Pre-registered study · November 2026 Code and data on GitHub

We gave an AI chart reader a fair exam on Japanese markets. It failed both tests.

Kronos is an open-source foundation model trained on billions of candlesticks. We asked two practical questions: can it tell which Tokyo stocks will beat the others, and can it warn when the Nikkei is about to get rough? We wrote the rules down before looking at any result, tested it only on months it had never seen, and made it beat simple free rules after trading costs. It did not.

✓Rules locked before any result ✓Unseen data only Jul 2024 – Apr 2026 ✕Volatility forecasting failed ✕Stock picking failed
1

What Kronos does

Kronos reads the last 40 daily candles and imagines many possible next 12 days, like a weather forecaster drawing a fan of possible futures. These are real forecasts it made for the Nikkei 225. It opens on 2 August 2024, the Friday before the market's worst day in decades. Slide through the dates and compare its fan with what happened.

2

How the exam works

A model can look brilliant on data it has already memorised, and a test can be quietly bent after the fact. We guarded against both. Every rule and pass mark went into PREREG.md, committed before any test data was downloaded. Kronos learned from data up to June 2024, so every test uses later months. A final stretch of months stayed sealed, to be opened only if Kronos passed.

The project, step by step

Tap a step to read what happened in it.

    Which months are used for what

    Hover or tap a bar.

     

    3

    First, we tested the tester

    A broken exam can make a bad student look good. Before grading Kronos, we checked the grading code on cases where the right answer is known in advance.

    Does it find signals only when they exist?

    A score of 0 means no predictive power. Whiskers show the margin of error.

    Passed. It found nothing in random fake prices, and found a hidden signal of 0.05 when we planted one.

    What does one trade really cost?

    Round trip for a large Japanese stock, in basis points (1 bp = 0.01%).

    10 bp kept. Tokyo's minimum price steps put a realistic round trip at 7 to 14 bp, under the 15 bp line that would have forced a change.

    The wrinkle we found and kept

    Scrambling which stock gets which signal gave zero, as it should. Scrambling which day gets which returns did not: some stocks stay strong or weak for months, and that pattern survives a date shuffle. We recorded this before grading and changed no rule.

    4

    Test 1: can it predict how rough the market gets?

    If Kronos could see a storm coming, a trader could bet smaller before it hits. We compared its 12-day volatility forecasts for the Nikkei with three textbook formulas over 435 trading days.

    Predicted vs actual swings, 12 days ahead

    Yearly volatility in %. Higher means rougher. Hover to read any day.

    Scoreboard: lower is better

    Average forecast error (QLIKE). It punishes missing a storm more than a false alarm.

    Betting with each forecast

    ¥1,000,000 in Nikkei micro futures, sized by each forecast. Profit in thousands of yen.

    Failed. Kronos lost to every simple formula. It kept underestimating storms, so sizing by it took bigger bets and the deepest drop (−20.5% against −15.4%). The extra profit came from carrying more risk in a rising market, not from skill.

    5

    Test 2: can it pick which stocks will do better?

    Every trading day from November 2024 to April 2026, Kronos ranked about 500 of Japan's largest companies by expected return over the next 5 or 12 days. We tested six ways of reading its forecasts. To pass, a version had to add information beyond simple price rules, make money after costs once we discounted for trying six versions, avoid being one lucky streak, and work in most months.

    Its top picks against the average stock

    Extra return per year from holding Kronos's top fifth of stocks instead of all of them, after 10 bp trading costs.

    Day by day

    Daily ranking score (rank IC). Above zero means its ranking matched what happened.

    ¥100 in its top picks against ¥100 spread over all stocks, after costs.

    Failed. Its ranking scores averaged between −0.001 and +0.008, nowhere near significant, although the test could detect scores down to about 0.01 to 0.03. It added nothing beyond simple price rules. Five of six versions lost to simply holding every stock, and the best one broke even. That matches an independent test on US stocks.

    All six versions, as a table
    6

    What this does and does not show

    What it shows

    Used out of the box on daily candles, Kronos-small gave a trader nothing usable on Japanese large caps or the Nikkei between mid-2024 and April 2026. Simple formulas forecast risk better, and its stock rankings were noise once costs were counted.

    What it does not show

    We tested the small model without fine-tuning, on daily bars, over about 18 months, through a free data plan. A fine-tuned or larger model, intraday data, or other markets could behave differently. The sealed final exam was never opened, because nothing passed.

    7

    Reproduce it

    Everything except the licensed market data is in the repository. You need Python 3.12 with uv, a free J-Quants account, and a Kaggle account with GPU access. Each step below writes files that the next step checks.

    1. Set up and run the tests
      git clone https://github.com/SIMETRAL55/kronos-zero-shot-japan
      cd kronos-zero-shot-japan
      uv sync && bash scripts/fetch_kronos.sh && uv run pytest
    2. Fetch the data
      uv run python scripts/fetch_dev_data.py nikkei
      uv run python scripts/fetch_dev_data.py jquants --free   # needs JQUANTS_API_KEY in ~/.config/kronos-explore/jquants.env
    3. Run the forecasts on Kaggle's free GPUs
      export KAGGLE_USERNAME=your-name
      uv run python scripts/push_h2_dev.py                   # H2, about 12 minutes
      uv run python scripts/build_h1_windows.py && uv run python scripts/kaggle_dataset.py create
      # then the five H1 jobs listed in README.md, about 23 GPU-hours
    4. Grade, in the pre-registered order
      uv run python scripts/eval_h2.py h2dev
      uv run python scripts/eval_h1.py costs && uv run python scripts/eval_h1.py p5 && uv run python scripts/eval_h1.py dev

    The free plan's two-year window moves forward every day, so an exact rerun of our dates now needs one month of the Light plan. A rerun on your current free window tests the same idea on slightly different dates.

    8

    Words used on this page

    Pre-registration
    Writing down the rules and pass marks before seeing any results, so nobody can quietly change them.
    Sealed final exam
    Months kept locked and used only once at the end, to confirm a pass.
    Rank IC
    How well a day's ranking of stocks matches the ranking of what they actually returned. 0 is no skill, 1 is perfect.
    Volatility
    How wildly prices swing. About 20% a year is normal for the Nikkei.
    HAR, GARCH, EWMA
    Free textbook formulas that predict future swings from recent swings.
    Placebo test
    Running the exam on fake or scrambled data where the right answer is "nothing", to prove the exam isn't fooling us.
    Basis point
    One hundredth of a percent. 10 bp on ¥1,000,000 is ¥1,000.