Managed-Futures-Carry-Whitepaper

MANAGED
FUTURES
CARRY:
A Practitioner's Guide
Adam Butler* and Andrew Butler†
*Adam Butler is Chief Investment Officer of ReSolve Asset Management SEZC (Cayman) ([email protected])
†
Andrew Butler, PhD is Chief Investment Officer of ReSolve Asset Management Inc. ([email protected])
Managed Futures Carry: A Practitioner’s Guide
Abstract
This whitepaper provides a comprehensive exploration of futures carry strategies from a practioner’s
perspective, dissecting the nuanced manifestations of carry across equities, bonds, commodities,
and currencies. By examining over three decades of data, we investigate the economic rationale
underpinning carry returns and their potential role in portfolio optimization. Our analysis employs
both classic and relative value approaches, utilizing cross-sectional and time series strategies,
as well as sophisticated ensemble and optimized portfolio constructions. We find that judiciously
implemented carry strategies may enhance returns while managing risk exposure, particularly
when accounting for market correlations and neutralizing exposure to key benchmarks such as
the S&P 500 and U.S. Treasury futures. The study also highlights the importance of considering
net returns after transaction costs, demonstrating the efficiencies achievable through trade netting
and weight smoothing. While the empirical evidence suggests that futures carry strategies can
contribute positively to portfolio performance, we emphasize the need for a measured perspective,
considering carry as part of a broader, diversified investment approach. This whitepaper aims to
contribute to the ongoing discourse on investment strategy optimization, offering insights into the
complexities of carry and its practical application in portfolio management.
2 ReSolve Asset Management

Contents
What is Carry? 4
Carry in Financial Markets 4
Economic Rationale for Carry Returns 5
Measuring Carry in Practice 5
Constructing Carry Strategies 7
Experimental Design 7
Data 7
Calculating Carry 8
Analytical Framework 10
Portfolio Construction 11
Performance Analysis 12
Sector Carry Strategies 12
Carry on the Full Liquid Futures Universe 15

Inverse Volatility Weighted Portfolios 15
Optimized Portfolios 17
Constrained Portfolios 19
Gross Versus Net Returns 22
Scenario Analysis 23
Carry in the Worst Benchmark Quarters 23
Carry During Benchmark Drawdowns 24
Stacking Case Studies 27
Conclusion 29
Disclaimers 31
References 32

What is Carry?
While investors generally prefer to sell investments for more than they purchased them for, price appreciation is not the only
source of returns. Landlords, for example, can earn rental income, seeking to generate income in excess of costs associated
with owning property.
This second source of return is called the expected yield or “carry,” and can be loosely defined as the expected return of
an investment assuming no change in its price. More thoroughly, carry is the economic benefit that one might expect to
achieve by holding – or “carrying” – a particular investment minus the costs associated with holding it. Carry strategies seek
to maximize exposure to this second component of returns.
We believe carry is a pertinent feature of all markets and a key driver of returns, allowing systematic carry strategies to serve as
a cornerstone of a well-diversified portfolio. While carry is almost always an expectation about what will happen in the future -
after all, tenants do not always make due on their rents! - we will mostly omit the ‘expected’ modifier going forward, for simplicity.
Carry in Financial Markets

Carry is a feature of all major asset classes. In equities, for example, an investor can expect to earn a dividend even when
price remains unchanged. What is the cost associated with holding equity? The return we can earn on cash, which can
either be viewed as our cost for financing the position, or the minimum opportunity cost of holding the position (Asness,
Moskowitz, and Pedersen 2013) (Frazzini and Pedersen 2014).
As a concrete example, at the time of writing, the expected 1-year dividend yield on the S&P 500 is approximately 1.5 percent.
By comparison, the 1-year U.S. Treasury rate is 4.75 percent. With the cost exceeding the expected benefit, U.S. equities
have a negative expected carry. We invoke the word ‘expected’ since both the financing rate on cash and the dividends paid
on stocks may change over the period we hold the securities.
Other asset classes, such as bonds, commodities, and currencies all have their own forms of carry. Bonds generate a
coupon, but they also benefit from potential roll yield as the bond matures over time. Again, the cost is the return of cash.
The definition gets a bit more complicated with currencies. For example, if a U.S. investor sells U.S. Dollars to buy British
Pounds and the exchange rate remains unchanged, they can expect to earn a return equal to the yield of short-term U.K.
Gilts minus the yield of short-term U.S. Treasury Bills. In other words, the return on cash held in England minus return on
cash the investor could have earned had they kept their money in U.S. Dollars. This difference is sometimes called “forward
points.”
There are additional complexities to consider with commodities due to their physical nature and use in industrial production.
The expected benefits of holding commodities are labeled under the umbrella term “convenience yield,” which captures
such effects as the ability to profit from temporary shortages and the ability to keep a production process running. The
costs of holding physical commodities not only include the cost of financing their purchase, but the costs associated with
transportation, storage, and insurance.

Asset Class Benefit of Holding Cost of Holding

Equities Dividend Yield Financing Rate
Bonds Yield Plus Roll Down Financing Rate
Commodities Convenience Yield Financing Rate, Storage, Transportation,
Insurance
Currencies Foreign Short Rate Local Short Rate
Economic Rationale for Carry Returns
Given that carry signals are explicitly measuring a component of expected returns, one possible explanation for the returns of
carry strategies is that it is the compensation for an investor’s willingness to bear some form of risk, or a portion of embedded
costs. In corporate bonds, for example, investors will demand higher yields when expected defaults - a significant driver of
price returns – increase (Ilmanen 2011). Additionally, carry premia may be related to time-varying inflation risk, as changes in
inflation expectations can impact asset prices and the returns to carry strategies (Boons et al. 2020).
We explore potential economic reasons for carry to exist in various asset classes below.
• Equity: If equity carry is a compensation for a fundamental risk, it follows that equity indices with a higher expected
dividend yield – or, more comprehensively, shareholder yield – must be fundamentally riskier. Given dividend yield’s
similarity to signals used within equity value investing, we might expect the return (and drivers) of an equity carry
strategy to correlate to the equity value premium.
• Government Bonds: Term spreads should compensate long-term bond holders for illiquidity risk, monetary policy
risk and inflation risk.
• Commodities: According to (Baltas 2017), “the theory of ‘normal backwardation’ suggests that commodity
producers take short futures positions in order to hedge against price drops and therefore pay a premium to an
investor that offers this insurance and takes a long position in the futures contract.” One can make the case that
commodity producers benefit from hedging the sale price of production forward because the asset owners who
provide capital for their operations will agree to better terms with the knowledge that the sale price is locked in.
There are also costs to inventory commodities, so the carrying costs implied in financial futures on the commodity
should reflect these costs. These and other dynamics contribute to positive or negative carry dynamics in different
commodites over time.
• Currencies: (Baltas 2017) suggests that higher short rates are generally “associated with rising inflation, funding
liquidity concerns, or consumption growth risk, which render higher-yielding currencies more vulnerable,”
necessitating a higher interest rate. Thus, currency carry has historically played out with investors funding higher
yielding emerging market currencies by borrowing in lower yielding developed market currencies, especially the
Japanese Yen and the U.S. dollar.
Measuring Carry in Practice
Estimating carry in a bottom-up fashion is no small task. In equities, for example, we need to develop dividend estimates for
thousands of companies around the world. For commodities, we would need to establish models of storage, transportation,
and insurance costs for dozens of unique markets.

