VolumeWeightedAveragePriceExecutionModel silently degenerates on daily resolution

Abierto
  • ¿Algo no está bien?
#10,011 1 comentario 0 reacciones 0 asignados Ver en GitHub

Los mantenedores suelen responder en 1 día

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
2/5
Tiempo estimado
1-3 horas
Aptitud para principiantes
55/100
Tipo de issue
Documentación
Claridad
Bien especificado
Estado de actividad
Activo
Stack tecnológico
csharp
Área
backend

Línea de trabajo

The relevant files are Algorithm.Framework/Execution/VolumeWeightedAveragePriceExecutionModel.cs, where PriceIsFavorable compares bid and ask against data.VWAP, and Indicators/IntradayVwap.cs, where the sums reset at each date boundary. Start by confirming the daily-resolution reasoning in the issue, then add a <remarks> block to both class documentations stating the model is intraday-only and that daily or coarser resolution degrades the indicator to the bar's typical price. Done when both remarks are present and build cleanly. Behaviour changes such as rejecting daily input need a maintainer decision first.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

Summary

VolumeWeightedAveragePriceExecutionModel is an intraday execution model, but it accepts any subscription resolution. On daily data the IntradayVwap indicator resets at every bar, so the "VWAP" it compares against is just that bar's typical price, and the model silently degrades into "buy below the bar midpoint, sell above it".

Nothing warns about it, and the class documentation does not say the model is intraday only, so this is easy to hit by accident.

What happens with daily data

IntradayVwap accumulates within a date only:

// Indicators/IntradayVwap.cs:41-47
// reset vwap on daily boundaries
if (_lastDate != input.EndTime.Date)
{
    _sumOfVolume = 0m;
    _sumOfPriceTimesVolume = 0m;
    _lastDate = input.EndTime.Date;
}

and for a TradeBar it accumulates (High + Low + Close) / 3 (:103).

With Resolution.Daily every bar starts a new date, so the sums are zeroed before each accumulation and the indicator returns exactly that bar's typical price. PriceIsFavorable then reduces to:

// Algorithm.Framework/Execution/VolumeWeightedAveragePriceExecutionModel.cs:147-168
if (unorderedQuantity > 0)
{
    if (data.Security.BidPrice < data.VWAP) { return true; }   // bid < (H+L+C)/3
}
else
{
    if (data.Security.AskPrice > data.VWAP) { return true; }   // ask > (H+L+C)/3
}

Both conditions are roughly a coin flip around the bar midpoint, so the model keeps trading at random while paying the spread. Order sizing is capped at 1% of the bar's volume (MaximumOrderQuantityPercentVolume, default 0.01m), so the damage is bounded, but the result has nothing to do with VWAP. Since the model's name promises intraday execution, a user who backtests it on daily data has no way to tell that the premise does not hold.

The indicator behaviour is intended (#9703 / #9705 made Reset() consistent with it), so this is not a bug in IntradayVwap. The gap is that the model does not validate its input, and the docs do not call it out.

StandardDeviationExecutionModel does not have this problem: an SMA and standard deviation over N daily bars are meaningful.

Options

  1. Documentation only — add a <remarks> to the model (and to IntradayVwap) saying it is an intraday model, and that at daily resolution or coarser the indicator degenerates to the bar's typical price. No behaviour change, no risk.

  2. Reject the input — throw in OnSecuritiesChanged when a security with Resolution.Daily or coarser is added, with a message pointing at ImmediateExecutionModel / StandardDeviationExecutionModel for daily algorithms. This is correct but breaks any existing algorithm that currently does this, so it needs a deliberate decision.

  3. Warn once — CONTRIBUTING.md says framework modules must stay silent ("production code should be silent unless there is a fatal exception ... logging or debugging is not allowed inside LEAN framework modules"), and the only exceptions left in Algorithm.Framework are PearsonCorrelationPairsTradingAlphaModel and OpenInterestFutureUniverseSelectionModel. So this would need an explicit waiver; mentioning it for completeness.

I would suggest 1 now and consider 2 for a future major release. Happy to open a PR with the documentation change, or with either of the others if you prefer.

Version

Verified against master at 138382568 (2026-11-02).

Lenguaje dominante
C#
Estrellas
22.4k
Forks
5.4k
Merge medio
1 d 22 h
PR fusionados (30 d)
55

Preparar el entorno

Abrir en Codespaces

Inicia el contenedor de desarrollo del proyecto en tu navegador, con tu propia cuenta de GitHub.

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de QuantConnect/Lean

Todos los issues de QuantConnect/Lean

Issues similares

Más issues de C#

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.