LiveMacroEval

Can LLMs Take the Pulse of the Economy? A Real-Time Evaluation of LLM Nowcasts on Macroeconomic Indicators

Live Last updated Next refresh
LiveMacroEval pipeline: sixteen indicators, hourly LLM nowcasts across the pre-release window, scored against the market and against Fed, Bloomberg and ARIMA baselines I. Supply & production Real GDP Industrial Production Durable Goods ISM Manufacturing PMI TASKS Four Macro Indicator Families II. Demand & inflation CPI, PPI PCE Price Index Real PCE Retail Sales ISM Services PMI III. Labor market Nonfarm Payrolls Unemployment Rate IV. Housing market Building Permits Housing Starts New Home Sales Existing Home Sales EVALUATION From forecast error to market-relevant payoff Market ReturnCapture Score · Did model-predictedshocks match realizedmarket moves? PolymarketBetting Return · Did the model beatthe wisdom-of-the-crowd? Theme Score · Where does eachmodel specializeacross the macroeconomy? NOWCASTING TIMELINE Anchored to the official release schedule of each indicator GroundTruthRelease GPT Claude Qwen High-frequency LLM nowcasting throughout the window BloombergConsensus · Large professional forecastsurvey ARIMA · Statisticalbenchmark forecast · Once per month Fed nowcasts · Atlanta Fed · New York Fed · St. Louis Fed · Cleveland Fed · Chicago Fed · Several times per month · Rolling estimates near release
LLM agents nowcast sixteen U.S. macro indicators in real time. We score them against the equity market, Polymarket, and professional economists.

Why a live benchmark?

Nowcasting means estimating an indicator before it is officially released. Central banks staff whole teams to do it, and an LLM agent can be asked far more often than any of them.

The issue is LLM memorization leading to data contamination. GDP and CPI are among the most reported numbers on earth, so their past values are almost certainly memorized during pretraining, and testing on old releases measures recall rather than nowcasting.

LiveMacroEval scores only nowcasts made before the release, which keeps it clean by construction.

The sixteen indicators

Who we compare against

Human experts

Federal Reserve regional banks

Institutional nowcasts, where they exist.

    Consensus & econometrics

    Bloomberg ECOS and ARIMA

    Bloomberg ECOS is the professional economist consensus, covering all sixteen indicators, and it is famously hard to beat.

    ARIMA is a univariate time-series baseline, refit at the start of every window.

    How we score

    Plain error on the released level ignores that indicators differ in economic weight. We use two market-grounded scores instead. Full definitions and estimation are in the paper.

    Metric 1

    LiveMacro Score

    Macro surprises move equity prices. We weight each indicator by its measured effect on S&P 500 futures at the moment of release, so sixteen accuracies collapse into one comparable number. +1 is perfect, 0 ties the consensus, −1 is arbitrarily worse. Each release is scored on the nowcast in force when it comes out, and only if that nowcast is less than a day old.

    Metric 2

    LiveBetting Score

    At fixed times each day before a release, one hour after the agents' scheduled runs, we bet $1 on the Polymarket bucket holding the agent's latest nowcast, provided that nowcast is less than a day old. Prices are the crowd's own odds, so liquidity is priced in: beating a confident crowd pays a lot, agreeing with it pays little. The score is unit-free, so it aggregates across very different indicators.

    Leaderboard: LiveMacro Score

    #Model LiveMacro90% CI vs. consensus
    Loading…

    Tool and agent design

    One base model, three ways of running it: a plain prompt, the same prompt with a financial-data plug-in, and a multi-agent team.

    #Configuration LiveMacro vs. consensus
    Loading…

    Score by theme: where the skill is

    Nine target months live. The overall score hides sharp differences across the economy: the same arm can beat the consensus on one block and trail it on another.

    Real-time return: LiveBetting Score

    Watching a belief update

    GPT-5's March 2026 CPI nowcast sits near 2.7% through early April, then jumps to about 3% on April 8 and tracks the realized print from there. Two scheduled events land inside that window: a large gasoline-inventory drawdown, and FOMC minutes flagging upside inflation risk. The size of the revision matches a back-of-the-envelope gasoline contribution under BLS expenditure weights.

    Curves are a Gaussian smooth of the hourly nowcasts.

    Citation

    @inproceedings{zhao2026livemacroeval,
      title         = {Can LLMs Take the Pulse of the Economy? A Real-Time Evaluation
                       of LLM Nowcasts on Macroeconomic Indicators},
      author        = {Zhao, Xinyue and Zhang, Ruiyi and Ye, Liqin and
                       Cao, Rui and Xie, Pengtao and Chava, Sudheer},
      booktitle     = {Proceedings of EMNLP},
      year          = {2026},
      eprint        = {2608.30110},
      archivePrefix = {arXiv},
      primaryClass  = {cs.CL},
      url           = {https://arxiv.org/abs/2608.30110}
    }