A founder, operator and lifelong autodidact.
Most days (and nights) you can find me investing at King Whale Capital located in Charlottesville, Virginia.
In my spare time I enjoy fly fishing and spending time with family & friends.
Get in touch with me at td@kingwhale.co.
“Mortality rates don’t swing with the Dow, and a Picasso’s worth doesn’t tank with tech stocks…”
- Unknown
When exploring uncorrelated returns, it’s worth distinguishing between strategies that thrive on explicit “directional bets” — predictions tied to an asset’s inherent fate — and those that don’t.
Some mainstream and common approaches to source uncorrelated returns include quant-driven HFT, merger arbitrage, or market-neutral strategies. These methods focus on capturing spreads or inefficiencies between assets, exploiting price discrepancies across international markets, leveraging deep M&A expertise, or pairing a long position with a short position to play their relative performance — insulating the bet from broader market swings. Such strategies, whether powered by algorithms or deep industry expertise, are inherently non-directional, seeking gains agnostic to the underlying assets’ absolute trajectory.
What sets directional positions apart is their embrace of an asset’s singular path, free from the short legs or market-neutral buffers that drive arbitrage plays. These investments stake a claim on an outcome while remaining insulated from the tides of broader economic currents. This directionality is a feature, not a bug: it opens the door to outsized gains when predictions hold true, compounding returns in ways non-directional strategies rarely can. Of course, that same exposure is not risk free and carries certain risks but it’s precisely this boldness that makes them intriguing.
I’d like to examine a couple of directional uncorrelated assets below, including the fine art and life settlement market. If I discover additional directional uncorrelated assets that I believe are worth exploring (or interesting), I’ll continue to refine and expand on this topic accordingly.
Fine Art
Fine Art stands out as one such directional uncorrelated asset. Two studies—one published by Columbia Business School (2015) and another from Citi Global Perspectives & Solutions (2020)—provide data on fine art as an asset class. I’ll focus on their reporting of annualized returns and correlations with equity markets.
Overview of the Studies
The Columbia Business School paper analyzes fine art returns for specific artist groups and individuals — Surrealists (for example Salvador Dalí, Max Ernst, and other frequently traded Surrealists), Impressionists (for example Claude Monet, Vincent Van Gogh, and other frequently traded Impressionists), Pablo Picasso, and Pierre-Auguste Renoir — using auction data from 1985 to 2014. It employs a bootstrap method, a resampling technique corrected for log-transformation bias, to estimate annualized returns and correlations with the S&P 500 (TR).
The Citi paper examines broad fine art categories — All Art, Contemporary Art, and Impressionist Art — using Masterworks.com’s price-weighted indices, which favor higher-valued works (generally defined as worth $500,000 or more), from 1985 to 2020. It provides estimated annualized returns and correlations with both developed equities.
Annualized Returns and Correlations of Fine Art
The Columbia paper implies a mean correlation of 0.171 with the S&P 500 (calculated as a simple average) across surrealists, impressionists, Picasso, and Renoir. Similarly, Citi’s analysis yields a mean correlation of 0.15 for all fine art, contemporary art, and impressionist art with developed equities. Below is a consolidated table of annualized returns and correlations.
| Asset Class | Annualized Return1 | Correlation2 with S&P 500 or Developed Equities | Source |
| All Art | 8.3% | 0.12 | Citi (2020) |
| Contemporary Art | 11.5% | 0.19 | Citi (2020) |
| Surrealist Artists | 4.61% | 0.144 | Columbia (2015) |
| Impressionist Art3 | 6.8% | 0.13 | Citi (2020) |
| Impressionist Artists | 13.05% | 0.184 | Columbia (2015) |
| Pablo Picasso | 21.28% | 0.149 | Columbia (2015) |
| Pierre-Auguste Renoir | 8.10% | 0.208 | Columbia (2015) |
Life Settlements
Life settlements offer another unique example of a directional non-correlated asset. One where, unfortunately, returns hinge on the starkly idiosyncratic event of mortality. A life settlement involves purchasing an existing life insurance policy from a policyholder. The buyer assumes the premium payments and collects the payout when the policy matures. In the short term, expected returns pivot on shifts in actuarial forecasts; over the long haul, they reflect the gap between expected and actual maturity. There is virtually no economic or fundamental reason why life settlement returns would be correlated with those of the capital markets(a), promising alpha when predictions align with reality.
