Asset Class: FUTURES

Run: 2026-05-16T06-52-05Z · Verdict: NO_EDGE
Schema v1 · citations: 8 verified / 12 hallucinated · candidates: 10 · backtested: 10 · independent: 10 · cost: $0.00 · wall: 7s

Citations

URLAccess dateTitleEvidenceStatus
https://www.aqr.com/Insights/Documents/Time-Series-Momentum.pdf… 2026-05-11 Time Series Momentum peer-reviewed ✗ unverified low-trust domain
https://www.aqr.com/Insights/Documents/Value-and-Momentum-Everywhere.pdf… 2026-05-11 Value and Momentum Everywhere peer-reviewed ✗ unverified low-trust domain
https://www.bis.org/publ/work790.pdf… 2026-05-11 Carry Trade and Momentum in Currency and Commodity Markets peer-reviewed ✓ verified low-trust domain
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2356501… 2026-05-11 Volatility-Managed Portfolios peer-reviewed ✗ unverified low-trust domain
https://www.federalreserve.gov/econres/feds/files/2018058pap.pdf… 2026-05-11 Regime-Dependent Commodity Futures Strategies empirical ✓ verified low-trust domain
https://www.sciencedirect.com/science/article/pii/S0304405X17301512… 2026-05-11 Commodity Futures Momentum: A Cross-Sectional Approach peer-reviewed ✗ unverified low-trust domain
https://www.aqr.com/Insights/Documents/Commodity-Value-and-Momentum.pdf… 2026-05-11 Commodity Value and Momentum peer-reviewed ✗ unverified low-trust domain
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2620984… 2026-05-11 Carry Strategies in Commodity Futures Markets peer-reviewed ✗ unverified low-trust domain
https://www.nber.org/papers/w22910… 2026-05-11 Volatility Targeting for Commodity Futures empirical ✓ verified low-trust domain
https://www.sciencedirect.com/science/article/pii/S0304405X19302345… 2026-05-11 Regime-Switching Momentum in Commodity Futures peer-reviewed ✗ unverified low-trust domain
https://www.aqr.com/Insights/Research/Journal-Article/Commodities-for-the-Long-R… 2026-05-11 Commodities for the Long Run peer-reviewed ✓ verified low-trust domain
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2607041… 2026-05-11 Time-Series Momentum in Futures Markets peer-reviewed ✗ unverified low-trust domain
https://www.sciencedirect.com/science/article/abs/pii/S0304405X14002376… 2026-05-11 Carry and Trend in Lots of Places peer-reviewed ✗ unverified low-trust domain
https://arxiv.org/abs/1904.04925… 2026-05-11 Mean Reversion in Futures Markets: Evidence and Implications empirical ✓ verified
https://www.aqr.com/Insights/Research/White-Paper/Value-and-Momentum-Everywhere… 2026-05-11 Value and Momentum Everywhere peer-reviewed ✓ verified low-trust domain
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2741651… 2026-05-11 Volatility Targeting in Futures Markets peer-reviewed ✗ unverified low-trust domain
https://www.sciencedirect.com/science/article/abs/pii/S0304405X17303158… 2026-05-11 Regime-Dependent Commodity Price Dynamics peer-reviewed ✗ unverified low-trust domain
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2879018… 2026-05-11 Commodity Futures Risk Premia peer-reviewed ✗ unverified low-trust domain
https://www.aqr.com/Insights/Research/Journal-Article/Superstar-Investors… 2026-05-11 Superstar Investors and Commodity Strategies peer-reviewed ✓ verified low-trust domain
https://arxiv.org/abs/2003.05906… 2026-05-11 Dynamic Asset Allocation in Commodity Futures empirical ✓ verified

Strategy candidates (P2)

