Forecast a Date, Then Grade It

Date mode returns percentiles rather than a single day, so the answer carries its own uncertainty. When the event resolves you can score the forecast against where the outcome landed in that distribution.

This one has resolved.

The question

Run it

import asyncio

from pandas import DataFrame
from futuresearch.ops import forecast


async def main():
    result = await forecast(
        input=DataFrame([
            {
                "question": "When will Waymo charge a member of the public for a ride in the Ojai?",
                "resolution_criteria": (
                    "The first day Waymo charges normal public fares for rides in the "
                    "Ojai, in any market. Free Trusted Tester rides do not count."
                ),
            },
        ]),
        forecast_type="date",
        output_field="first_paid_ride",
        effort_level="HIGH",
    )
    print(
        result.data[
            ["first_paid_ride_p10", "first_paid_ride_p50", "first_paid_ride_p90", "rationale"]
        ]
    )


asyncio.run(main())

Results

Waymo began charging for Ojai rides in Phoenix on July 29, 2026, nine days after the p10 and about two months ahead of the median. The rationale had the mechanism right, a slip against Waymo's summer messaging followed by fast monetization once the Trusted Tester phase stabilized, and underweighted the state split: Arizona imposes no fare-authorization requirement, while California's CPUC advice-letter suspension was still in force.

Going deeper

• Long-form analysis: Waymo Profitability Forecast: Rides, Margins, and Losses Through 2027
• Published output: the first paid public Ojai ride date
• Track record: evals.futuresearch.ai