Hedging Benefits from Individual Assets

Questions

Questions:

Commodity portfolios are often characterized by assets with either high or low PnL correlation, e.g.
- High 𝜌 between intra-commodity futures, such as Corn Mar vs Corn May.
- Low 𝜌 between inter-commodity futures, such as Silver Dec vs Cocoa Jul.
The ability to have both long and short allocations among contracts of different as well as the same commodity can have significant impact on the portfolio risk.
How large impact does this hedging effect have on an AiLA long-short portfolio?

Approach:

Construct portfolio scenarios with different intra-commodity allocation signals to indicate the impact from hedging.
Compare the AiLA asset allocations applied to the individual contracts vs benchmark commodity returns to demonstrate the relevance of how the AiLA allocations are applied.

Long vs Short Hedging

In the case of two assets (𝑎,𝑏) with market return 𝜌 ~ 1, where 𝑎 (𝑏) are allocated long (short) according to a signal 𝑠 ^𝐿 (𝑠 ^𝑆).
The PnL correlation is a combination of the market correlation together with the signal overlap.

𝜌(𝑃𝐿 ^𝑎 ,𝑃𝐿 ^𝑏 ) ~ 𝑐𝑜𝑛𝑠𝑡 ∙𝜌(𝑟 ^𝑎 ,𝑟^𝑏)∙𝜌(𝑠 ^𝐿,𝑠 ^𝑆)
𝜌(𝑠 ^𝐿 ,𝑠 ^𝑆 )= 0: if the 𝑠 ^𝐿 signals are always generated on different days than 𝑠 ^𝑆 , the two PnL streams are complementary but without any hedging benefit w.r.t. their combined risk.
𝜌(𝑠 ^𝐿 ,𝑠^𝑆 ) ≠ 0: if the 𝑠 ^𝐿 and 𝑠 ^𝑆 signals can occur on the same days, hedging will reduce their combined risk with a potential benefit w.r.t. their risk adjusted performance.
This scenario applies to (high 𝜌 ) inter-commodity assets, such as WTI vs RBOB, however, can also be applied to intra-commodity assets, such as WTI Jul vs WTI Dec.
The possibility of reducing the portfolio risk is one reason for generating AiLA allocations for individual commodity contracts.

The initial portfolio created for this exercise is based on multiple contracts per commodity, but where the intra-commodity allocation signals are effectively all long or short on a given day.
The portfolio is comprised by a diverse set of commodities, so the individual long and short portfolios are expected to benefit from diversification, however, the long vs short PnL correlation (hedging) is expected to be minimal.
A negative correlation of about -30% is observed between the long and short portfolios, which is expected roughly due to the common USD component among all assets used in the portfolios.
The long and short portfolios are constructed with similar risk and the total performance obtained is consistent with that expected, which also confirms the small hedging benefit w.r.t. a scenario of two independent long and short portfolios.

Two other portfolios were constructed with a common set of oil and metal markets, both with the same nr of contracts for each commodity.
“Same” portfolio: produced by allocations which effectively are either long or short for the same day across all the commodity contracts.
In contrast to the previous portfolio the same portfolio is not diverse, e.g. oil markets are expected to behave similarly, and therefore little diversification as well as hedging is expected.
“Mixed” portfolio: produced by allocation signals able to be long and short among commodities as well as contracts on a given day.
The enhanced long vs short correlation of the mixed portfolio demonstrates a significant risk adjusted performance benefit,

Since the allocation signals are produced differently for the two scenarios above, their long and short portfolio performances are different, but that is beside our point regarding their hedging benefit.

To address the relevance of applying the allocations to the individual assets, the allocations were used in three different scenarios,
- Long-Front: as a long-only reference, the absolute value of all contract allocations for a given commodity were summed, and the corresponding total allocation was applied to the front contract of that commodity.
- Net-Front: the net allocation across all contracts for a given commodity was allocated to the front contract of that commodity.
- AiLA: based on the AiLA allocations for each contract asset individually.
The exercise was made for an AiLA portfolio comprising 76 assets across a diverse set of 18 commodity markets.
The results suggest that while the AiLA allocation signals obtained for individual intra-commodity assets provide a significant portfolio risk benefit, most of the performance can be retained by applying their net allocation to the front contract of the corresponding commodities.

The AiLA allocations are generated for individual commodity assets across the curve, where long and short positions are possible for different contracts of the same commodity market.
The ability of having simultaneous long and short allocations among different intra-commodity contracts, can significantly enhance the correlation between the long vs short part of the portfolio.
The impact from hedging between the long vs short part of an AiLA portfolio was studied by constructing equivalent portfolios with different AiLA allocation signal overlap to maximize/minimize the effect.
The enhanced correlation resulted in a reduced portfolio risk and a significantly improved risk adjusted return, with only minor impact on leverage.
The results also indicated that most of the performance can be retained when the net allocation is applied to the rolling front month contracts, instead of the individual contracts.
In order to isolate the effect investigated all performance results in this study were produced ignoring realistic aspects, e.g. related to liquidity and transaction costs.