The Cycle-Tested Construction Doctrine

Date: 2026-05-25 Author: OLTA Research Abstract: Most crypto baskets cannot be backtested through a full cycle because their youngest constituent dates back only to 2020 or later. The intersection-of-availability rule on which any honest backtest is built forces the basket window to start at the youngest constituent's listing date, and for most baskets that date is inside the most recent cycle. This paper documents the cycle-tested construction principle, the reference basket OLTA built to make the principle operational, the mathematical limits a cycle-tested basket faces against BTC, and the implications for the catalogue's overall composition.

1. The intersection rule and what it means for backtest length

A basket's backtest window is bounded by the intersection of its constituents' available daily histories. The window starts on the latest day on which every constituent has at least one observed close and ends on the earliest day on which every constituent's series ends. If a basket holds ten constituents and nine have ten years of daily history but one listed in 2023, the basket's backtest starts in 2023.

The intersection rule is not a methodological choice. It is the only honest position. Extrapolating a basket's NAV before one of its constituents existed would require synthesising the missing constituent's price history, and any synthesis imports an assumption that the basket's published Sharpe should not contain. Better to publish a shorter window than a longer window with synthetic constituents.

The consequence for the OLTA catalogue is concrete. Most actively constructed crypto baskets in 2026 hold at least one constituent that listed in 2020 or later. For those baskets the maximum honest backtest is six years or less. Several baskets hold constituents that listed in 2022 or 2023, which bounds them to three or four years. A few of the most recent baskets hold constituents that listed in 2024 or 2025, which bounds them to one or two years.

Six years covers one complete cycle: 2021 mania peak, 2022-2023 crash and base, 2024 early bull, 2025 sideways. That is sufficient to characterise behavior across one regime turn but not across multiple. Three years covers half a regime turn. One year covers none.

A desk that wants to make claims about a basket's multi-cycle behavior cannot do so from a single-cycle window. The headline numbers are real but they describe a regime, not a basket. This is the structural limitation the cycle-tested doctrine addresses.

2. The cycle-tested principle

A cycle-tested basket holds only constituents whose daily price history extends back across at least two complete cycles. The intersection rule then produces a backtest that itself extends across at least two complete cycles. The basket's published Sharpe, drawdown, and correlation profile measure behavior across multiple regime turns.

The principle is restrictive by design. A constituent that listed in 2020 or later is excluded regardless of how well it has performed. A constituent that listed before 2017 but is illiquid or carries other operational issues is also excluded. The eligible set is necessarily small. For crypto in 2026 the eligible set is roughly a dozen names: BTC, ETH, LTC, XRP, ADA, BCH, plus a handful depending on the deep-history filter.

The small eligible set defines the cycle-tested basket's role. The basket is not a product. It is a reference point. The Sharpe an actively constructed basket reports on its own short window can be compared against the cycle-tested reference's multi-cycle Sharpe, and the comparison tells the allocator something about how regime-dependent the active basket's number actually is.

3. The OLTA reference basket

OLTA built a six-constituent reference basket from the eligible set. The basket holds BTC, ETH, LTC, XRP, ADA, and BCH at weights following a convention scaled to roughly two-thirds BTC plus ETH and one-third the remaining four majors. The exact weights are documented in the catalogue; the principle is that the basket is heavily weighted to the two majors and lightly weighted to the long tail.

The basket's backtest window extends from 2017 through the present. That is eight and a half years of daily history covering 2017 mania, 2018 crash, 2019 base, 2020-2021 bull, 2022 crash, 2023-2024 recovery, 2025 sideways. The window includes two complete cycles plus the front and back halves of two more.

The basket sits in the Benchmark family alongside the equal-weight benchmark. Both are reference points rather than products. Neither is intended for allocator-level investment. Both serve as control groups against the actively constructed baskets. The cycle-tested basket's role in particular is to anchor the catalogue's multi-cycle claims: a comparison of an active basket's recent-window Sharpe against the cycle-tested reference's multi-cycle Sharpe is one of the few apples-to-apples comparisons the catalogue can support.

4. The mathematical limit against BTC

A cycle-tested basket constructed from BTC, ETH, and four major altcoins is heavily BTC-correlated. BTC is the largest weight, and the remaining altcoins each carry high realised beta to BTC across the multi-cycle window. The basket's net correlation to BTC over its full window is in the 0.85 to 0.95 band.

A basket 0.85 to 0.95 correlated to BTC cannot generate large alpha against BTC. The arithmetic is direct: if variance is dominated by BTC's variance and return tracks BTC's return at high correlation, the basket's risk-adjusted profile is bounded above by BTC's. The basket can outperform BTC by a small margin (better diversification across the altcoin tail, weight discipline through rebalance) or underperform by a small margin (drag from constituents whose realised return underperforms BTC's), but it cannot deliver multi-cycle Sharpe materially above BTC's own.

This is not a failure of the basket. It is a structural property of the eligible set. Any cycle-tested crypto basket from the pre-2017 universe will be BTC-dominated. The doctrine identifies what is achievable from that universe and what is not.

Achievable: a Sharpe ratio in the 0.6 to 0.9 band across the multi-cycle window, depending on weights and rebalance cadence. Not achievable: a Sharpe that beats BTC by a large margin on the same window. The cycle-tested basket can report a slightly higher Sharpe than BTC because altcoin diversification reduces idiosyncratic risk; it cannot report a Sharpe two or three times BTC's.