Instead of diving deep into complex calculations for each asset, we can look at futures markets for insights. Futures contracts
are agreements to buy or sell an asset at a future date for a set price. These contracts inherently factor in the costs and
benefits of owning the asset.
Here’s a straightforward example using the S&P 500:
Imagine you want to put on a three-month trade in the S&P 500, which we will assume is trading at $5000. Instead of using
your own money to buy, however, you decide to borrow the required capital. Over the next three months, you earn any
dividends paid from the underlying stocks. At the end of three months, you sell the shares for some amount of money (F),
and pay back the $5000 plus some interest.
Your profit or loss is calculated as:
F + Dividends − 5000 × (1 + Interest Rate)
In an efficient market, a contract that allows you to lock in the price at which you will sell the S&P 500 for in the future (F)
should be priced such that you have the expectation of making neither a profit nor a loss.
If we simplify this further, we can say:
This equation shows us the concept of equity carry in simpler terms. You would expect to pay more for the S&P 500 in the
future if the cost to borrow money to buy the S&P 500 exceeds the expected dividend payout. This is because holding onto
the future contract and cash would earn you more than owning the S&P 500 and its dividends directly. As the future contract
gets closer to its end date, its price will move to match the S&P 500’s current price. If you bought the future at a higher price
than the S&P 500’s price today, you’d expect the future’s price to drop as it nears its expiration. This drop should equal the
difference between your borrowing costs and the dividends you expected to earn.
In simpler terms, using futures to estimate carry helps investors understand the potential costs and benefits of holding an
asset without getting lost in the weeds of complex calculations for each individual asset.
The same exercise can be carried out for bonds, currencies, and commodities with a similar result.
Fortunately, rather than having to guess at F , it is provided to us by futures markets. Simply comparing the futures price to
the current spot price, or the prices of futures contracts that expire in the distant future to those which expire in the near term,
allows us to efficiently capture the concept of carry across assets (Koijen et al. 2018) (Baltas 2017).
While we can in theory look to futures markets exclusively to provide information about carry, it is often more effective to
compute carry directly from cash markets. This is especially true for markets where deferred futures contracts are not
regularly traded (like in currencies), or where the cash market provides more granular information and is more efficient from
a pricing perspective (like in bonds). We explore the nuances of computing carry for markets in different sectors below.

Constructing Carry Strategies
There are three primary ways in which carry signals can be transformed into carry strategies:
• Calendar spreads: If deferred contracts are trading at a higher(lower) price than near-term contracts, sell(buy) the
back-month against the front month. This approach isolates expected supply/demand dynamics in a given market
at different points in the future (Szymanowska et al. 2014) (Gorton, Hayashi, and Rouwenhorst 2013).
• Cross-sectional carry: Futures markets with carry higher than the sector average will be held long against a
similarly sized basket of assets with below-average carry. Through its long/short construction, this approach aims to
neutralize directional market sector exposure and profit from the carry-driven, relative movement across markets in
each sector (Koijen et al. 2018) (Baltas 2017). Importantly, in this approach markets with positive(negative) expected
carry may be held short(long), if the expected carry is below the average carry expected across other markets in
the same sector. Indeed, there may be periods where all of the assets in a sector have positive expected carry, but
half of them are held short.
• Time-series carry: Futures markets with positive expected carry are held long and those with negative expected
carry are held short. As a result, the portfolio will maintain long and short sector biases commensurate with the
proportion of markets in each sector with positive or negative carry (Moskowitz, Ooi, and Pedersen 2012) (Baltas
2017). This approach has parallels to tactical allocation strategies that combine momentum and term structure
signals in commodity futures (Fuertes, Miffre, and Rallis 2010). For some sectors, like government bonds where
the yield curve has historically implied positive carry, this may result in a slightly larger average correlation to
respective betas. Indeed, as discussed below, time-series carry strategies have exhibited a slightly larger average
beta to government bonds. We also investigate ways to ameliorate this exposure at the portfolio level, and discuss
tradeoffs in a portfolio context.
Regardless of approach, there is no way to profit from carry without taking some price risk. As the last section suggests, an
investor who tries to go long futures with high carry and hedge by shorting spot will find themselves with no profit after all
costs are considered.
Experimental Design
Data
Futures price data across currencies, energies, metals, bonds, and equity indices was obtained for the period December 31,
1989 through December 31, 2023. Given the practitioner focus of this paper, we narrowed our focus to only those futures
contracts with the potential to scale to levels required by large institutions. We included the following list of highly liquid futures
markets in our analysis:
Currencies: Australian Dollar (AD), Canadian Dollar (CD), British Pound (BP), Euro (EC), Japanese Yen (JY)
Energies: Crude Oil (CL), Brent Crude Oil (CO), Heating Oil (HO), Natural Gas (NG), Gasoil (QS), RBOB Gasoline (XB)
Metals: Gold (GC), Silver (SI), Copper (HG)
Bonds: 10-Year Treasury Note (TY), U.K. Gilt (G), German Bund (RX)
Equities: E-Mini S&P 500 (ES), E-Mini Nasdaq 100 (NQ), Nikkei 225 (NX), S&P/TSX 60 (PT), German DAX (GX), Euro Stoxx
50 (VG), FTSE 100 (Z)

Futures contracts were rolled on a fixed trade date each month reflecting the historical average date of transfer of both
volume and open interest to the next contract in each market. Continuous futures price series were constructed by splicing
together the individual contract series, with returns calculated as the percentage change in daily closing prices.
For all simulations, markets are added when there is sufficient data post-inception to calculate the relevant carry signal. The
vast majority of markets in our liquid universe had started trading by January 1st, 1990 but a few markets were launched in
the mid-1990s. The inception dates for each market are illustrated in Figure 1. Note that markets that start trading prior to
January 1st, 1990 are truncated at this date for the purpose of our analysis.
Calculating Carry
We computed carry for markets of different sectors in slightly different ways, commensurate with their unique sources of carry
and available information. For example cash rates in global Treasury curves, and forward points for global currency markets
are readily available at daily frequencies with long histories. In addition, these markets have sparse deferred contracts, so it’s
difficult to rely on futures term strucure. On the other hand, commodity markets typically have liquid and granular deferred
contracts out many months and sometimes years. As a result, the carry on these markets can be reliably computed from
futures curves alone (Szymanowska et al. 2014) (Gorton, Hayashi, and Rouwenhorst 2013) (Bakshi, Gao, and Rossi 2017).
Equity carry was computed as deferred minus front month prices1.
1
It can be argued that a better measure of equity carry is shareholder yield. However, long-term data for this metric across global equity indices is not readily available, so we
settled for futures term structure, implicitly assuming futures traders will adjust for this in pricing.

Figure 1. Inception Dates for Markets in our Experimental Universe

The following describes the calculation of carry for each sector in detail:
For government bonds, carry is calculated by first converting annual yields of benchmark cash bonds into prices. This
conversion uses the formula:
for maturities of 1Y, 2Y, 3Y, 5Y, 7Y, 10Y, 20Y, and 30Y. Carry is then determined as the average of the pairwise differences
in the implied prices (adjusted for the time to expiry) of the bonds for all combinations of pairs, with the assumption that all
government bonds for the same country share the same carry value.
In the currency markets, carry is derived from the spot and forward term structure prices. Again, carry is calculated as the
average of the pairwise differences in the forward prices (adjusted for the time to expiry) observed over spot, 1M, 3M and
6M terms.
For equity indices and all other commodity markets, carry is calculated by taking the average of all the pairwise settlement
prices over the first 12 months of the term structure, excluding deferred contracts that fall below a certain liquidity threshold.
This approach captures the average expected cost or benefit of holding a commodity over the near term.
In general, the carry between two prices, and , with time to expiration tm and tn, respectively, is given as:
The pairwise average carry is then given as:
Analytical Framework
The analysis below seeks to provide a comprehensive examination of the performance of futures carry strategies across
multiple dimensions:
• carry definition: classic and relative value carry

• different transforms on the raw carry signals
• simple cross-sectional and time series carry strategies
• sector and total portfolio level
• several layers of ensembles
• naive vs optimized portfolios
• impact of certain practical constraints
It is useful to examine carry signals both in absolute terms and in relation to any long-term biases in the carry structure of
a market (Koijen et al. 2018) (Baltas 2017) (Baz et al. 2015). For example some markets have a structural negative (gold) or
positive (bonds) carry, but this does not lead to constant negative (positive) returns because investors demand (supply) the
asset for other reasons.

