Comment une limite maximale de poids affecte les rendements des portefeuilles ERC

S'abonner à notre lettre d'information

Adding a maximum weight limit, or cap, to an equal risk contribution (ERC) portfolio does not affect its information ratio – though lower weight caps are associated with higher return and volatility.

Le ratio d'information mesure les rendements excédentaires par unité de risque de volatilité, évaluant essentiellement la régularité de la surperformance d'une stratégie par rapport à un indice de référence. Plus le ratio d'information est élevé, mieux c'est.

A study by Quantilia found that maximum weight caps do not tend to affect the information ratios of ERC-weighted portfolios or inverse-volatility portfolios. In both cases, however, lower weight caps tended to generate higher volatility and higher mean rolling returns.

This is because under ERC and inverse-volatility weighting schemes, a few strategies often make up most of the total portfolio’s weight. By restricting the weightings of these strategies, more weight is shifted to strategies that contribute higher volatility and risk – leading to higher portfolio volatility and returns.

Les fonds ERC sont pondérés de telle sorte que chaque titre contribue de manière égale au risque total du portefeuille, réduisant ainsi la concentration du risque du portefeuille. Cette approche est similaire à celle de l'inverse de la volatilité, bien que la stratégie ERC tienne également compte des corrélations entre les titres au sein du portefeuille.

The Quantilia study used a range of weighting caps, from 10% to 50%. The report was based on 20 baskets of 20 strategies each, with a split between alpha and beta strategies and exposure to different asset classes and strategy providers. The study assumed transaction costs of 20 basis points and used a semi-annual rebalancing strategy.

It also found that implementing lower weight limits, or weight floors, affected these portfolios’ return, volatility and information ratio figures. If weight floors are raised, information ratios tend to decrease. The study suggests using lower weight limits of only 0.5% or 1%.