XRP–XLM Relative-Jump Exhaustion Across Hyperliquid and OKX

Hypothesis

Test whether a large relative repricing between payment-network tokens partially unwinds after its first observable reversal. Trade synchronized 4-HOUR XRPUSD.HYPERLIQUID and XLMUSDT.OKX bars, shorting the relative winner and buying the laggard. This is cross-asset relative value, not guaranteed arbitrage. Market orders execute at the signal bar close, the supported equivalent to next-bar-open execution on continuous venues. Individual protective stops trigger intrabar through platform matching. Include actual funding and execution costs on both legs. All capture and firing-rate estimates are unmeasured research predictions.

Long-only exposure is the largest portfolio imbalance at 76.9% versus a 55% ceiling. This paired book also adds underrepresented Hyperliquid, OKX and cross-venue exposure. corpus_concentration is empty, so recent symbol and timeframe shares cannot be verified. Prior XRP–XLM experiments traded relative-momentum persistence and lost; this proposal tests the opposite direction after a large recent jump begins reversing, with bounded exposure and a fixed holding limit. Their failures do not establish that reversal works. The supplied options commentary overlaps existing skew-spread hypotheses, so this proposal selects a simpler paired contract.

Negative observed expectancy across 73 paired episodes, including losses before modeled execution costs. Funding is credited and QA’s concern is borne out on full available paired history. This rejects advancing the measured XRP–XLM relative-jump exhaustion baseline, not reversal universally or every untested threshold. No accounting or data defect is established.

Implementation

On exactly synchronized 4-hour XRP Hyperliquid and XLM OKX bars, arm after a large XRP/XLM ratio change over event_bars, then short the relative winner and buy the laggard after the first opposite ratio move of at least 0.1%. Each leg targets 12.5% of combined venue equity with at most 1% initial notional mismatch. Exit on a 3% favorable relative-ratio move, a 3% adverse move, 18 observed bars or 72 hours, or either leg's intrabar 8% protective stop. Wait six synchronized bars after closing. Market entries execute at the signal close, the supported continuous-market equivalent of next-bar-open execution.

Implements relative-jump exhaustion rather than persistence, retaining both declared instruments and all fixed research parameters. The continuous signal is the unbounded XRP/XLM ratio return over event_bars; entry confirmation uses the one-bar ratio return. All position, event and cooldown state is scalar, while price history comes from bounded framework buffers and exact timestamp matches. Equal-dollar sizing uses the combined balances of two separately funded venue accounts, with no leverage. At the baseline, an 8% stop on a 12.5%-of-total-equity leg risks 1% of total equity. The second leg is submitted only after the first entry fills; either leg's stop closes the surviving exposure. Actual venue fees, impact and per-instrument funding accrual are supplied and audited by the backtest runner; no funding proxy or synthetic cash credit is added. Only entry_gap and event_bars are tunable under the attached research plan. Capture and loss thresholds measure signed fractional XRP/XLM ratio change from filled entry prices.

Verification Results

Before trusting full-backtest economics, verify funding_rates_by_instrument contains XRPUSD.HYPERLIQUID and XLMUSDT.OKX settlement histories over the held windows, and confirm their payments are booked. The full-backtest accounting loader already requests these series independently of strategy signal reads.

Sandbox accounting has funding_events_available=0, funding_folded=false and zero posted funding despite multi-day perp positions. It therefore does not validate the hypothesis's requirement to include actual funding on both legs. This is a sandbox input limitation, not demonstrated absence of historical funding or a strategy coding defect.

The smoke test reports -11.324% total return, PF 0.736 and -0.6523% average net leg return before funding. Assess expectancy on full history with actual funding on both legs; these losses are not evidence of a code defect.

Backtest Review

Sharpe

-1.39

Total return

-11.28%

Max drawdown

12.06%

Trades

146

Win rate

46.6%

Profit factor

0.74

All 73 episodes contain synchronized opposing legs with entry-notional mismatch below 0.003%. Code implements winner-fading after reversal confirmation.

Accounting reconciles, funding is credited, and no positions remain open. Buy-and-hold is not a meaningful benchmark for this paired book.

QA’s concern is confirmed: 146 closed legs lose $22,550.69, with return -11.275%, PF 0.737 and Sharpe -1.389.

Average net leg return -0.6493% falls below the +0.15% Hyperliquid review reference. Approximately $16,023 of losses remain before commissions, modeled impact and funding.

Every reported calendar year loses. No specific implementation defect or evidence-backed improvement region warrants optimization.

Outcome Summary

This measured baseline needs evidence of positive expectancy before further tuning, since losses persisted before execution costs.

Backtest review voted to abandon because expectancy was negative even before modeled execution costs, and every reported calendar year lost money. No accounting or data defect or evidence-backed improvement region was established; optimization and later reviews were not reached.

Fade large XRP–XLM relative jumps after the first observed reversal by shorting the winner and buying the laggard on synchronized four-hour Hyperliquid and OKX bars.

Across 989 days, 73 paired episodes produced 146 closed leg trades, a net return of -11.275%, Sharpe of -1.389, and maximum drawdown of 12.065%. Average net leg return was -0.6493%, with a profit factor of 0.737.