A Descriptive Cross-Asset Performance Comparison

Research question

What did realized cross-asset performance actually look like over 2000–2026 — and how much of any apparent “ranking” is a real difference versus an artifact of unequal histories, mismatched return bases, the currency lens, or how a crisis window is drawn?

Method

A USD investor’s lens over a configured universe of 29 assets (26 core + 3 extended controls): equity indices, selected US stocks, commodities, currencies, cash, and bonds. Monthly returns are the primary basis; daily NAV drives drawdown and recovery; weekly returns drive rolling correlation. The study is deliberately descriptive and ex-post — no forecast, no portfolio optimization, no advice.

The discipline is in the guardrails:

• Label every return basis. Total return, price return, ETF-proxy return, spot-FX, cash accumulation, and front-futures proxy are not the same economic claim. The S&P 500 is total return; the Dow, Nasdaq 100, and non-US indices are price return — an implicit ~1.5–2%/yr dividend gap that is stated, not hidden.
• Equal windows before ranking. FULL_AVAILABLE rows are never equal length, so headline rankings are recomputed on a common window.
• A risk-free-consistent Sharpe. The risk-free series (FRED DGS3MO) is unified to start in 2000 so risk-adjusted metrics span the same window as raw returns.
• Significance on conditional claims. “Defensive” behavior in down months is reported with n, t, and p — not just a mean.

Findings

Three results survive the discipline and reframe the headline numbers:

• The window trap. Tesla tops full-available CAGR (40.9%) only because its series starts in 2010; on a common window Nvidia leads. A ranking across unequal histories partly ranks start dates.
• Risk-free consistency lowers Sharpe. Unifying the risk-free window moves the common-window Sharpe leader from Apple to Nvidia and lowers headline ratios, because the prior start silently excluded the 2000–2002 drawdown from Sharpe but not CAGR.
• “Defensive” is mostly not significant. Gold has the highest average return in S&P 500 down months (+0.9%) but is not significant (p = 0.09); Treasuries are the only reliably positive hedge in the set (p = 0.04).

Supporting results: US large-cap indices are nearly redundant (S&P 500 / Dow = 0.95); the strongest inverse pair is JPY vs the US dollar (−0.46), which is distinct from the weakest (closest-to-zero) pair; regime leaders rotate with no single all-weather winner; and WTI is excluded from standard NAV rankings because its 2020 negative prices invalidate long-only NAV.

What this is, and is not

It is descriptive evidence about a realized sample dominated by a handful of ex-post successful US firms — sensitive to four choices a reader can verify: the sample window, the return basis, the currency lens, and the crisis/regime definition. It is not a forecast, a strategy, or a recommendation. The value is the labelling discipline that turns an attractive-but-fragile set of rankings into honest descriptive statements.