Strategies

Derived by the Research Lead from Zhang & Urquhart, 'Pairs trading across Mainland China and Hong Kong stockmarkets' (International Journal of Finance and Economics, 2018), as summarised on BigQuant (https://bigquant.com/wiki/doc/zNkp9XcBmh). The full paper PDF could not be read (HTTP 404), so only the abstract's findings were available. The page's only image is the BigQuant site favicon (a blue logo icon), not a chart, so it adds no evidence. Findings from the source: pairs trading inside either market alone earns no significant abnormal return; pairs trading ACROSS the mainland and Hong Kong markets earned about 9% a year in abnormal returns after risk and costs from 1996 to 2017; profitability varies over time and fell after 2012; and the strategy does best in long range-bound markets. The source attributes the edge to irrational demand combined with limits to arbitrage between the segmented markets. The rules below are the Research Lead's operationalization, not the authors' strategy. I use the cleanest cross-market pair available: the A-share and H-share of the SAME company, which are claims on identical cash flows kept apart by capital controls and Stock Connect quotas. Constraint: CNEQ and HKEQ are both CASH accounts, so the rich twin cannot be shorted and a true two-leg spread cannot be expressed. Prior art: run 17503f75 held only the cheap twin with no hedge. Its analyst ruled that this unhedged form cannot measure A/H convergence, because the long leg carries full China beta, and asked for a hedged version. This hypothesis is that hedged version. A short in FXI.USEQ (the iShares China Large-Cap ETF, whose top holdings are H-shares of the same banks, insurers and oil majors) is sized against the long book, so the remaining P&L is the cheap twin's catch-up plus single-name noise, not China market direction. Universe: 11 verified A/H twins with 1-DAY bars on both venues from 2018-01-02: ICBC 1398.HKEQ/601398.CNEQ, CCB 0939/601939, Bank of China 3988/601988, Agricultural Bank 1288/601288, Bank of Communications 3328/601328, China Merchants Bank 3968/600036, Ping An 2318/601318, China Life 2628/601628, PetroChina 0857/601857, Sinopec 0386/600028, BYD 1211/002594. The universe is fixed. These are long-standing large-cap dual listings, all listed on both exchanges before 2018, so there is no survivorship selection within the test window. Execution: decisions are made at the close of the primary 2318.HKEQ daily bar (HK close 08:00 UTC); the A-share close (07:00 UTC) of the same date is already known by then. Every order fills at its own leg's next session open (next_open_fill, supported): HK legs at the next HK open, A-share legs at the next mainland open, and the FXI hedge at the next US open (13:30 UTC the same calendar day). The hedge is therefore placed about 12 hours before the HK/CN entry fills. That is a known timing mismatch and stays in the test. Each leg is held for many sessions (expected median ~15-25), so CNEQ T+1 is never binding. Costs: HKEQ about 0.22% statutory round trip (stamp duty on both sides) plus about 0.08% spread; CNEQ about 0.10% plus 0.04% spread; FXI about 0.03% spread plus modeled borrow. Precondition: the previous A/H run showed total_commission 0.0 on HKEQ/CNEQ. The backtest must show non-zero commissions on both venues before its economics are read. Risk route (b): signal exits and capped sizing. Each long twin is 8% of reference equity (max 8 concurrent, so 64% gross long), and the FXI short is at most 64% of reference equity. Unlevered per leg; the only margin used is Reg-T on the USEQ hedge.

This hypothesis traded the up-day counterpart of the factory's flush-vs-build classifiers. It shorted Bybit alt rallies that came with falling Binance open interest, a sign that shorts were covering rather than new buyers arriving, and hedged each short with BTC for 3 days. The Research Lead's pre-registered study showed a hedged 3-day return of -1.11% across 134 events on 62 dates, which is a gain for the short. The strategy was coded in 1 iteration, but it never reached a backtest, backtest review, optimization, analyst review or risk review. It sat in data review with an unavailable input for 352 h, past the 144 h limit, and was abandoned without any judgment on its edge. It can be revived once the data issue is resolved.

This hypothesis took a published SPY–Bitcoin LSTM diversification idea and swapped quarterly rebalancing for fixed per-venue capital sleeves, so it was a feasibility test rather than a profitability claim. Across 4 iterations the developer could not implement the required cost-net benchmark and standalone attribution within the runner, so the strategy code raises an unsupported_capability error on purpose. The capability hold went unresolved for 418 hours against a 216-hour limit, and the run was abandoned without a backtest, optimization or analyst verdict. The record says the run can be revived if the registry gains the capability or the contract is rewritten.

This one-iteration run tested whether perp-first tokens sell off after their delayed Binance spot listing, using a BTC-hedged short of the USD-M perp held for a fixed period across 65 events. The strategy was coded, but the run stalled at the data stage: it waited 442 hours in a blocked-data review against a 144-hour limit. It was then abandoned as hold_expired, with no backtest, optimization, analyst or risk verdict. The idea is still untested and can be revived once the data issue is fixed.

