HKUDS/Vibe-Trading · error · ValueError

returns and var must be the same length, got {ret_values.siz

Error message

returns and var must be the same length, got {ret_values.size} and {var_values.size}

What it means

Once both inputs are 1-D (and var was not a broadcast scalar), _align requires equal element counts; a length mismatch raises ValueError showing both sizes. This typically happens when one side lost rows (dropna, rolling-window warmup) or plain arrays without indexes were passed with different lengths.

Source

Thrown at agent/src/quantlib/var_backtest.py:286

                "partial join silently compares each day against another day's "
                "forecast."
            )

    ret_values = np.asarray(returns, dtype=float)
    if ret_values.ndim > 1:
        raise ValueError(f"returns must be 1-D, got shape {ret_values.shape}")
    ret_values = ret_values.ravel()

    var_values = np.asarray(var, dtype=float)
    if var_values.ndim == 0:
        var_values = np.full(ret_values.shape, float(var_values))
    else:
        if var_values.ndim > 1:
            raise ValueError(f"var must be 1-D or scalar, got shape {var_values.shape}")
        var_values = var_values.ravel()

    if ret_values.size != var_values.size:
        raise ValueError(
            f"returns and var must be the same length, got {ret_values.size} "
            f"and {var_values.size}"
        )
    if ret_values.size == 0:
        raise ValueError("returns is empty")

    keep = np.isfinite(ret_values) & np.isfinite(var_values)
    dropped = int((~keep).sum())
    if not keep.any():
        raise ValueError("no observation has a finite return and a finite var")

    index = ret_index if ret_index is not None else var_index
    kept_index = index[keep] if index is not None else None
    return ret_values[keep], var_values[keep], kept_index, dropped


def violation_indicator(
    returns: pd.Series | np.ndarray | Sequence[float],

View on GitHub (pinned to 80ffdda44c)

Solutions

Trim returns to the var series' valid range or re-attach indexes and use pandas alignment.
Regenerate var over the full returns sample so lengths match by construction.
Slice both to the overlapping suffix: rets[-len(var):].

Example fix

# before
var_backtest(rets, var)  # 500 vs 480
# after
var_backtest(rets[-len(var):], var)  # aligned tail sample

Defensive patterns

Strategy: validation

Validate before calling

import numpy as np
assert np.asarray(returns).size == np.asarray(var).size

Type guard

def same_length(returns, var) -> bool:
    import numpy as np
    return np.asarray(returns).size == np.asarray(var).size

Try / catch

except ValueError as e:
    if 'same length' in str(e): trim both to the overlapping tail

Prevention

Trim returns to the VaR warmup window before backtesting
Use indexed Series so alignment errors surface as label errors early

When it happens

Trigger: Passing a numpy returns array of 500 values and a var array of 480 (e.g. after a 20-day rolling warmup dropped rows), or lists built from different date ranges when indexes are absent so the earlier label check cannot fire.

Common situations: Rolling VaR estimators that return shorter series than the input returns; dropping NaNs from one array only; concatenating train/test segments inconsistently.

Related errors

AI-assisted analysis of HKUDS/Vibe-Trading@80ffdda44c (2026-08-28). Data as JSON: /api/errors/b7c78502694097ab. Report an issue: GitHub.