Carry relative value, therefore, is defined as the z-score of a market’s carry over 1, 3, 5, 10 years and since-inception historical
rolling windows. This approach normalizes the carry signal by its historical volatility:
where: - Current Carry is the carry calculated for the current period, - Mean Historical Carry is the average carry over a
specified historical period, - Standard Deviation of Historical Carry is the standard deviation of carry over the same historical
period.
In all instances, the z-score value for a market on each of the lookbacks are averaged to provide a final carry z-score signal
for each day.
For both carry and carry z-score methods, we build carry strategies from the ground up, starting with sectors and then
integrating the full target universe.
We explore the use of raw carry and carry z-score as signals, along with a quasi-‘binary’ transform of each score as either 1
or -1 for markets with positive or negative values, respectively. We also employ a rescaled ranking transform, where assets
with positive carry scores are assigned positive ranks ordinally on carry score, while those with negative scores receive
negative ranks. These transforms are applied cross-sectionally each day across all markets in the target universe. For
example, at the sector level the transforms are applied cross-sectionally to markets in each sector. At the full universe level
they’re applied across the full universe of markets.
Portfolio Construction
Naïve and optimized carry strategies were constructed in a cross-sectional and time series manner using both carry and
carry z-score signals.
Cross-sectional (XS) carry strategies hold long positions in the assets with carry signals above the cross-sectional average
on a given day, and short positions in assets below the average. All positions are scaled to the same expected volatility as
described below.
Time series (TS) carry strategies hold long or short positions in individual markets based on whether the carry or carry-z
signal is positive or negative, respectively. Thus, the cross-sectional strategy assumes little sector risk while the time-series
strategy will assume sector risk in proportion to the carry of the markets in the sector.
For the purpose of portfolio formation, the variance-covariance matrix was formed by combining a shorter-term volatility
estimate with an intermediate term correlation estimate. Volatility was computed using a 40-day half-life with a floor at 1/2 the
expanding-window long-term average standard deviation. Correlations were computed over a rolling window of 252 trading
days.
We explored inverse-volatility weighted portfolios at both the sector and the portfolio level. We also examined portfolios
formed by mean-variance optimization on the full asset universe, using carry and carry-z (and transforms) as expected
returns. For the optimization we employed a covariance matrix shrunk 50 percent towards the diagonal of the covariance
matrix, and then 20 percent toward the average of the off-diagonal covariances in the style of Ledoit-Wolf.

where is the vector of portfolio weights for each market, is the covariance matrix of returns among markets and µ is the
vector of signals (e.g., carry or carry-z scores) for each market.
In all cases, a 5-day moving average filter was applied to the weights prior to scaling the weights to target volatility in order
to reduce transaction costs from minor day-to-day fluctuations.
Portfolio weights were scaled each day to target 10 percent annualized volatility computed from current smoothed weights
and a simple one-year moving average.
Ensemble strategies were created by averaging either the cross-sectional or time series strategy weights for strategies
constructed from the various carry metrics (raw carry signals, binary carry signals, carry ranks, carry z-scores), and
optimization methods. Volatility scaling was always performed on final average weights after all averaging, as a final step.
Performance Analysis
Our performance analysis will focus on absolute and risk-adjusted performance metrics, as well as correlations to relevant
benchmarks. All returns are excess returns, that is they are in excess of the prevailing cash rate. We provide summary
performance tables for all simulations and select plots of cumulative equity. Simulation results are calculated from the first
date a signal can be calculated across all methods for apples-to-apples comparisons.
Importantly, all simulated results in this section are gross of transaction costs, except where noted. We focus on gross
returns for all but our final meta-ensemble strategies because costs are a confounding variable that makes it difficult to
contrast performance across strategy constituents, which is the focus of early sections. We provide estimated results net
of transaction costs for the final ensemble strategy. Note that we smooth all weights over the previous 5 trading days for all
strategies, so most strategies stand up well after accounting for estimated trade frictions.
Sector Carry Strategies

We start with an examination of how carry strategies have historically played out in Cross Sectional (Figure 2) and Time Series
(Figure 3) implementations. Sectors are divided into currencies, energy commodities, metals, global government bonds, and
equity indices. Market signal transforms are within each sector, and each market is weighted in proportion to the inverse of
its volatility. Portfolio weights are then scaled at the sector portfolio level to target 10 percent annualized standard deviation
by taking current smoothed weights and a long-term covariance estimate.

Figure 2. Performance by sector for Cross-Sectional Carry and Carry Z-Score strategies.
Carry Carry Carry Carry Carry Carry Z Carry Z Carry Z Carry Z Carry Z
Currencies Energies Metals Bonds Equities Currencies Energies Metals Bonds Equities
Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90
Annualized Return 2.57% 2.10% 0.33% 1.32% -2.06% 3.27% 2.05% 2.80% 2.38% 0.90%
Annualized Volatility 10.19% 10.17% 10.19% 10.24% 10.03% 10.23% 10.48% 10.27% 10.29% 10.35%
Annualized Sharpe 0.30 0.26 0.08 0.18 -0.16 0.37 0.25 0.32 0.28 0.14
Maximum Drawdown -48.94% -43.93% -57.11% -43.78% -66.28% -30.57% -32.60% -31.44% -34.06% -44.71%
Correlation with U.S.
0.03 -0.03 0.05 -0.06 -0.02 0.09 0.00 0.01 -0.02 0.02
Dollar Index
Correlation with WTI
0.20 0.15 -0.06 -0.04 0.05 -0.06 -0.10 -0.03 -0.01 0.03
Crude
Correlation with Nymex
0.20 0.05 0.17 -0.01 0.08 -0.03 -0.03 -0.15 -0.00 0.03
Copper
Correlation with US 10Y -0.17 -0.04 -0.01 0.22 0.00 0.05 0.02 0.05 0.02 -0.02
Correlation with S&P 500 0.25 0.03 -0.01 -0.03 -0.13 -0.03 -0.01 -0.12 -0.04 -0.08
Data from CSI Data. Authors’ calculations. Performance is HYPOTHETICAL and GROSS of all trading slippage and commissions. Past performance is not
indicative of future results. See also disclaimers at the end of the document for more information.
It’s clear that carry strategies have historically produced better performance across most sectors when sector exposures are
allowed to rise and fall in proportion to the carry of the markets within the sector.
Figure 3. Performance by sector for Time-Series Carry and Carry Z-Score strategies.
Carry Carry Carry Carry Carry Carry Z Carry Z Carry Z Carry Z Carry Z
Currencies Energies Metals Bonds Equities Currencies Energies Metals Bonds Equities
Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90
Annualized Return 2.68% 5.69% 0.37% 4.58% -0.03% 1.58% 5.57% 2.91% 4.32% 2.41%
Annualized Volatility 10.34% 10.40% 10.21% 10.45% 10.38% 10.46% 10.25% 10.67% 10.31% 10.23%
Annualized Sharpe 0.31 0.58 0.09 0.48 0.05 0.20 0.58 0.32 0.46 0.28
Maximum Drawdown -55.56% -30.21% -43.36% -38.98% -61.12% -50.02% -27.76% -43.82% -39.78% -34.39%
Correlation with U.S.
-0.13 -0.00 0.19 -0.03 -0.10 -0.13 -0.01 -0.14 0.02 0.04
Dollar Index
Correlation with WTI
0.09 0.24 -0.13 -0.11 0.15 -0.01 0.09 0.10 -0.04 0.15
Crude
Correlation with Nymex
0.09 -0.01 0.03 -0.13 0.09 0.03 -0.02 0.18 -0.05 0.21
Copper
Correlation with US 10Y -0.08 -0.02 -0.05 0.67 -0.16 0.04 -0.02 0.03 0.18 -0.18
Correlation with S&P 500 0.07 -0.03 -0.03 -0.13 0.00 -0.02 -0.05 -0.05 -0.06 0.54
Data from CSI Data. Authors’ calculations. Performance is HYPOTHETICAL and assumes $100 invested with reinvestment of profit GROSS of trading
slippage and commissions. Past performance is not indicative of future results. See also disclaimers at the end of the document for more information.
Perhaps surprisingly, time-series carry strategies are not consistently or materially more correlated on average to their
respective sectors than cross-sectional strategies. Bonds represent a notable exception, where the time-series bond
sector carry strategy has an average correlation of 0.64 to 10-year U.S. Treasury futures. This validates the intuition that the
term structure of government bonds was mostly positive over our simulation horizon. We examine ways to neutralize bond
exposure in later sections.
We next form portfolios by aggregating all sectors into a single portfolio, conditioned on type of carry signal and transform.
Each sector strategy receives equal risk weighting in the portfolios. An interesting observation from Figure 4 is that when
aggregated, carry z-score strategies produced consistent stronger performance than regular carry strategies. In general, this

