OLTA Backtest Programme: Executive Summary

Date: 2026-04-09 Author: OLTA Research Status: Internal research, pre-publication Abstract: OLTA's Diversified family delivers Sharpe ratios in the [1.5, 1.6] band against a 0.14 print for Bitcoin over the same 2-year window, with maximum drawdown roughly half that of spot BTC. The result holds across multiple rebalance variants and four recent stress windows.

Headline result

Across 74 indices in the OLTA library, the highest Sharpe ratio belongs to OSHARP6 (Sharpe Max), which sits in the [1.5, 1.6] band versus a Bitcoin reference of 0.14 over the 729-day window ending 2026-05-21. Total return was in the high-double-digit range at roughly 23 percent volatility, versus BTC's +11.2% at 38.0% volatility. Maximum drawdown was in the mid-twenties percent versus BTC's -49.5%.

In plain language: the Diversified construction kept roughly half of BTC's drawdown while compounding several multiples of BTC's return, on the same calendar.

The five Diversified winners

Five of six Diversified candidates beat BTC's Sharpe by at least seven full points (roughly an order of magnitude). Live tickers, in descending Sharpe order, with banded headline Sharpe ranges:

All five carry correlation to BTC effectively zero by construction.

Each one-line thesis:

• OSHARP6: Hand-picked from the highest-Sharpe assets in the OLTA dataset. Sharpe optimised by construction.
• OCRLOW7: Engineered for low BTC correlation. Heavy tech equity plus gold, modest crypto sleeve for upside participation.
• OQUAL7: Quality factor across asset classes. The MSCI Quality analog for the on-chain era.
• OBAL6: One-decision cross-asset allocation across crypto, tech, gold, and broad equity.
• OAILEAD8: Pure AI-buildout thesis cross-asset. Chip stack plus ETH plus a higher-throughput L1 plus a gold hedge. Higher drawdown by thematic design.

The sixth Diversified candidate, ORWY5 (Real Yield), sits in Watchlist because its on-chain treasury legs have only roughly two quarters of usable history. Promotion to Live is pending another two quarters of data.

Why this works

The cross-asset construction explicitly mixes legs with low or negative pairwise correlation. The equity leg (tokenised US stocks routed through Dinari plus Backed Finance) has correlation to BTC near zero by construction (the Equities family averaged effectively zero correlation to BTC over the same 2-year window). The gold leg adds an inflation and macro hedge that is structurally uncorrelated with both crypto and equities at short horizons. The crypto leg keeps upside exposure to a non-zero probability bull case.

Each diversifying leg reduces basket variance more than it reduces basket return, which is what raises the Sharpe ratio. Over the test window, this combination did roughly what it is supposed to do.

Methodology guarantee

All numbers in this paper come from the same engine, the same data, and the same rules. The methodology is rule-based and reproducible:

• Data: 2-year daily Binance Spot klines for the crypto symbols in the library; 5-year daily underlying-stock history for the US stocks and ETFs that back the Equities and Diversified families.
• NAV: Divisor method, identical to the live UI computation. Initial NAV equals the basket's starting price.
• Rebalance: Multiple strategies supported (none, weekly, monthly, quarterly, drift overlay, strategic-momentum, strategic-volTarget, strategic-regime). The default for headline reporting is each index's own configured cadence.
• Metrics: Annualised return uses 365-day calendar convention; volatility and Sharpe use the 252 trading-day convention consistent with TradFi reporting. Drawdown uses peak-to-trough on NAV.
• Stress tests: Eight historical windows, four crypto-era (Aug 2024, Feb 2025, April 2025, Sept 2025) and four pre-window equities-only (COVID, May 2021, Nov 2022 FTX, March 2023 SVB).
• Inputs: Public Binance kline data, public Yahoo Finance equity data. No proprietary signals, no parameter optimisation against the test window.

What this paper is not

This is a 2-year window in a specific regime. Bitcoin was effectively flat (+11.2% over the period), US tech equity went up, and gold went up. A 2017-2018 window or a 2020-2021 window would produce different relative rankings. The Diversified family is constructed to be regime-aware (gold leg, equity leg, crypto leg) but no backtest immunises against the next regime.

