FutureSearch Logofuturesearch
  • Pricing
  • Research
  • Docs
  • Evals
  • Markets
  • Blog
  • Company
  • Try it for free
FutureSearch Logo

General inquiry? You can reach us at hello@futuresearch.ai.

Company

TeamCareersPressPrivacy PolicyTerms of Service

Developers

SDK DocsAPI ReferenceCase StudiesGitHubSupport

Integrations

Claude CodeCursorChatGPT CodexClaude.ai

Track Record

Trading ResultsAccuracy EvalsTournament Standings

Follow Us

X (Twitter)@dschwarz26LinkedIn
FutureSearchdocs
Frontier forecasting
Installation
  • All install methods
  • Claude.ai
  • Claude Code
  • Web App
  • Python SDK
  • Skill
Reference
  • API Key
  • forecast
  • decision
  • multi_agent
  • agent_map
  • World Modeling
  • Published Forecasts
  • MCP Server
  • Progress Monitoring
Guides
  • Turn Claude into an Accurate Forecaster
  • Forecast Outcomes for a List of Entities
  • Forecast Conditional Scenarios
  • Forecast Categorical and Threshold Questions
  • Find Profitable Prediction Market Trades
  • Research a Question with a Team of Agents
  • Add a Column via Web Research
  • Error Handling in FutureSearch: Failed Rows and Partial Results
Case Studies
  • Forecast a Decision: Grant Funding at Three Levels
  • Forecast a Decision: Which CEO Replacement Maximizes Share Price
  • Forecast a Binary Question End to End
  • Forecast a Date, Then Grade It
  • Forecast Categorical Outcomes for Two Stealth Labs
  • Forecast Conditional Scenarios for OpenAI's IPO
  • Forecast Anthropic and OpenAI IPOs: Dates and Valuations
  • Forecast a Sum-of-the-Parts SpaceX IPO Valuation
  • Forecast Founder Seed Valuations for AI Researchers
  • Find Startups Selling to Frontier AI Labs
  • Run 10,000 LLM Web Research Agents
FutureSearchby futuresearch
by futuresearch

Forecast Conditional Scenarios for OpenAI's IPO

A conditional forecast runs both branches of a scenario in one call. The outcome type stays whatever the question calls for, numeric here, and the condition is a separate parameter. Every output column comes back twice, suffixed _given_condition and _given_not_condition.

The question

FieldValue
QuestionWhat will OpenAI's market capitalization be 90 days after its IPO?
Outcome typenumeric, trillions USD
ConditionOpenAI ships a step-change capability release before its IPO
Step-changeA model that leads a majority of major AI benchmarks for four consecutive weeks after release

Conditional forecasts are HIGH effort only.

Run it

import asyncio

from pandas import DataFrame
from futuresearch.ops import forecast


async def main():
    result = await forecast(
        input=DataFrame([
            {
                "question": "What will OpenAI's market capitalization be 90 days after its IPO?",
                "resolution_criteria": (
                    "Closing market capitalization on the 90th calendar day after "
                    "OpenAI's first day of public trading."
                ),
            },
        ]),
        forecast_type="numeric",
        output_field="market_cap",
        units="trillions USD",
        condition=(
            "OpenAI ships a step-change capability release before its IPO: a model "
            "that leads a majority of major AI benchmarks for four consecutive "
            "weeks after release."
        ),
        effort_level="HIGH",
    )
    print(
        result.data[
            [
                "market_cap_p50_given_condition",
                "market_cap_p50_given_not_condition",
                "rationale",
            ]
        ]
    )


asyncio.run(main())

condition applies one condition to every row. When rows carry distinct conditions, name the column holding them with condition_field instead. The two are mutually exclusive.

Results

BranchWeightMedian 90-day market cap
Step-change model ships before the IPO23%$1.33T
No step-change model77%$800B
Probability-weighted$920B

The 23% weight comes from a thresholded forecast that prices every step-change date in a single run, read off at the July 2027 median IPO date. The no-step-change median of $800B sits about 6% below OpenAI's last private mark.

When the condition is a choice you control rather than something you observe, use decision instead.

Going deeper

  • Long-form analysis: OpenAI Revenue, Losses, and IPO Valuation
  • Published output: post-IPO market cap in the step-change case
  • Guide: Forecast conditional scenarios

Ready to try it yourself? Run it in the app →