suggests many markets exhibit consistent bias in carry over time, and it is more effective to condition positioning on carry
relative to this bias than on raw carry alone.
In addition, while the average Sharpe across all cross-sectional and time-series strategies is 0.64, the Sharpe of the ensemble
of all strategies is .84, suggesting attractive diversificiation from different formation methods and transforms. None of the
strategies exhibit material average correlation to equities, but some time-series strategies have been mildly correlated to
bonds for reasons discussed above.
Figure 5 reveals further context on how cross-sectional strategies have fared relative to time series strategies over time.
While time-series carry strategies show a relatively consistent character over time, cross-sectional version performance
peters out after 2012. Given that cross-sectional carry attempts to remove all directional risk and produce what some might
characterize as “pure alpha” without time-varying beta, one might induce that the market effectively arbitraged this market-
risk-free anomaly. Since 2012, investors have only been rewarded to taking on time-varying beta risk in carry strategies, at
least when constructed with equal sector weighting.
Figure 4. Performance of Cross-Sectional (XS) and Time-Series (TS) Equal Sector Weighted Carry and Carry z-score Strategies by Signal and
Transform.
Carry Multi Carry Z Multi Mult Strat Multi Carry Multi Carry Z Multi Mult Strat Multi Sector Meta
Sector XS Sector XS Sector XS Sector TS Sector TS Sector TS Ensemble
Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90

Annualized Return 2.46% 5.82% 5.28% 7.40% 7.47% 9.55% 9.09%
Annualized Volatility 10.21% 10.37% 10.22% 10.48% 10.39% 10.64% 10.52%
Annualized Sharpe 0.29 0.60 0.55 0.73 0.75 0.91 0.88
Maximum Drawdown -60.16% -34.00% -52.73% -30.40% -32.08% -30.81% -34.09%
Correlation with S&P 500 0.05 -0.13 -0.05 -0.05 0.14 0.06 0.01
Correlation with US 10Y 0.01 0.06 0.04 0.18 0.03 0.14 0.12
Figure 5. Cumulative Excess Returns of Cross-Sectional (XS) and Time-Series (TS) Equal Sector Weighted Carry and Carry Z-Score Strategies
by Signal and Transform.

Carry on the Full Liquid Futures Universe
The balance of our analysis will form portfolios from the full universe of liquid futures without reference to sectors. We start
with simple methods of portfolio construction and compare against optimized and constrained portfolios.
It is not clear how one might construct cross-sectionally neutral portfolios once you move on from sector strategies. As a
result, our focus narrows to time-series versions.
Inverse Volatility Weighted Portfolios
Recall that to form inverse volatility weighted portfolios, at each day we divide the carry signal (raw or transformed) for each
market, by the market’s current estimated volatility. This ensures each market’s weight is proportional to both the strength
(and direction) of its carry signal and inversely proportional to its volatility. Markets with larger signals and lower estimated
volatilty will earn a larger absolute weighting in the portfolio than markets with smaller signals and higher volatility. We then
scale these weights so that the overall portfolio has an expected volatility of 10 percent annualized conditioned on long-term
covariance estimates.
Figure 6. Performance of Time-Series (TS) Inverse Volatility Weighted Carry and Carry Z-Score Strategies by Signal and Transform.
Carry Binary Carry Rank Carry Carry-Z Binary Carry-Z Rank Carry-Z
Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90

Annualized Return 6.77% 8.21% 9.15% 7.75% 8.00% 8.21%
Annualized Volatility 10.52% 10.17% 10.15% 10.29% 10.14% 10.10%
Annualized Sharpe 0.68 0.83 0.91 0.78 0.81 0.83
Maximum Drawdown -21.55% -23.99% -24.52% -27.91% -27.78% -33.81%
Correlation with S&P 500 -0.15 -0.18 0.06 0.05 0.07 0.33
Correlation with US 10Y 0.06 0.14 0.15 -0.05 -0.06 -0.04
Figure 7. Cumulative Excess Returns of Time-Series (TS) Inverse Volatility Weighted Carry and Carry Z-Score Strategies by Signal and
Transform.

Evaluating Figure 6 and Figure 7 we note that, in contrast with sector-weighted versions, raw carry strategies fare a little better
than carry z-score versions. This is mostly due to performance in the recent period, as they were qualitatively indistinguishable
through approximately 2016. Still, even in recent years we can not reject the view that the observed performance differential
is an artifact of statistical noise. Indeed, we may check back in 5 years to see quite a different rank ordering than we note in
our snapshot at the end of 2023.
Intuitively, when we aggregate weights across all transforms, and again across all transforms and signal types, it’s clear the
raw carry strategies have slightly dominated carry z-score strategies in the last few years. Per Figure 8 and Figure 9, due to
relatively low correlations between the carry and carry z-score inverse volatility ensemble strategies, the combination exhibits
better performance than either. Indeed, the IV ensemble of all Inverse Volatility weighted strategies sports a gross Sharpe
ratio of 0.98 with a 25 percent maximum drawdown at a 10 percent annualized target volatility.
Generally, portfolios that enable all markets to directly compete based on their carry and volatility metrics outperform those
constructed from sector-specific strategies, where markets only vie for allocations within their respective sectors. This makes
sense: while markets might earn a high score relative to other members of its sector, if the carry score of the sector is weak,
the markets in the sector should earn lower weight in the portfolio.
Figure 8. Performance of Time-Series (TS) Inverse Volatility Weighted Ensemble Carry Strategies.
IV Ensemble Carry IV Ensemble Carry-Z IV Ensemble
Inception Date MAR 90 MAR 90 MAR 90
Annualized Return 8.76% 8.41% 10.46%
Annualized Volatility 10.29% 10.15% 10.47%
Annualized Sharpe 0.87 0.85 1.00
Maximum Drawdown -19.00% -30.20% -28.16%
Correlation with S&P 500 -0.09 0.16 0.04
Correlation with US 10Y 0.13 -0.05 0.06
Figure 9. Cumulative Excess Returns of Time-Series (TS) Inverse Volatility Weighted Ensemble Carry Strategies.

