Alt Breadth Thrust on OKX Spot: Buy an Equal-Weight 12-Alt Spot Basket When Market Breadth Swings From Washed-Out to Broad Strength Within 10 Days, Hold Up to 30 Days, Exit Early on Breadth Collapse (OKX_SPOT 1-DAY, long-only, unlevered 60% gross)

Derived by the Research Lead from Martin Zweig's breadth-thrust indicator. Zweig's idea: when market breadth moves from oversold to broadly advancing within about 10 sessions, a lasting advance is starting. Here it is applied to a basket of crypto alts. The knowledge base has no test of a breadth-THRUST event. The only related test used continuous breadth-scaled allocation, which is a different rule. Universe: 12 alt spot markets on OKX_SPOT with daily bars from 2020: SOL, XRP, ADA, DOGE, LINK, LTC, DOT, AVAX, BCH, ATOM, ETC, FIL. Selection rule: the largest non-BTC, non-ETH, non-stablecoin alts with an OKX USDT spot market listed by 2020-12. The universe was fixed before any test. Point-in-time handling: on each daily close a name is ELIGIBLE only if it is listed (is_listed / universe_at) and its SMA has a full window. Breadth and entries use eligible names only. Breadth is computed only when at least min_eligible=8 names are eligible. Remaining survivorship bias: OKX keeps no delisted spot pairs, so a name that later died cannot be included. The report must say so. Signal: breadth_t = fraction of eligible names whose daily close is above their own sma_len-day SMA. A THRUST fires on daily close t when all three hold: breadth_t >= thrust_high; breadth_{t-1} < thrust_high; and min(breadth over the prior thrust_window closes, excluding t) <= thrust_low. On a thrust, while flat, buy every eligible name in equal notional. Execution: decisions are made on the 1-DAY close. Orders are market orders at the signal bar's close. On a 24/7 venue that price is the next bar's open (registry next_bar_open_fill: use_equivalent). All legs exit together after hold_days daily closes, or earlier on the first daily close where breadth < exit_breadth. No new thrust is taken while the basket is held. Risk control route (b): signal/time exit with capped sizing and no stop-loss. Each leg = gross_exposure / N_eligible of equity (60% / 12 = 5% per leg). Exposure never exceeds 100% of equity; unlevered cash account. Pre-study on local OKX_SPOT daily bars 2020-2026. It is gross, close-to-close and before costs. It was NOT a decision input and was used only to set predictions. - With sma 50, low 0.25, high 0.75, window 10, the thrust fires about 29 times in 6.5 years (about 4-5 per year), in every calendar year including 2026. - Mean 20-day equal-weight basket return after a thrust is +9.6% (median +5.2%). Unconditional: +4.4% (median +0.0%). - With the breadth<0.35 exit, the 30-day cap and 60% gross: 354 leg-trades, mean leg return about +12% but median about -3.5%. The edge is right-tailed: 2021 and 2024 dominate. Max drawdown about 25%; 2022, 2025 and 2026 were negative. - The SMA-20 variant was much worse (drawdown 54%), so sma_len is the key sensitivity. Fee arithmetic: an OKX spot round trip costs about 0.20% of notional. The mean leg captures several percent, carried by trend legs, while the median leg loses. Honest predicted average is about 3% per leg, more than 10x the cost.