Test convergence of BTC-versus-ETH implied-minus-realized volatility premia. Buy the cheaper-premium asset's ATM straddle and sell the richer-premium asset's same-expiry straddle, matching dollar vega. Hedge each asset's delta with Bybit linear perpetuals. ETH-PERPETUAL.DERIBIT-4-HOUR-LAST-EXTERNAL is the untraded primary clock; synchronized BTC perpetual bars supply BTC realized volatility. Historical Deribit chain expansion must resolve all four actual option contracts point in time and persist their full IDs in extra_instruments and the run's market context before verification. The listed perpetuals are permanent clock and hedge instruments, not substitutes for options. Continuous-market orders execute at synchronized signal-bar closes, the registry's equivalent of next-bar-open execution. Test only actual overlapping chain, quote and hedge history. Standard-tier option fees are underlying-based: [Deribit fees](https://support.deribit.com/hc/en-us/articles/25944746248989-Fees).

This hypothesis proposed collecting the ETH FOMC event-insurance premium with a protected weekly Deribit iron condor, using 4-hour bars and ETH-PERPETUAL as the clock. After one iteration, the developer delivered code that deliberately refused to start, because the rolling options template could not express per-strike leg directions or FOMC-based entry and exit without prohibited overrides. The capability gap stayed with the research lead for 454 hours, past the 216-hour limit, so the run was abandoned before any backtest, review or optimization took place. The abandon reason recommends a capability key or a contract revision if the idea is revived.

HowtraderEthPcpArbitrage translated an MIT-licensed howtrader put-call-parity strategy into a three-leg ETH trade: a Binance call/put synthetic forward against a Bybit perpetual. After one iteration the run was held in data review because an input was unavailable. The hold lasted 514 h against a 144 h limit, expired unresolved, and the run was abandoned before any backtest metrics, backtest review, optimization, analyst or risk review were produced. The abandonment reflects the data problem, not the strategy's merit, and the run can be revived once the data is fixed.

HlSpotPerpAggressorDisagreement coded a spot-versus-perpetual aggressor-flow disagreement signal on Binance tick data to trade HBAR, XLM and ETC perpetuals on Hyperliquid. After one iteration the run was blocked in data review and never reached a backtest, backtest review, analyst or risk review. The block held for 524 hours against a 144-hour limit, so the run was abandoned on hold expiry rather than on its merits. The recommended next steps are to re-verify after the deployed sufficiency fix or to drop the leg that fails the check.

This EURUSD 1-hour EMA(5) cross strategy was built as a fixed-rule plumbing test to exercise orders, flips and session-boundary fills, and it was registered in advance as expected to lose to costs. The backtest showed the mechanics working: balanced long and short trades, no liquidations, a clean input audit and no order failures. Economically, though, it lost steadily (-9.34% total, Sharpe -30.6, average trade -0.154%) because the $2 minimum commission swamped a gross edge of about zero. The analyst abandoned it at backtest review after one iteration and noted that the operator's plumbing goal is better served outside the promotion pipeline.

This strategy ported the FMZ article's actual RSI2 code rather than its prose: it faded RSI2 extremes against the SMA70 trend on BTC COIN-M 15-minute bars at 1x with no stop. In the single full-history backtest it made 1,606 trades at -0.172% per trade (PF 0.78, Sharpe -1.33) and was liquidated in April 2023. The analyst abandoned it at Backtest Review for negative expectancy. Gross capture was negative before fees, the short arm drove the losses, and every year lost money. The verdict covers only this literal code port; the article's prose version and the held-back period were not judged.