Optimized Portfolios
Let’s pause quickly to examine the assumptions that are implied by inverse volatility weighted portfolios. Since weights are
exclusively a function of a market’s carry signal and volatility, there is no accounting for correlations. Thus, if there are more
equities than energies in the portfolio then all things equal the portfolio will be overweight equities. In fact, inverse volatility
weighted portfolios are only optimal if there is no useful information in correlations between the markets. This condition is
met if correlations are effectively random from day to day, or if all markets always have the same pairwise correlations. Since
neither of these conditions typically hold, we would expect portfolios that account for correlations - (thoughtfully) optimized
portfolios - to outperform inverse volatility portfolios.
We simulated portfolios formed by passing each market’s signal along with an up-to-date estimated covariance matrix into
a mean-variance optimizer. As noted above, to mediate known challenges with optimization, we stabilized the covariance
matrix using standard methods.
For further intuition into this solution, consider that an unconstrained mean-variance optimizer effectively takes the
following steps. First, the covariance matrix is used to change the basis of the problem from asset space into independent
“eigenportfolio” space, where each eigenportfolio can be interpreted as a portfolio of assets that represents a statistically
independent risk factor. For example, one eigenportfolio may capture equity beta while another eigenportfolio may capture
the return spread between energies and metals. Each eigenportfolio is then scaled to the same volatility level and held in
proportion to its volatility-scaled carry. In doing so, the process effectively applies time-series carry to the independent risk
factors, recognizing that many of those risk factors will capture pertinent cross-sectional information!
Inverse volatility weighting, then, simply collapses this solution to assume that each market represents an independent risk
factor and that cross-sectional effects can be ignored.
Figure 10. Performance of Time-Series (TS) Optimized Carry and Carry Z-Score Strategies by Signal and Transform.
Carry Binary Carry Rank Carry Carry-Z Binary Carry-Z Rank Carry-Z
Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90
Correlation with S&P 500 -0.11 -0.09 0.05 0.03 0.05 0.20
Correlation with US 10Y 0.09 0.15 0.23 -0.00 -0.02 0.07
As expected, optimized strategies in general compare favourably to inverse volatility weighted strategies, but the differences
are fairly moderate. Scrutinizing Figure 10 and Figure 11 there’s no clear benefit to either raw carry or carry z-score methods.
Regular carry z-score was a notable exception, with an unfortunate stretch of weakness over the 2006 - 2012 period. We
leave it to the reader to speculate about the source of this underperformance. Still, over the long-term the strategies are
mostly statistically indistinguishable, with all methods showing promise.

Figure 11. Cumulative Excess Returns of Time-Series (TS) Optimized Carry and Carry Z-Score Strategies by Signal and Transform.
When we combine weights across all inverse volatility and optimized strategies, and again across all transforms and signal
types in Figure 12 and Figure 13, the complementarity of the methods shines through. While the average of the Sharpe ratios
of the carry ensemble and the carry z-score ensemble are .93 and .89 respectively, the Sharpe of the meta ensemble of all
sub-strategies is 1.09. Adding to the strong performance, the strategies all show effectively zero correlation to equities and
very low correlation to bonds.
Figure 12. Performance of Time-Series (TS) Optimized Ensemble Carry Strategies.
Optim Ensemble Optim Ensemble

Optim Ensemble
Carry Carry-Z
Inception Date MAR 90 MAR 90 MAR 90

Annualized Return 9.54% 9.17% 11.57%
Annualized Volatility 10.42% 10.45% 10.54%
Annualized Sharpe 0.93 0.89 1.09
Maximum Drawdown -21.04% -27.28% -25.05%
Correlation with S&P 500 -0.06 0.10 0.03
Correlation with US 10Y 0.18 0.01 0.12

Figure 13. Cumulative Excess Returns of Time-Series (TS) Optimized Ensemble Carry Strategies.
While average correlations to stocks and bonds are low, a more nuanced picture emerges from an analysis of rolling
correlations over time. Figure 14 demonstrates that the strategy spends multi-year periods of time being long equity and
Treasury beta (i.e. with positive correlations to stocks and bonds), followed by long periods being short. In other words, it
seems to align with long-term regimes. While some investors may be agnostic about this character when holding the strategy
in isolation, others may prefer a more neutral disposition to stocks and bonds for diversification purposes.
Constrained Portfolios
Given a common preference for diversifiers to act as diversifiers to common betas most of the time, we explored ways to
structurally neutralize expected exposure to stocks and bonds at portfolio formation. A straightforward but sub-optimal
approach would be to explicitly trade equity markets and bond markets cross-sectionally in portfolios. As discussed above,
this often leads to situations where markets with strong positive carry are held short. Instead, we can lean on correlation
relationships between all the assets in the portfolio to form portfolios that simultaneously seek to maximize total carry at
minimal risk, while constraining aggregate expected portfolio beta to equities and Treasuries to zero. Importantly, market
betas are updated daily to represent current market relationships using a rolling window of 252 trading days.

The optimization problem can be formulated as follows:
Figure 14. Rolling Pearson Correlations: Time-Series (TS) Optimized Ensemble Carry Strategies vs S&P 500 and U.S. 10-Year Treasury
Futures.
Where and are vectors of portfolio betas to the S&P 500 futures and 10-year Treasury futures,
respectively. The constraint set therefore constrains the portfolio to be neutral to these benchmarks. The portfolio is otherwise
unconstrained.
We neutralize both inverse volatility weighted and mean-variance optimal portfolios. To neutralize inverse volatility weighted
portfolios, we employed mean-variance optimization with beta constraints as above, but set off-diagonal covariances to zero.
Note that we need a year (252 trading days) to calculate market betas, so simulations of beta neutralized portfolios start in
January 1991 rather than March 1990, which is why the long-term performance numbers in Figure 15 below do not match
those in the table above.

Figure 15. Performance of Time-Series (TS) Optimized Ensemble Carry Strategies.

Optimization Optimization IV Ensemble Optimization Meta Ensemble
Ensemble Equity Ensemble Treasury Treasury Equity Ensemble Treasury Treasury Equity Meta Ensemble
Neutral Neutral Neutral Equity Neutral Neutral
Inception Date JAN 91 JAN 91 JAN 91 JAN 91 JAN 91 JAN 91
Correlation with S&P 500 -0.02 0.10 -0.04 -0.03 -0.04 0.05
Correlation with US 10Y 0.19 0.05 0.05 0.05 0.05 0.09
Figure 16. Cumulative Excess Returns of Time-Series (TS) Optimized Ensemble Carry Strategies.
By taking advantage of correlation relationships between markets, the performance of equity and Treasury-neutral portfolios
rival the performance of unconstrained strategies. This is quite an attractive feature, as the beta-neutralized strategy delivers
on our objective of substantially reducing periods of sustained exposure to the target markets. Note that the rolling correlations
between the sector constrained strategy returns and core markets in Figure 17 have much smaller peaks and troughs to
target betas, and revert toward our target of zero much more quickly. In fact, the correlations we see mostly represent basis
risk of holding equity and bond markets with imperfect correlation to S&P 500 and 10-year Treasuries, which is unavoidable.

Figure 17. Rolling Pearson Correlations: Time-Series (TS) Equity and Treasury Neutral Carry Strategies vs S&P 500 and U.S. 10-Year Treasury
Futures.
Gross Versus Net Returns
So far we’ve discussed performance gross of expected trading costs so as not to confound the many take-aways from
examining the various dimensions of carry across signal types, transforms, sectors, and optimization methods. Of course
ultimately the value of carry strategies is a function of economic returns, net of costs.
In this section we compare gross returns to production-ready ensemble strategies to returns after subtracting estimated
trade slippage and commissions. Importantly, trade-netting plays a material role in minimizing trading slippage. By offsetting
buy and sell orders for the same assets across various carry strategies, we can significantly reduce transaction costs and
market impact. This efficiency not only lowers operational overhead but also enhances portfolio performance by minimizing
unnecessary trades.
Recall also that for all strategies we smooth portfolio weights using a 5-day moving average, which also reduces unnecessary
noise trading.
Per Figure 18 and Figure 19, after trade-netting and smoothing, and given our highly liquid investment universe estimated
trade frictions reduce returns by about 1 percent per year, which works out to approximately 10 basis points in annualized
Sharpe ratio.
For the balance of this paper we’ll refer to the Meta Ensemble Treasury Equity Neutral Carry Strategy as simply the Carry
Strategy.

