HKUDS/Vibe-Trading · error · ValueError

start_date must be YYYY-MM-DD

Error message

start_date must be YYYY-MM-DD

What it means

autopilot_tool validates backtest inputs before writing run artifacts; start_date must parse strictly as YYYY-MM-DD via datetime.strptime. Any other format (slashes, month names, YYYYMMDD) fails.

Source

Thrown at agent/src/tools/autopilot_tool.py:241

    candidate = (data_sources or ["auto"])[0]
    try:
        from backtest.loaders.registry import VALID_SOURCES
    except Exception:  # pragma: no cover - registry import is environment-stable
        return candidate, None
    if candidate in VALID_SOURCES:
        return candidate, None
    return "auto", (
        f"hypothesis data_source {candidate!r} is not a known loader source; "
        "fell back to 'auto'"
    )


def _validate_backtest_dates(start_date: str, end_date: str) -> None:
    """Validate backtest dates before writing any run artifacts."""
    try:
        start = datetime.strptime(start_date, "%Y-%m-%d").date()
    except ValueError as exc:
        raise ValueError("start_date must be YYYY-MM-DD") from exc
    try:
        end = datetime.strptime(end_date, "%Y-%m-%d").date()
    except ValueError as exc:
        raise ValueError("end_date must be YYYY-MM-DD") from exc
    if start > end:
        raise ValueError("start_date must be on or before end_date")


def _run_dir_for_hypothesis(hypothesis_id: str) -> Path:
    """Return a path-contained run directory for any persisted hypothesis id."""
    suffix = hashlib.sha256(hypothesis_id.encode("utf-8")).hexdigest()[:12]
    return Path.home() / ".vibe-trading" / "runs" / f"autopilot_{suffix}"


class GenerateBacktestConfigTool(BaseTool):
    """Generate backtest config.json from a research hypothesis.

    Reads a hypothesis, derives config fields from its universe and

View on GitHub (pinned to 80ffdda44c)

Solutions

1. Format dates as %Y-%m-%D strictly (date.strftime('%Y-%m-%d'))
2. Normalize incoming datetimes to .date() then format
3. Add a regex pre-check r'^\d{4}-\d{2}-\d{2}$' before calling the tool

Example fix

// before
execute(start_date='2023/01/01', ...)
// after
execute(start_date='2023-01-01', ...)

import re
assert re.fullmatch(r'\d{4}-\d{2}-\d{2}', start_date), 'start_date must be YYYY-MM-DD'

def is_iso_date(s: str) -> bool:
    import re, datetime
    if not re.fullmatch(r'\d{4}-\d{2}-\d{2}', s or ''):
        return False
    try:
        datetime.date.fromisoformat(s); return True
    except ValueError:
        return False

try:
    _validate_backtest_dates(start_date, end_date)
except ValueError as e:
    start_date = normalize_date(start_date); retry

When it happens

Trigger: Calling autopilot execute with start_date='2023/01/01', '01-02-2023', '20230101', or a datetime object stringified with time components.

Common situations: Frontend date pickers emitting locale formats; LLM-supplied dates in natural formats; ISO datetime strings ('2023-01-01T00:00:00') passed untrimmed.

Related errors

• end_date must be YYYY-MM-DD
• alpha_id not found
• period {period!r} must be YYYY-YYYY or YYYY-MM-DD/YYYY-MM-DD
• invalid date: {date!r}; expected YYYY-MM-DD
• invalid date for tushare fallback: {value!r}

AI-assisted analysis of HKUDS/Vibe-Trading@80ffdda44c (2026-08-28). Data as JSON: /api/errors/f3a495309b3c75c9. Report an issue: GitHub.