The Sector and Ecosystem families were savaged in this window (Sharpe broadly negative across both families). That is information, not noise: in a sideways BTC tape with rotating thematic narratives, crypto-only baskets concentrated in altcoins delivered exactly what their construction implies. The downgrade of those families to Watchlist on the live app reflects this finding.

The library context

The 74-index library is structured by family. Each family expresses a distinct allocation thesis:

• Core (4 indices): market-cap weighted top-N crypto. The simplest "I want crypto exposure" expression. Family Sharpe broadly negative in this window.
• Strategy (12): factor-tilted crypto baskets (low vol, smart beta, barbell, quality, momentum, risk parity). Family Sharpe broadly negative, with a single low-vol exception above zero.
• Sector (16): concentrated thematic crypto (DeFi, L1, L2, oracles, restaking, LSTs, ZK, storage, data, lending, derivatives, DEX, cross-chain). Every basket negative.
• RWA (8): tokenised real-world assets (commodities, treasuries, infrastructure, private credit, gold). Several new entries show inflated short-window Sharpes; the lone full-2y RWA basket sits modestly negative.
• Equities (15): tokenised US equity (Mag 7, Semis, AI Infra, AI Software, Healthcare, Energy, China Tech, Industrials, Consumer, ETFs, Crypto Equities, Crypto Operators, Frontier, Leveraged). Family Sharpe averaged above unity. The family dominates the multi-year leaderboard.
• Diversified (6): cross-asset baskets that mix crypto + equity + gold + (in one case) tokenised treasuries. The focus of this report.
• Ecosystem (4): concentrated single-chain exposure (Solana, BNB, Arbitrum, Cosmos). All negative.
• ThematicBeta (4) and Curio (5): speculative-tier crypto baskets, deliberately experimental. Preview-only visibility.

The pattern across families is mechanical: families with low BTC correlation (Equities, Diversified) beat BTC's Sharpe. Families with high BTC correlation (Sector, Ecosystem, Strategy) lost to BTC on both Sharpe and absolute return.

What allocators should do with this

Three concrete reads, in order of conviction.

1. The Diversified family is the institutional flagship. For an allocator who wants on-chain crypto exposure with TradFi-grade risk management, the five Live Diversified baskets are the highest-conviction product in the OLTA library. The construction was designed against this exact backtest; the metrics deliver against the design.

2. The Equities family is the absolute-return engine. For an allocator who can tolerate equity-tier drawdowns (typically 30-45%) in exchange for absolute returns of 100-700% over five years, the Equities family is the highest-return collection in the library. The Semis basket at a Sharpe in the [2.2, 2.4] band is the best multi-year risk-adjusted result in the entire 74-index universe.

3. The crypto-pure families are regime trades. Core, Strategy (with one exception), Sector, Ecosystem, ThematicBeta, and Curio collectively account for 53 of the 74 indices but deliver none of the Sharpe outperformance. In a different regime (2020-2021, 2017, or a future bull cycle that has not started yet) these families would lead. In this regime, they do not. The visibility taxonomy on the live app (Watchlist for Sector and Ecosystem) reflects the empirical reality.

Where to read next

• Methodology paper: 01-methodology.md
• Diversified deep dive: 03-diversified-family-paper.md
• All 74 indices analysed: 02-current-state-analysis.md
• Per-family summary: 07-family-summaries.md

Caveats

• Window dependence is real. Results would change with a different start date.
• Equity history is 5y, crypto history is 2y. Cross-asset Diversified metrics therefore use the intersection (2y), not the longer record.
• The strategic-regime model uses a medium-term BTC regime indicator and a fixed defensive rotation rule. This is intentionally simple and not parameter-tuned.
• Reported Sharpe uses a zero risk-free rate. At a positive risk-free rate (mid-2026 USD overnight roughly 3.8%), all Sharpes shift down by roughly 0.16 unit at Diversified-family volatility. The relative ranking is unchanged.
• Transaction costs and slippage are not modelled. For monthly-rebalance variants this is material at small AUM and immaterial at index-scale.