Figure 18. Performance of Time-Series (TS) Carry Strategies: Gross vs Net.

Unconstrained Meta Ensemble Meta Ensemble Meta Ensemble
Unconstrained
Meta Ensemble Treasury Equity Treasury Equity Treasury Equity
Meta Ensemble Net
Gross Neutral Gross Neutral Net Neutral
Inception Date JAN 91 JAN 91 JAN 91 JAN 91 JAN 91
Annualized Return 10.90% 9.83% 10.07% 9.05% 10.15%
Annualized Volatility 10.52% 10.47% 10.58% 10.51% 10.54%
Annualized Sharpe 1.04 0.95 0.96 0.88 0.97
Maximum Drawdown -25.05% -27.01% -22.60% -21.71% -20.99%
Correlation with S&P 500 0.04 0.06 -0.03 -0.03 -0.04
Correlation with US 10Y 0.12 0.10 0.05 0.05 0.05
Data from CSI Data. Authors’ calculations. GROSS performance is HYPOTHETICAL and GROSS of all trading slippage and commissions. NET performance
is HYPOTHETICAL and NET of estimated trading slippage and commissions. Past performance is not indicative of future results. See also disclaimers at
the end of the document for more information.
Figure 19. Cumulative Excess Returns of Time-Series (TS) Carry Strategies: Gross vs Net.
Data from CSI Data. Authors’ calculations. GROSS performance is HYPOTHETICAL and GROSS of all trading slippage and commissions. NET performance
is HYPOTHETICAL and assumes $100 invested with reinvestment of profit NET of estimated trading slippage and commissions. Past performance is not
Scenario Analysis
Carry in the Worst Benchmark Quarters
To assess the diversification potential of carry strategies, we analyze their performance conditional on the worst quarterly
returns for the S&P 500 Total Return Index and the Bloomberg Aggregate Total Return Index. Figure 20 and Figure 21 show
the annualized excess returns of the final optimized Carry Strategy, a trend following strategy, and futures representing the
underlying asset class itself (S&P 500 futures and 10-Year Treasury Note futures, respectively) sorted into quintiles based on
quarterly asset class performance.
The results demonstrate that both carry and trend returns are largely independent of the performance of either equities or
bonds, and each other. Average carry returns are consistently positive across all quintiles, with little variation based on the
magnitude of benchmark losses. This suggests that carry can provide meaningful diversification and positive returns even in
the most challenging market environments (Koijen et al. 2018) (Moskowitz, Ooi, and Pedersen 2012) (Baltas 2017).

Figure 20. Annualized Excess Returns of Carry (Net), Trend Replication (Net), and S&P 500 Futures Conditioned on S&P 500 Total Return
Quintile.
Carry (Net) S&P 500 Futures Trend (Net)
Quintile Summary
60% 58%
40%
Annualized Return
26%
20% 17% 15%

12% 12% 11%
9% 10% 8%
6% 5%
2%
0%
-1%
-20%
-34%
Worst Quintile 2 3 4 Best Quintile
Quintile
Data from CSI Data, Standard & Poors. Authors’ calculations. S&P 500 futures are GROSS of roll costs. Carry and Trend performance is HYPOTHETICAL
and NET of estimated trading slippage and commissions. Past performance is not indicative of future results. See also disclaimers at the end of the
document for more information.
Figure 21. Annualized Excess Returns of Carry (Net), Trend Replication (Net), and U.S. 10Y Treasury Futures Conditioned on Bloomberg Agg
Total Return Quintile.
Carry (Net) US 10Y Treasury Futures Trend (Net)
Quintile Summary
25% 24%
23%
20% 18%
15%
12%
10%
Annualized Return
10% 9%
10%
7%
5% 5%
5%
2% 3%
1%
0%
-5% -4%
-10%
-13%
-15%
Worst Quintile 2 3 4 Best Quintile
Quintile
Data from CSI Data, Bloomberg. Authors’ calculations. U.S. 10Y Treasury futures are GROSS of roll costs. Carry and Trend performance is HYPOTHETICAL
and NET of estimated trading slippage and commissions. Past performance is not indicative of future results. See also disclaimers at the end of the
document for more information.
Carry During Benchmark Drawdowns
While the quintile analysis is informative, investors are often most concerned with performance during extended drawdowns
rather than single quarter returns. To address this, Figure 22 to Figure 24 plot the cumulative excess return of the final Carry
(Net) Strategy and S&P 500 futures during the three worst drawdowns for the S&P 500 Total Return Index. Figure 25 to Figure
27 do the same for U.S. 10Y Treasury futures conditioned on quarterly quintile returns of the Bloomberg Barclays 7-10 Year
Government Bond Total Return Index (Koijen et al. 2018) (Moskowitz, Ooi, and Pedersen 2012) (Baz et al. 2015).
Carry mostly shrugged off the Global Financial Crisis of 2008 (Figure 22) and the ‘Tech Wreck’ following the late 1990s
technology and telecom bubble (Figure 23), demonstrating a perhaps surprising ability to provide “crisis alpha”. The Carry
Strategy also held in relatively well in the fastest and most severe equity drawdown during COVID (Figure 24).

Figure 22. Cumulative Excess Growth of S&P 500 Futures and Carry Strategy (Net) from October 10, 2007 to August 16, 2012.
Data from CSI Data and Standard & Poors. Authors’ calculations. Carry and Trend performance is HYPOTHETICAL and NET of estimated trading slippage
and commissions. Past performance is not indicative of future results. See also disclaimers at the end of the document for more information.
Figure 23. Cumulative Excess Growth of S&P 500 Futures and Carry Strategy (Net) from March 27, 2000 to October 26, 2006.
Figure 24. Cumulative Excess Growth of S&P 500 Futures and Carry Strategy (Net) from February 20, 2020 to August 10, 2020.

The bond drawdown analysis is particularly relevant given the historic losses in fixed income in 2022. Figure 25 shows that
carry was mostly resilient during this period, and returns to carry almost perfectly mirror returns to bonds for most of this
inflationary shock. Despite the Carry Strategy’s attempts at neutralizing beta to Treasuries, strategy returns have been mixed
in the other two worst Treasury market drawdowns, with one positive and one slightly negative result over relatively short
horizons.
Taken together, the scenario analysis suggests that futures carry strategies can play a unique role in investor portfolios,
providing meaningful returns and diversification during precisely the periods where traditional asset classes suffer the most.
By generating crisis alpha, carry can help investors hedge left-tail risk and smooth portfolio drawdowns.
Figure 25. Cumulative Excess Growth of U.S. 10Y Treasury Futures and Carry strategy (Net) from August 05, 2020 to December 29, 2023.
Data from CSI Data and Bloomberg. Authors’ calculations. Performance is HYPOTHETICAL and GROSS of all trading slippage and commissions. Past
performance is not indicative of future results. See also disclaimers at the end of the document for more information.
Figure 26. Cumulative Excess Growth of U.S. 10Y Treasury Futures and Carry strategy (Net) from December 31, 2008 to June 28, 2010.