This is informative. A basket that does report a Sharpe two or three times BTC's on a recent window is necessarily drawing alpha from a source the cycle-tested basket does not have. The candidates for that source are limited in number, and the next section enumerates them.

5. Implications for catalogue strategy

If a cycle-tested crypto basket cannot beat BTC's Sharpe by a large margin, then any actively constructed basket that does must rely on one or both of two sources of alpha.

The first source is low-correlation assets. A basket that mixes crypto with assets near-zero correlated to BTC (tokenised US equity, gold, tokenised treasuries, real-world assets) can structurally reduce volatility for the same return profile. The diversification benefit is bounded only by the correlation structure of the underlying assets, and for cross-asset baskets that includes the very low correlations between crypto majors and TradFi sectors. The Diversified family draws its alpha from this source.

The second source is sophisticated weighting. A basket constructed with risk-parity, hierarchical risk parity (HRP), or another optimisation-based weighting scheme can reduce idiosyncratic risk versus an equal-weight or market-cap-weighted alternative. The benefit is real but bounded; the math recovers some of the Sharpe lift that comes from diversification when constituents have heterogeneous volatility profiles, but it cannot beat the structural limit imposed by the basket's correlation structure.

Active baskets reporting Sharpe materially above the cycle-tested reference therefore draw from one or both of these sources. Cross-asset Diversified baskets draw from the low-correlation source. Risk-tilted Strategy baskets draw from the sophisticated-weighting source. Sector and ecosystem baskets that hold only high-BTC-correlated crypto constituents draw from neither, and their reported Sharpes are bounded by the cycle-tested limit.

The doctrine: a basket's reported Sharpe should be readable against the catalogue's cycle-tested reference. If the active basket sits materially above the reference, the source of the alpha should match one of the two structural sources above. If the active basket sits at or below the reference, the basket is not generating alpha against BTC over the multi-cycle horizon and should be positioned as a beta exposure rather than as an alpha source.

6. Reading the catalogue through this lens

The Diversified, RWA, and Equities families all draw alpha from the low-correlation source. Their reported Sharpes sit materially above the cycle-tested reference because the underlying assets are structurally uncorrelated to BTC over the relevant window.

The Strategy family draws from the sophisticated-weighting source. Its reported Sharpes sit in a moderate band above the cycle-tested reference for baskets where the weighting scheme captures meaningful Sharpe lift, and at or below the reference for baskets where it does not. This is informative about which schemes do structural work.

Sector and Ecosystem draw from neither source. Their constituents are crypto-only and their weighting schemes are conventional. Their reported Sharpes sit in a band consistent with the cycle-tested limit; they cannot beat BTC by a material margin and should be positioned as concentrated beta exposures rather than as alpha sources.

The Core family is similarly bounded. Market-cap-weighted top-N crypto baskets are essentially BTC plus an altcoin tail; their multi-cycle Sharpe sits within a small band of the cycle-tested reference.

The Curio and ThematicBeta families are explicitly speculative; they do not claim to be alpha sources and the doctrine does not apply.

This is the catalogue's structural story. A reader who wants to understand why OLTA holds certain families at Live tier and others at Watchlist can read it directly from the cycle-tested reference. Live tier corresponds to families that draw from one or both structural alpha sources. Watchlist tier corresponds to families bounded by the cycle-tested limit.

7. What is not disclosed

The published framework documents the principle, the eligible set, the reference basket's role, the mathematical limit, and the catalogue implications. The methodology brief documents the exact constituent weights of the cycle-tested reference, the per-basket Sharpe attribution against the cycle-tested benchmark, the precise HRP and risk-parity parameters used in the Strategy family, and the per-basket correlation matrices against the cycle-tested reference. The brief is available to institutional counterparties and grant reviewers under standard research-distribution terms.

The institutional pattern is to publish the framework and reserve the recipe. A desk that wants to construct its own cycle-tested reference can do so from the published material. The eligible set is public; the principle is publishable. A desk that wants to lift the OLTA per-basket attribution verbatim cannot, because the per-basket inputs are not disclosed.

8. Caveats

• The eligible set is small. The basket's role is reference, not product. It is held in the Benchmark family and surfaced on the research surface but not promoted as an investable.
• The cycle-tested basket is heavily BTC-correlated by construction. Its multi-cycle Sharpe is bounded above by BTC's plus a small margin. This is a structural property of the eligible set, not a finding about the catalogue.
• Banded ranges for the cycle-tested basket's Sharpe across the multi-cycle window run in the 0.6 to 0.9 band. The exact value depends on the weighting scheme and the rebalance cadence.
• The doctrine reads forward as well as backward. A basket whose constituents include a token that listed in 2024 will not be cycle-tested for several more years. Its reported Sharpe should be read against the cycle-tested reference with the understanding that the underlying window is shorter and the regime fingerprint more visible.
• The doctrine does not claim that recent-window Sharpe is uninformative. Recent-window Sharpe is the only number available for many baskets and the right number for many allocator questions. The doctrine claims it should be read alongside the cycle-tested reference, not in isolation.

The methodology brief documents the per-basket parameters and is available on request.

OLTA Research Desk · 2026-05-25