Skip to content
All hypotheses

Hypotheses

OKX Alt-Swap Basket 4H Aggressor-Burst Continuation, Long-Short (12 pre-2021 OKX USDT swaps: follow a >=3.5-sigma, >=4x-volume, extreme-close-location 4H bar for a fixed 24h, flat ~95% of the time)

Cross-sectional transplant of the factory's live FIL aggressor-burst continuation idea (FILUSDT.BINANCE 1H, live_stage) onto a different venue, different bar size and a different universe: 12 OKX USDT perpetual swaps…

Hypothesis

Cross-sectional transplant of the factory's live FIL aggressor-burst continuation idea (FILUSDT.BINANCE 1H, live_stage) onto a different venue, different bar size and a different universe: 12 OKX USDT perpetual swaps with continuous 4-HOUR history since 2020 (ETC, BCH, DOT, LINK, ATOM, AAVE, UNI, XLM, ALGO, SUSHI, CRV, COMP). These are almost unused in the corpus (OKX is 2.5% of experiments), and the strategy is long-short, which is also under-represented (18.8%). Mechanism: a 4H bar that is simultaneously (a) an……Show moreShow less

Cross-sectional transplant of the factory's live FIL aggressor-burst continuation idea (FILUSDT.BINANCE 1H, live_stage) onto a different venue, different bar size and a different universe: 12 OKX USDT perpetual swaps with continuous 4-HOUR history since 2020 (ETC, BCH, DOT, LINK, ATOM, AAVE, UNI, XLM, ALGO, SUSHI, CRV, COMP). These are almost unused in the corpus (OKX is 2.5% of experiments), and the strategy is long-short, which is also under-represented (18.8%). Mechanism: a 4H bar that is simultaneously (a) an extreme return versus that coin's own trailing volatility, (b) a volume burst versus its trailing median volume, and (c) closes near its extreme marks one-sided aggressive flow: forced liquidations, stop runs, or information-driven takers. These moves are not absorbed within the bar. On mid-cap alts, where liquidity providers withdraw after such bars, the move tends to keep drifting for hours as late followers, liquidation chains and cross-venue arbitrage on Binance and Bybit catch up. The strategy follows the burst direction for a fixed 24h (6 bars), with an intrabar protective stop. It uses OHLCV only, needs no supplementary feed (L189), and is event-driven and flat most of the time. Each trigger is evaluated independently per coin. Execution: decisions happen on the 4H bar close, with a market order at that close. On a 24/7 venue that is the faithful equivalent of the next-bar-open fill, and this substitution is stated here. The stop is an intrabar protective stop: a reduce-only stop-market order matched inside the bar by the platform's simulated venue, which live trading mirrors. No exchange-resting order is required. Sizing: 10% of equity notional per position at leverage 1, at most 4 concurrent positions (max gross 40%), with one netted position per instrument. Fee arithmetic: OKX taker round trip is about 0.10%, plus about 0.05% modeled impact on mid-caps, so about 0.15% all-in. Extreme 4H bars of this type on these alts typically move 5-9%. If 15-20% of that continues over the next 24h, the gross capture is about 0.8-1.2% mean and the net capture is about 0.5% per trade. That is more than 3x the cost. Firing rate: about 6-10 qualifying bars per coin per year across all regimes (both up and down bursts qualify, so it is not bear-only), or about 80-120 trades per year across the basket and about 450-650 over the 2021-2026 history.

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