Derived by the Research Lead from Cordoni, Sancetta and Zhao, 'Dynamic Strategy Selection' (arXiv 2307.04754, Sections 2-4 and Supplement S.5): each month, pick ONE of several competing factor strategies with a fitted-Q (iterated ridge regression) policy whose reward penalises switching by a regularising cost larger than the real trading cost. The selection algorithm, state design (12-month factor trends + market-cycle dummies), reward (quadratic utility of net return), gamma=0.9, cap B=0.5, GCV ridge grid, NA=10 simulated action sequences and 120-month expanding initial window are taken from the paper. What I changed and why: the paper's actions are Kenneth French MOM/HML/RMW plus a CRSP low-vol factor; we hold no fundamentals (book value, operating profitability) and no CRSP, so I build the four actions as PRICE-ONLY long-short books from our own point-in-time S&P 500 daily panel: (1) MOM = 12-1 month return winners minus losers; (2) DEF = low 252-day realized volatility minus high volatility (the paper's DEF, Novy-Marx style); (3) LTR = 36-to-12-month long-term reversal (past losers minus winners), the standard price-only proxy for HML value (De Bondt-Thaler / Fama-French LT_Rev); (4) STR = 1-month short-term reversal (last month's losers minus winners), replacing RMW which cannot be built without fundamentals. Universe AS OF EACH month-end rebalance: S&P 500 members at that date via index_member()/universe_at() (former members included where bars exist), at least 400 sessions of bars, then the 150 most liquid by trailing 60-session median dollar volume (close x volume). Each factor book: equal-weight long its top quintile (30 names) and equal-weight short its bottom quintile (30 names) of the 150-name universe. The strategy computes all four factor books' daily returns internally every month from the panel (whether held or not), so the factor-return history used for the policy is point-in-time and available from the first month the 36-month LTR lookback is filled (~2000 with membership from 1996). Only the CHOSEN book is traded. Fills: decision on the last session close of the month, market orders filled at the next session's open (platform next_open_fill, supported). Risk control route (b): signal/rebalance exit, dollar-neutral, gross 100% of equity (50% long / 50% short, 1.67% equity per name), no single position above 2% of equity, so leverage 1x gross on the Reg-T 2x account; a long-short price factor at this gross runs roughly 8-15% annual volatility, inside the conservative 60% band. Dividends are not in PnL (both sides, roughly offsetting for a dollar-neutral book; flagged). Short borrow costs are not modelled; universe is restricted to the 150 most liquid large caps to keep hard-to-borrow names out. Benchmarks the analyst must see beside the policy (pre-registered controls, computed by the strategy as report-only series): the Greedy policy (gamma=0, one iteration) and each fixed factor held throughout the same OOS span, as in the paper's Table 1.

This fourth frozen replication of the absorbed-flush rule went back to leveraged perps on a Bybit panel with no shared names, after the Binance spot test came out underpowered. It cleared backtest review (verdict: optimize) and ran one evaluation of the fixed rule, ending with +46.5% over 228 trades, PF 1.58, Sharpe 0.65 and 11.1% drawdown. These numbers broadly reproduce the two promoted siblings. OOS significance was just short (t = 1.81 vs 2.0), the holdout was negative but inconclusive (Sharpe -0.91), and profit was heavily concentrated in a handful of 2024 flush days. The analyst still recommended promotion, and the strategy is now in paper trading, which the analyst noted adds little diversification next to its two siblings.

This strategy tested whether the promoted Bybit perp down-shock rebound template carried over to 14 post-2022 Bybit spot alts hedged with a BTC perp short. The first backtest showed per-event economics far above the ~0.31% two-venue cost, though the backtest review found 2024 carried about 85% of return. After optimization the final parameters (shock_z 2.15, hold 2 days) gave PF 1.82 and positive but non-significant OOS (t 1.41) and holdout (t 1.14), and the largest year carried 45% of PnL. The analyst recommended promotion on iteration 1 and the strategy is now in paper trading as low-confidence research, with survivorship bias, low capital deployment and signs of 2026 decay noted as open risks.

Absorbed-Flush Rule, Fourth Breadth Replication on a Name-Disjoint Bybit Panel with a Same-Date Concurrency Cap: buy a Bybit USDT perp at the close of a down day whose range blew out (>=2x its 60-day return std) but closed at >=0.55 of its range, fixed 3-day clock, 10% per leg, max 2 new entries per date and max 4 open (28 names sharing none with the Bybit/HL/spot panels, 1-DAY, long-only, unlevered)

EurusdLondonBandFade was a faithful port of a user's EURUSD 1-hour London-session 2-sigma band-fade rule. In its single iteration the full backtest lost 6.78% (Sharpe -0.66, 500 trades, average -0.0088% of notional), and it lost most heavily in stressed volatility regimes. The backtest review found no reversion edge even before costs and voted abandon, so optimization and the later reviews never ran. The pipeline still records the final stage as live_stage, which conflicts with that abandon verdict and should be checked.

The strategy took the volume-up / open-interest-down conditioning from a futures short-term-reversal paper and traded only its long side on one Binance DOGE perp, holding each trade for 3 days. Its supporting evidence was a pre-study cell picked from a small grid (+1.84% mean over 54 trades, t≈2.3), which QA noted makes selection, not fees, the main risk; the hypothesis itself asked for an XRP cross-check before any promotion. The code never passed verification: after 3 iterations, QA still found an off-by-one in the time stop that held trades for 4 bars instead of 3. The run was abandoned as a verification-loop code defect before any backtest, optimization or analyst review.

Original research hypothesis: trade unexpected revisions to already public token release schedules, rather than predictable unlock dates. On synchronized 4-HOUR bars, buy reductions and short increases in the next 90 days of scheduled circulating supply, hedged with ETH. Use only revisions publicly announced while the relevant perp was listed. Market orders execute at the signal-bar close, the registry equivalent of next-bar-open execution on continuous markets; protective stops are platform-matched intrabar. Historical publication vintages are required before implementation can test this contract. Tokenomist documents release schedules and event data, but does not establish that historical publication vintages are available: [Tokenomist API introduction](https://docs.unlocks.app/api-documents/introduction). No event counts or forward-return studies have been measured for this proposal.