Selecting Futures Price Series for Portfolio VaR
Summary
The document discusses how to build price histories for calculating Value at Risk on a portfolio that may include interest-rate, commodity, and equity futures. Its central concern is that contract rolls, changing liquidity, and commodity seasonality can distort returns and therefore affect a parametric VaR estimate and its covariance matrix. It raises the choice between using a nearby contract series and constructing a constant-maturity series, noting that the suitable approach depends on the application.
Two ways to form a continuous return history are described: use the return of the most active contract on each date, or combine returns from leading contracts using open-interest weights. The latter can smooth the transition across contracts. The document gives no empirical comparison, implementation details for roll dates or seasonal contracts, or VaR backtest. These choices therefore remain modeling judgments that should reflect the portfolio’s exposures and the intended risk measure.
Key ideas
- Nearby-contract and constant-maturity series are alternative ways to represent futures prices over time.
- A rolling series based on the contract with the highest open interest selects the most active contract each day.
- Returns from leading contracts can be combined using open interest as the weighting basis.
- Contract selection and roll construction affect the return data used in VaR covariance estimates.
Tags
Full text
# Value at Risk for Futures Contracts # Value at Risk for Futures Contracts I would like to know how you would compute Value at Risk on a portfolio of futures i.e rates futures, commodity futures and equity. How do you deal with the discontinuous form of commodity futures for example, how do you select the right time series for these futures when liquidity and seasonality aspects kicks in. Do we need to build a constant maturity futures contract to be able to calculate the VaR? Or we can directly calculate the VaR on the continuous price time series( GFUT as example on Bloomberg). I am afraid that as Parametric VaR involves the calculation of CovVar matrix of returns, the selection of good n fair prices of futures in critical in order not to get misleading results of VaR? Any experience on that? Thank you very much S ## Answer by nbbo2 (score 1) https://quant.stackexchange.com/a/19209 This link has a worthwhile discussion of two possible approaches: the Nearby approach (paragraph 6.6.1) and the Constant Maturity approach (para 6.6.2). http://www.value-at-risk.net/futures-prices/ . With the pluses and minuses of each. Ultimately it is going to come down to your judgement of what is best in your situation. ## Answer by mt_christo (score 0) https://quant.stackexchange.com/a/19205 There are ways to get a continuous time series from switching futures prices. These include: 1) Taking return of a leading future contract (max open interest) on every date, and 2) Taking a weighted average return among a group of leading contracts, with weights based on open interest of each contract. For example: ``` R_average = (OI1*R1 + OI2*R2)/(OI1 + OI2) ``` on any given date, where OI is Open Interest and R is return for contracts with top 2 OIs among the contracts traded on that date.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.