Skip to content
All hypotheses

Hypotheses

OKX Cross-Sectional 52-Week-High Anchoring Spread: Dollar-Neutral Long Names Nearest Their Trailing High / Short Names Furthest Below It (15 OKX USDT Swaps, Daily Bars, Weekly Rebalance)

A dollar-neutral, multi-instrument, long-short cross-sectional strategy on 15 OKX USDT perpetual swaps (OkxCollector USDT-swap OHLCV, catalog ids <BASE>USDT.OKX, 1-DAY bars, every leg verified in the catalog with daily…

Hypothesis

A dollar-neutral, multi-instrument, long-short cross-sectional strategy on 15 OKX USDT perpetual swaps (OkxCollector USDT-swap OHLCV, catalog ids <BASE>USDT.OKX, 1-DAY bars, every leg verified in the catalog with daily history starting 2020-01 to 2020-12). Signal: the George & Hwang (2004) 52-week-high ratio PTH_i = close_i / max(high_i over the trailing N daily bars). The mechanism is anchoring rather than return momentum. Traders use the trailing high as a reference point and under-react to news that pushes a……Show moreShow less

A dollar-neutral, multi-instrument, long-short cross-sectional strategy on 15 OKX USDT perpetual swaps (OkxCollector USDT-swap OHLCV, catalog ids <BASE>USDT.OKX, 1-DAY bars, every leg verified in the catalog with daily history starting 2020-01 to 2020-12). Signal: the George & Hwang (2004) 52-week-high ratio PTH_i = close_i / max(high_i over the trailing N daily bars). The mechanism is anchoring rather than return momentum. Traders use the trailing high as a reference point and under-react to news that pushes a coin toward it, because resistance sellers and break-even sellers cluster there. Coins that are deep below their high carry overhang from trapped holders selling into every bounce. This is NOT a return-ranked momentum rotation: two coins with identical 30-day returns can have opposite PTH ranks. Every Monday 00:00 UTC daily close, rank the 15 names by PTH. Go long the top K and short the bottom K at equal dollar weight per leg, so the book is net-zero dollar. Names already in the right bucket are held, not churned. Data check per L186/L181: all 15 1-DAY bar dirs exist, with first bars from BTC/ETH/BCH/ETC 2020-01-02 to AAVE 2020-12-12, giving ~5.8 years of overlapping history. Implementation note to avoid sandbox_timeout: keep a fixed-length deque of highs per instrument, never rescan the full history per bar, and rank only on the rebalance bar.

Backtest and paper results are hypothetical. Trading involves risk of loss.