HKUDS/Vibe-Trading · error · ValueError

valuations[{index}].value must be finite

Error message

valuations[{index}].value must be finite

What it means

Raised when a valuation's value coerces to a float that is not finite (NaN, +inf, -inf, per math.isfinite). Infinite or NaN results poison every downstream statistic, so the tool rejects them upfront.

Source

Thrown at agent/src/tools/cashflow_analytics_tool.py:332

            or an entry lacks a usable date or value.
    """
    if not isinstance(raw, list) or len(raw) < 2:
        raise ValueError("valuations must be an array of at least two {date, value} objects")
    if len(raw) > _MAX_VALUATIONS:
        raise ValueError(f"valuations may contain at most {_MAX_VALUATIONS} entries")

    pairs: list[tuple[date, float]] = []
    for index, item in enumerate(raw):
        if not isinstance(item, dict):
            raise ValueError(f"valuations[{index}] must be an object with date and value")
        if "date" not in item or "value" not in item:
            raise ValueError(f"valuations[{index}] needs both 'date' and 'value'")
        try:
            value = float(item["value"])
        except (TypeError, ValueError) as exc:
            raise ValueError(f"valuations[{index}].value must be numeric") from exc
        if not math.isfinite(value):
            raise ValueError(f"valuations[{index}].value must be finite")
        pairs.append((item["date"], value))
    return pairs


def _resolve_flows(kwargs: dict[str, Any]) -> CashFlowSeries | None:
    """Build the cash-flow series from inline records or from a file.

    Args:
        kwargs: The tool's raw inputs.

    Returns:
        A ``CashFlowSeries``, or ``None`` when no flows were supplied.

    Raises:
        ValueError: If both sources were given, an inline record is malformed,
            or a currency is missing. File problems surface as
            ``CashFlowIngestError``, which is a ``ValueError``.
    """

View on GitHub (pinned to 80ffdda44c)

Solutions

1. Filter out non-finite values before calling the tool (e.g. filter out NaN from pandas series with .dropna())
2. Fix the upstream division-by-zero or overflow producing inf/NaN
3. Use json.dumps(..., allow_nan=False) upstream to catch these at serialization time

Example fix

# before
valuations = series.to_dict()  # may include NaN
# after
valuations = [{"date": d, "value": float(v)} for d, v in series.dropna().items()]

import math
valuations = [v for v in valuations if math.isfinite(float(v["value"]))]

import math

def is_finite_value(item: dict) -> bool:
    try:
        return math.isfinite(float(item["value"]))
    except (TypeError, ValueError):
        return False

When it happens

Trigger: valuations=[{"date": "2024-01-01", "value": "NaN"}] or value="Infinity" (float() accepts these strings in Python), or a computed value that overflowed to inf before being passed in.

Common situations: Aggregations that divided by zero upstream, pandas/NumPy pipelines emitting NaN for missing data then serializing to JSON, or literal 'NaN'/'Infinity' tokens in JSON payloads (Python json.loads accepts them).

Related errors

• valuations[{index}] needs both 'date' and 'value'
• valuations[{index}].value must be numeric
• pass either flows or flows_path, not both
• flows_columns must be an object mapping field to column name
• flows must be an array of {date, amount, kind} objects

AI-assisted analysis of HKUDS/Vibe-Trading@80ffdda44c (2026-08-28). Data as JSON: /api/errors/089cbe31abafca0d. Report an issue: GitHub.