Yet, pinning down the performance of life settlements as an asset class is not straightforward. These products lack a universal, industry-wide benchmark. The absence of a standardized system muddies the waters — origination risks, legal uncertainties, secondary versus tertiary or distressed versus arm’s-length transactions all confound comparisons. Valuation approaches vary wildly, from fund to fund or deal to deal, making it tough to distill a cohesive picture of returns. Still, the directional nature of the underlying and their insulation from market swings compel me to explore what returns might look like. To that end, I’ve turned to a selected pair of approaches.
Overview of the Studies
One examination comes from a 2013 study published by the London Business School (LBS). Covering January 2001 to December 2011, it analyzes 9,002 life settlement policies insuring 7,164 individuals, with a total net maturity benefit of $24.14 billion. Sourced from Coventry First, a major industry player, the dataset includes detailed transaction records — policy types, costs, maturity expectancy estimates from up to four medical underwriters, and projected cash flows for 7,890 policies.
Another examination comes from a 2015 paper published in The Journal of Risk and Insurance (JRI). Spanning 1993 to 2009, the study constructs an index of 1,724 viatical and life settlement4 policies (VSLI). The portfolio has a face value of roughly $300 million, built using data from New York State Insurance Department filings.
Valuation Framework
VLSI tracks growth over time from acquiring a policy and holding it until maturity, adjusted for all costs, including purchase price, fees and premiums. The fees and premiums are discounted at the 3-month Treasury bill rate (averaging 4.73% annually) to reflect time value. VSLI employs the repeat sales method, a technique borrowed from real estate (Bailey, M.J., Muth, R.F. and Nourse, H.O. 1963), to build a quarterly index of returns5. This approach assumes policy values grow via a log-linear model, where portfolio performance is inferred through a weighted least squares (WLS) regression index. The WLS regression weights policies with shorter holding periods more heavily, reducing the volatility of returns as holding times lengthen6. With only 1,724 policies across 64 quarters (~27 transactions per quarter) this weighting stabilizes the index, much like a weighted stock benchmark. Since consistent actuarial updates aren’t available7, VLSI models these unobservable health changes as a random walk with a zero mean, focusing returns on timing rather than policy specific health shifts.
The LBS study estimates expected returns by calculating the IRR of the portfolio of life settlement policies. Outflows begin with the total cost of purchase, recorded as the cash paid to the policyowner at funding plus any premiums paid to the carrier at or just before funding. The inflow is the net maturity benefit, the face value paid upon maturity, adjusted for any retained benefit8 kept by the seller. Since many policies hadn’t matured by 2011, cash flows are probabilistic, with assumed maturity date probabilities anchoring the IRR calculation. These probabilities are derived from actuarial estimates provided by up to four third-party medical underwriters per insured. The framework assumes no mid-term health updates, fixing estimated maturity at purchase, and dismisses adverse selection, this way returns don’t primarily stem from random actuarial misestimates but on economic factors like cost-benefit tradeoffs, premium convexity, and policy size. Returns are reported in two forms: cost of purchase-weighted (CP-weighted), scaling returns by investment size for a portfolio view, and equal-weighted (EQ-weighted), offering a per-policy average.
Annualized Returns and Correlations of Life Settlements
The LBS study does not directly report S&P 500 or equities correlations but emphasizes life settlements’ exposure to longevity risk, suggesting low correlation with financial markets.
| Asset Class | Annualized Return | Correlation with S&P 500(a) | Source |
| VLSI (1993–2009) | 8.0% | 0.067 | JRI (2015) |
| VLSI (1993–2007) | 7.3% | -0.038 | JRI (2015) |
| CP-Weighted (2001–2011) | 12.5%(b) | N/A | LBS (2013) |
| EQ-Weighted (2001–2011) | 12.9%(b) | N/A | LBS (2013) |
Notes:
1: The Annualized returns found in the Columbia Business School paper are derived from 1985–2014 auction data and corrected for log-transformation bias to capture the uncertainty and variability in the historical data. This correction helps account for the influence of “trophy hunters” who make extremely high-priced purchases, perhaps for prestige rather than intrinsic value, ensuring that the data reflects a more balanced view of the overall market rather than being skewed by a few big sales.