Figure 27. Cumulative Excess Growth of U.S. 10Y Treasury Futures and Carry strategy (Net) from October 18, 1993 to May 04, 1995.
Stacking Case Studies
The concept of return stacking, or combining uncorrelated return streams to enhance risk-adjusted performance, is a core
principle of modern portfolio construction. Given the diversifying properties of carry demonstrated in the prior section, and
the well documented low correlation of managed futures trend to major indices, we now explore the potential benefits of
stacking carry and trend strategies on top of traditional stock and bond allocations.
Figure 28 shows the cumulative growth of the S&P 500 and hypothetical portfolios investing 100 percent in the S&P 500 and
collateralizing carry and trend strategies stacked on top. In simulation, stacking carry and/or trend on top of U.S. equities
produced substantial performance gains with minimal excess volatility, and the potential for similar or smaller drawdowns.
Figure 29 conducts the same hypothetical analysis for the Bloomberg Barclays Aggregate Bond Index. The improvement in
simulated performance is even more striking, with the stacked bond portfolio delivering large boosts in returns with minimal
impact on drawdowns. The 2022 bond drawdown is also significantly reduced in simulation by the strong positive returns to
carry and trend during this challenging period for fixed income.
Figure 30 summarizes the hypothetical return stacking results, showing that carry and trend each may have the potential
to add material excess return to the underlying asset class while maintaining a low correlation and beta. The impact is
particularly meaningful for bonds in simulation.
These case studies demonstrate the potential for carry and trend to enhance traditional portfolios. By stacking uncorrelated
risk premia across global futures markets, these strategies can raise the efficient frontier of stock and bond holdings.

Figure 28. Cumulative Growth of S&P 500 Total Return Index Stacked with Carry and Trend.
Data from CSI Data and Standard & Poors. Authors’ calculations. Trend and Carry performance is HYPOTHETICAL and NET of estimated trading slippage
Figure 29. Cumulative Growth of Bloomberg Aggregate Total Return Index Stacked with Carry and Trend.
Data from CSI Data and Bloomberg. Authors’ calculations. Trend and Carry performance is HYPOTHETICAL and NET of estimated trading slippage and
commissions. Past performance is not indicative of future results. See also disclaimers at the end of the document for more information.

Figure 30. Performance of S&P 500 Total Return Index and Bloomberg Aggregate Total Return Index Stacked with Carry and Trend.
S&P 500
Bloomberg Bloomberg
Bloomberg 50/50 S&P 500
S&P 500 S&P 500 Total Return Bloomberg Aggregate Total and Bloomberg
S&P 500 Aggregate Aggregate Total
Total Return Total Return Stacked Aggregate Return Stacked Aggregate
Total Return Total Return Return Trend
50/50 Carry & Stacked Carry &
Carry Stack Trend Stack 50/50 Carry Total Return Stack
Carry Stack Trend Trend
& Trend
Inception Date JAN 91 JAN 91 JAN 91 JAN 91 JAN 91 JAN 91 JAN 91 JAN 91 JAN 91
Annualized Return 10.06% 20.04% 18.63% 19.61% 4.68% 14.07% 12.71% 13.65% 17.11%
Annualized Volatility 18.09% 20.66% 21.59% 19.99% 4.95% 11.93% 13.65% 10.85% 13.05%
Annualized Sharpe 0.62 0.99 0.90 1.00 0.95 1.16 0.95 1.23 1.28
Maximum Drawdown -55.25% -56.61% -49.26% -51.71% -18.43% -29.42% -21.29% -20.26% -28.83%
Data from CSI Data, Standard & Poors and Bloomberg. Authors’ calculations. Performance is HYPOTHETICAL and NET of estimated trading slippage and
commissions. Past performance is not indicative of future results. See also disclaimers at the end of the document for more information.
While the appropriate sizing will vary based on investor objectives and constraints, this framework of return stacking represents
a compelling way for investors to utilize alternative risk premia to potentially improve portfolio outcomes. By combining the
best of both worlds, return stacking may enable investors to maintain their strategic asset allocation while still gaining the
benefits of diversifying alternative strategies.
Conclusion
The exploration of futures carry strategies within this study illuminates the nuanced and multifaceted nature of carry as a
source of potential returns across various asset classes. Through a meticulous examination of carry and its manifestations
in equities, bonds, commodities, and currencies, we have endeavored to dissect the economic rationale underpinning carry
returns, thereby offering a granular understanding of its potential role in portfolio optimization (Koijen et al. 2018) (Baltas 2017)
(Asness, Moskowitz, and Pedersen 2013) (Frazzini and Pedersen 2014) (Ilmanen 2011).
Our analysis, spanning over three decades, leverages a comprehensive dataset to delve into the intricacies of carry, employing
both classic and relative value approaches to capture the essence of carry across a diverse range of market conditions. The
methodologies employed—ranging from simple cross-sectional and time series strategies to more sophisticated ensemble
and optimized portfolio constructions—reveal the dynamic interplay between carry signals and market behavior, highlighting
the importance of adaptability in strategy formulation.
The findings suggest that carry strategies, when judiciously implemented, may offer a valuable component to a well-
diversified portfolio, potentially enhancing returns while managing risk exposure. The comparative analysis of inverse volatility
weighted and optimized portfolios underscores the significance of accounting for market correlations, which, when ignored,
may inadvertently skew portfolio allocations. Moreover, the introduction of constraints to neutralize exposure to key market
benchmarks such as the S&P 500 and U.S. Treasury futures further refines the approach, aiming to deliver a more balanced
and risk-aware portfolio composition.
It is imperative to acknowledge that the performance of carry strategies, as detailed in this study, is presented on a gross
basis, with a subsequent examination of net returns after accounting for estimated trading costs. This distinction is crucial,
as it highlights the operational efficiencies achievable through trade netting and weight smoothing, which may mitigate the
impact of transaction costs on overall returns.
In conclusion, while the empirical evidence presented herein suggests that futures carry strategies may contribute positively
to portfolio performance, it is essential to approach these findings with a measured perspective. The potential benefits

of carry must be weighed against the inherent risks and uncertainties characteristic of financial markets. Investors are
encouraged to consider carry strategies as part of a broader, diversified investment approach, always mindful of the dynamic
and evolving nature of market conditions.
This study, by dissecting the complexities of carry and its practical application in portfolio management, aims to contribute
to the ongoing discourse on investment strategy optimization. As markets continue to evolve, so too will the strategies
employed by investors to navigate them. Carry, with its foundational role in the economic underpinnings of asset returns,
remains a compelling area for further exploration and potential inclusion in the sophisticated investor’s toolkit.

Disclaimers
Confidential and proprietary information.

The contents hereof may not be reproduced or disseminated without the express written permission of ReSolve Asset
Management Inc. and/or ReSolve Asset Management SEZC(Cayman) (“ReSolve”). These materials do not purport to be
exhaustive and although the particulars contained herein were obtained from sources ReSolve believes are reliable, ReSolve
does not guarantee their accuracy or completeness. The material has been provided to you solely for information purposes
and does not constitute an offer or solicitation of an offer or any advice or recommendation to purchase any securities
or other financial instruments and may not be construed as such. This information is not intended to, and does not relate
specifically to any investment strategy or product that ReSolve offers. It is being provided merely to provide a framework
to assist in the implementation of an investor’s own analysis and an investor’s own view on the topic discussed herein. The
investment strategy and themes discussed herein may be unsuitable for investors depending on their specific investment
objectives and financial situation.
Hypothetical Performance Results

