Grade
NRF back-to-school spending Issue 01 · July 7, 2026	$856.83 / $870.37	$863.86	Win
Michigan consumer sentiment Issue 05 · September 9, 2026	55.4 index points	47.8 index points	Miss
NY Fed household spending growth Issue 06 · September 15, 2026	4.9%	4.704714298%	Partial
Deloitte holiday spending Issue 07 · October 1, 2026	$1,589	$1,654	Partial

Numeric headline outcomes in the Report as checked October 21, 2026. Units differ across rows. Partial was added on October 18 after grading; the original misses for Issues 06 and 07 remain visible. The categorical Issue 02 win and three awaiting grades are listed in the full record. Values and grading rules are linked by row.

Earlier evaluations

April 2026 · 7.47 points MAE. Earlier five-place study: 404 non-electoral questions and 10,000 simulated respondents per question. This is preserved published history; the surviving legacy export cannot reproduce the complete summary.

Post-freeze check · 9.97 points MAE. Fourteen scored questions from a 22-question batch after exclusions. The study design differs from the current retrospective benchmark.

Read the dated historical methods