2: Correlations are measured on a scale of 1 to -1, where 1 = two asset classes move in the same direction all of the time; -1 = two asset classes move in the opposite direction to each other all the time.
3: It’s reasonable to approach Masterworks’ data with some skepticism, given their role as a platform for buying and selling fractional artwork ownership, which could incentivize them to present inflated returns. Notably their return figure is lower than Columbia’s. This discrepancy might stem from methodological differences or data gaps (e.g., differing time periods), but perhaps it tempers concerns about potential bias in Masterworks’ reporting.
4: Viatical settlements (“VS”) involve policies sold by individuals with life expectancies of less than 2 years, while the broader life settlements (“LS”) usually involve healthier policyholders with longer expectancies, typically over 2 years. The study includes both, with VS dominant early (1990s) and LS growing later, reflecting market evolution. Life settlements is often a blanket term encompassing both VS and LS. However, VS specifically denote these short-term cases and are not used broadly.
5: The repeat sales method, traditionally used for real estate where price changes reflect two negotiated sales, adapts to life settlements by addressing their fixed payoff structure. Unlike real estate, a VLSI policy’s payoff is fixed upfront at purchase, not determined by economic changes or market woes as in real estate. This fixes the endpoint, shifting all return variability to timing uncertainty. The index anchors to this known payoff, starting from the purchase price (inclusive of discounted premiums) and uses regression to infer quarterly growth based on maturity dates.
6: Policies with shorter holding periods—often VS—are weighted more heavily in WLS to address heteroskedasticity, where longer periods exhibit greater return variance due to unpredictable events (e.g., health shifts, medical breakthroughs). With sparse data, weighting shorter periods curbs skew from a large spread of time horizon outliers (e.g., 90 vs. 10 months). VS, with their higher probability of return certainty and quicker resolution, also offer statistical stability, unlike longer-term life settlements prone to variance.
7: If actuarial updates were available or used, estimated maturity dates would be periodically adjusted based on evolving health data. In practice, many life settlement funds reevaluate policies quarterly or annually, using fresh medical records to refine projections and shift book value.
8: The retained benefit represents a fractional ownership of the total maturity benefit, reducing the investor’s inflow and subsequently outflow. This split effectively divides the benefit’s ownership between the seller and buyer at maturity.
(a): The slight correlation with the S&P 500 arises from supply and demand dynamics. Institutional buying in 2003–2004, led by firms like AIG and Berkshire Hathaway, increased demand, driving prices up and returns down. Conversely, distressed sales in 2008–2009, as cash-strapped policyholders flooded the market during the recession, coupled with reduced investor demand, lowered prices and lifted yields. In Q1 2009, fears of insurer defaults briefly aligned VLSI dips with equity plunges. These episodic shifts, alongside broader supply/demand volatility, create fleeting ties to market conditions.
(b): The framework tests LE sensitivity by extending estimates by 12, 24, and 36 months, reflecting potential underestimation of maturity. The CP-weighted IRR drops to 9.0%, 6.1%, and 3.2%, respectively, while EQ-weighted IRR falls to 9.2%, 5.9%, and 2.6%. Thus, even with a 36-month extension, returns remain positive.
Sources
Braun, A., Cohen, L. H., Malloy, C. J., & Xu, J. (2018, June). Introduction to life settlements. Harvard Business School (link)
Citi GPS: Global Perspectives & Solutions. (2020, December). The global art market. Citi (link)
Giaccotto, C., Golec, J., & Schmutz, B. P. (2015). Measuring the performance of the secondary market for life insurance policies. The Journal of Risk and Insurance (link)
Januário, A. V., & Naik, N. Y. (2013). Empirical investigation of life settlements: The secondary market for life insurance policies. London Business School (link)
(2015). The art market: What do we know about returns? Columbia Business School (link)