HYPOTHETICAL PERFORMANCE RESULTS HAVE MANY INHERENT LIMITATIONS, SOME OF WHICH ARE DESCRIBED
BELOW. NO REPRESENTATION IS BEING MADE THAT ANY ACCOUNT WILL OR IS LIKELY TO ACHIEVE PROFITS
OR LOSSES SIMILAR TO THOSE SHOWN. IN FACT, THERE ARE FREQUENTLY SHARP DIFFERENCES BETWEEN
HYPOTHETICAL PERFORMANCE RESULTS AND THE ACTUAL RESULTS SUBSEQUENTLY ACHIEVED BY ANY
PARTICULAR TRADING PROGRAM. ONE OF THE LIMITATIONS OF HYPOTHETICAL PERFORMANCE RESULTS IS THAT
THEY ARE GENERALLY PREPARED WITH THE BENEFIT OF HINDSIGHT. IN ADDITION, HYPO- THETICAL TRADING
DOES NOT INVOLVE FINANCIAL RISK, AND NO HYPOTHETICAL TRADING RECORD CAN COMPLETELY ACCOUNT FOR
THE IMPACT OF FINANCIAL RISK IN ACTUAL TRADING. FOR EXAMPLE, THE ABILITY TO WITHSTAND LOSSES OR TO
ADHERE TO A PARTICULAR TRADING PROGRAM IN SPITE OF TRADING LOSSES ARE MATERIAL POINTS WHICH CAN
ALSO ADVERSELY AFFECT ACTUAL TRADING RESULTS. THERE ARE NUMEROUS OTHER FACTORS RELATED TO THE
MARKETS IN GENERAL OR TO THE IMPLEMENTATION OF ANY SPECIFIC TRADING PROGRAM WHICH CANNOT BE
FULLY ACCOUNTED FOR IN THE PREPARATION OF HYPOTHETICAL PERFORMANCE RESULTS AND ALL OF WHICH
CAN ADVERSELY AFFECT ACTUAL TRADING RESULTS.
Not an Offer or Solicitation

This document is for informational purposes only and does not constitute an offer or invitation to subscribe for or purchase
any securities, products or services. This material is not intended for distribution to, or use by, any person or entity in any
jurisdiction or country where such distribution or use would be contrary to local law or regulation. The securities, products
and services described herein may not be available to all persons in all jurisdictions.
Not Investment Advice

This document is provided for informational and discussion purposes only and is not intended to be, and should not be
construed as, investment, tax or legal advice, an opinion regarding the appropriateness or suitability of any investment, or
an offer to sell, or a solicitation of an offer to buy an interest in any security, including an interest in any investment vehicle
managed or advised by ReSolve. This document is not and should not be construed as a recommendation to buy or sell
any securities. There is no assurance that any securities discussed herein will remain in any account’s portfolio at the time
you receive this document or that securities sold have not been repurchased. Recipients should consult their own advisors
regarding the legal, tax, and financial suitability of investing in any securities or following any investment strategies.
General Risks of Investing

All investments in securities involve risks, including the risk of losing all of the invested capital. Past performance of a security,
market, portfolio or investment strategy does not guarantee future results. There is no assurance that a portfolio or strategy
will achieve its investment objective. Diversification does not ensure profit or protect against loss in declining markets.

References
Asness, Clifford S, Tobias J Moskowitz, and Lasse Heje Pedersen. 2013. “Value and Momentum Everywhere.” The Journal
of Finance 68 (3): 929–85.
Bakshi, Gurdip, Xiaohui Gao, and Alberto G Rossi. 2017. “Commodity Futures Prices: More Evidence on Forecast Power,
Risk Premia and the Theory of Storage.” The Review of Financial Studies 30 (10): 3337–82.
Baltas, Nick. 2017. “Optimising Cross-Asset Carry.” Available at SSRN 2968677.
Baz, Jamil, Nicolas Granger, Campbell R Harvey, Nicolas Le Roux, and Sandy Rattray. 2015. “Dissecting Investment
Strategies in the Cross Section and Time Series.” Available at SSRN 2695101.
Boons, Martijn, Fernando M Duarte, Frans de Roon, and Marta Szymanowska. 2020. “Time-Varying Inflation Risk and Stock
Returns.” Journal of Financial Economics 136 (2): 444–70.
Frazzini, Andrea, and Lasse Heje Pedersen. 2014. “Betting Against Beta.” Journal of Financial Economics 111 (1): 1–25.
Fuertes, Ana-Maria, Jo"elle Miffre, and Georgios Rallis. 2010. “Tactical Allocation in Commodity Futures Markets: Combining
Momentum and Term Structure Signals.” Journal of Banking & Finance 34 (10): 2530–48.
Gorton, Gary B, Fumio Hayashi, and K Geert Rouwenhorst. 2013. “The Fundamentals of Commodity Futures Returns.”
Review of Finance 17 (1): 35–105.
Ilmanen, Antti. 2011. “Expected Returns: An Investor’s Guide to Harvesting Market Rewards.”
Koijen, Ralph SJ, Tobias J Moskowitz, Lasse Heje Pedersen, and Evert B Vrugt. 2018. “Carry.” Journal of Financial Economics
127 (2): 197–225.
Moskowitz, Tobias J, Yao Hua Ooi, and Lasse Heje Pedersen. 2012. “Time Series Momentum.” Journal of Financial Economics
104 (2): 228–50.
Szymanowska, Marta, Frans De Roon, Theo Nijman, and Rob Van Den Goorbergh. 2014. “An Anatomy of Commodity
Futures Risk Premia.” The Journal of Finance 69 (1): 453–82.

Managed-Futures-Carry-Whitepaper

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Managed-Futures-Carry-Whitepaper

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Managed-Futures-Carry-Whitepaper

Uploaded by

Copyright:

Available Formats

MANAGED

2 ReSolve Asset Management

Carry in Financial Markets 4

Economic Rationale for Carry Returns 5

Measuring Carry in Practice 5

Constructing Carry Strategies 7

Sector Carry Strategies 12

Carry on the Full Liquid Futures Universe 15

Stacking Case Studies 27

3 ReSolve Asset Management

Carry in Financial Markets

4 ReSolve Asset Management

Asset Class Benefit of Holding Cost of Holding

Economic Rationale for Carry Returns

Measuring Carry in Practice

5 ReSolve Asset Management

Here’s a straightforward example using the S&P 500:

Your profit or loss is calculated as:

F + Dividends − 5000 × (1 + Interest Rate)

If we simplify this further, we can say:

6 ReSolve Asset Management

Constructing Carry Strategies

7 ReSolve Asset Management

8 ReSolve Asset Management

Figure 1. Inception Dates for Markets in our Experimental Universe

9 ReSolve Asset Management

The pairwise average carry is then given as:

• carry definition: classic and relative value carry

10 ReSolve Asset Management

11 ReSolve Asset Management

Sector Carry Strategies

12 ReSolve Asset Management

13 ReSolve Asset Management

Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90

14 ReSolve Asset Management

Carry on the Full Liquid Futures Universe

Inverse Volatility Weighted Portfolios

Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90

15 ReSolve Asset Management

IV Ensemble Carry IV Ensemble Carry-Z IV Ensemble

Inception Date MAR 90 MAR 90 MAR 90

Annualized Return 8.76% 8.41% 10.46%

Annualized Volatility 10.29% 10.15% 10.47%

Annualized Sharpe 0.87 0.85 1.00

Maximum Drawdown -19.00% -30.20% -28.16%

Correlation with S&P 500 -0.09 0.16 0.04

Correlation with US 10Y 0.13 -0.05 0.06

16 ReSolve Asset Management

Inception Date MAR 90 MAR 90 MAR 90 MAR 90 MAR 90 MAR 90

Annualized Return 8.80% 7.92% 8.90% 8.59% 8.81% 8.91%

Annualized Volatility 10.59% 10.37% 10.32% 10.45% 10.39% 10.49%

Annualized Sharpe 0.85 0.79 0.88 0.84 0.86 0.87

Maximum Drawdown -26.01% -22.06% -26.02% -26.77% -26.85% -29.88%

Correlation with US 10Y 0.09 0.15 0.23 -0.00 -0.02 0.07

17 ReSolve Asset Management

Figure 12. Performance of Time-Series (TS) Optimized Ensemble Carry Strategies.

Optim Ensemble Optim Ensemble

Inception Date MAR 90 MAR 90 MAR 90