Correlation in an asset is a condition. In a strategy, it is behaviour.

The asset layer

Bitcoin against the S&P 500, 2019 to 2026.

Pearson correlation of daily log returns, bitcoin against the S&P 500, by calendar year:

In the two stress windows of the sample, when a diversifier earns its allocation, the correlation held at the top of its range: 0.52 through the COVID crash (19 February to 30 April 2020) and 0.56 through the January to June 2022 drawdown.

Two more numbers frame that table. Bitcoin against ether: between 0.78 and 0.91 every single year since 2019, the same trade under two tickers. Bitcoin against gold: near zero throughout, ranging -0.05 to 0.30 and mostly around 0.1. Diversification is not dead. Gold still does it. Bitcoin stopped.

Our own number, published against interest

The record's least flattering number belongs in public.

The OSIQ 30's correlation to bitcoin, 1 July to 16 August 2026, interval-matched, 45 observations: 0.48. About the same as bitcoin's own correlation to the S&P 500 this year.

We publish it unprompted; publishing the record's least flattering number is the benchmark's job. And the number is unstable in a way worth showing: July alone was 0.64, August alone 0.23. That swing is what short windows do: forty-five observations prove nothing, in either direction. Beta to bitcoin over the same matched window: 0.37.

Same market, different path

Correlation compresses two return paths into one number.

Correlation compresses two return paths into one number. The paths themselves say more. In July, bitcoin's deepest drawdown was 5.52%. Over the same days, the index drew 3.26%, and it closed the month at +11.09%. Same market, different path.

Why the layers differ

Assets inherit markets. Strategies express decisions.

An asset's correlation is a condition of its market: bitcoin cannot decide to stop following risk appetite, any more than a stock can decide to leave its index. A strategy's returns are made of decisions instead: long, short, flat, funding, timing, rather than holding. Decisions are behaviour, and behaviour can be measured, scored and selected on. An asset has no choice but to follow its market. A behaviour has nothing it must follow.

That is the layer this benchmark exists to measure. The 0.48 says the cohort's opening weeks were not especially uncorrelated. The point is that at the strategy layer, correlation is observable per strategy, daily, in public, and can be selected on, rather than inherited by holding the asset.

Method, so you can check it

Match each return to the window that produced it.

Bitcoin against the S&P 500, ether and gold: Pearson correlation on daily log returns over calendar-year windows, aligned on common trading days, computed from Yahoo Finance daily closes and cross-checked against Coin Metrics, CME and NYDIG published figures. OSIQ 30 against bitcoin: the published index ledger matched to Hyperliquid 15-minute candles at each snapshot's actual capture time. The matching matters: a naive date join returns 0.16, a more flattering and wrong number, because the ledger's snapshot times shifted in mid-July. We publish the matched 0.48. The full index series is at openstrategy.trade/api/index/record.json; the candles are on Hyperliquid's public info API.

Limitations

A short window moves quickly.

Forty-five matched observations is a short sample, and the July/August split (0.64 against 0.23) is the proof of how much such windows swing. All figures are frozen to the windows stated above and will not be silently refreshed; later windows get their own note. Correlation describes the past; it is not a forecast. OSIQ 30 is a public research benchmark and is not currently investable.