spec_idEntryExitSizingUniverseSourcesSource engine
futures_momentum_v1 At market close, compute 60-day total return for each ETF in ['DBC','GSG','DBA','USO','UNG','PDBC']; rank them descending. Open long positions in the top 2 ETFs and short positions in the bottom 2 ETFs. Positions are entered at the next day open. Close each position at the next day open when either (a) the ETF drops out of its respective top-2/bottom-2 rank, or (b) 30 calendar days have elapsed since entry, whichever comes first. Allocate equal dollar amount to each leg; then scale the entire portfolio to target 8% annualized volatility using the 30-day rolling realized volatility of the portfolio (risk-parity scaling). DBC, GSG, DBA, USO, UNG, PDBC [1], [2] cerebras
futures_carry_v1 For each ETF, compute the 30-day simple moving average (SMA30). Go long an ETF when its price is at least 1% below SMA30 (indicating potential positive roll-yield carry) and go short when its price is at least 1% above SMA30. Enter positions at next day open. Close the position at next day open when the price crosses back within ±0.5% of SMA30 or after 45 calendar days, whichever occurs first. Equal dollar allocation per leg; apply risk-parity scaling to achieve a portfolio-wide target volatility of 10% annualized, using a 30-day rolling realized volatility estimate. DBC, GSG, DBA, USO, UNG, PDBC [1], [2] cerebras
futures_regime_switch_v1 Determine macro regime daily: if NBER recession flag = false (expansion) use the momentum rule from futures_momentum_v1; if recession flag = true use the carry rule from futures_carry_v1. Enter positions at next day open accordingly. Apply the respective exit rule of the active sub-strategy (30-day rank change or 45-day SMA cross). Risk-parity across all active legs targeting 9% annualized volatility, using a 30-day rolling realized volatility estimate. DBC, GSG, DBA, USO, UNG, PDBC [1], [2] cerebras
futures_vol_target_momentum_v1 Compute 60-day total return for each ETF; go long the top 2 and short the bottom 2 ETFs. Positions are entered at next day open. Close each leg at next day open when the ETF exits its top-2/bottom-2 rank or after 20 calendar days, whichever occurs first. Scale each leg so that the portfolio's 30-day rolling realized volatility equals a target of 10% annualized (volatility-targeting as in Harvey et al.). DBC, GSG, DBA, USO, UNG, PDBC [1], [2] cerebras
futures_pair_trend_v1 Form two pairs: (USO vs UNG) and (DBC vs GSG). For each pair, compute 30-day momentum for both ETFs; go long the ETF with higher momentum and short the other. Enter at next day open. Close the pair when the momentum difference reverses sign or after 40 calendar days, whichever occurs first. Allocate equal risk to each pair; within each pair, size long and short legs equally. Apply portfolio-wide risk-parity scaling to target 9% annualized volatility (30-day rolling vol). DBC, GSG, USO, UNG [1], [2] cerebras
futures_carry_vol_dbc_v1 long DBC when 60d carry signal (implied yield from futures curve) > 0 AND 30d realized volatility < 15% exit when carry signal flips negative OR after 60d hold period fixed position size targeting 10% annualized volatility based on 30d historical vol DBC [1] xai
futures_momentum_pdbc_v1 long PDBC when 120d momentum > 0 AND 60d momentum > 0 exit when 60d momentum < 0 OR after 90d hold period fixed position size targeting 10% annualized volatility based on 60d historical vol PDBC [1] xai
futures_regime_dba_v1 long DBA when 60d momentum > 0 during economic expansion (based on 6m moving average of ISM PMI > 50) exit when momentum < 0 OR economic regime shifts to recession (ISM PMI < 50) risk-parity allocation targeting 8% annualized volatility based on 60d historical vol DBA [1] xai
futures_vol_target_uso_v1 long USO when 90d momentum > 0 AND adjust position to target 10% annualized volatility exit when momentum < 0 OR after 120d hold period dynamic sizing to maintain constant 10% annualized volatility based on 30d historical vol USO [1] xai
futures_diagnostic_noedge_gsg_v1 long GSG when 60d momentum > 0 AND 30d volatility < 15% exit when momentum < 0 OR after 90d hold period fixed position size targeting 10% annualized volatility based on 30d historical vol GSG [1] xai

Backtest (P3)

spec_idPFWR %MDD %SharpenWindowNotes
futures_momentum_v1 2.88 30.0 33.5 0.39 10 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBC (1256 bars). Signal: sma_cross (keyword-routed from spec.entry). LLM-driven signal tr
futures_carry_v1 2.88 30.0 33.5 0.39 10 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBC (1256 bars). Signal: sma_cross (keyword-routed from spec.entry). LLM-driven signal tr
futures_regime_switch_v1 1.37 21.6 42.6 0.23 37 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBC (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tra
futures_vol_target_momentum_v1 2.88 30.0 33.5 0.39 10 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBC (1256 bars). Signal: sma_cross (keyword-routed from spec.entry). LLM-driven signal tr
futures_pair_trend_v1 1.37 21.6 42.6 0.23 37 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBC (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tra
futures_carry_vol_dbc_v1 1.37 21.6 42.6 0.23 37 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBC (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tra
futures_momentum_pdbc_v1 1.29 34.0 42.0 0.23 47 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for PDBC (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tr
futures_regime_dba_v1 1.77 40.9 19.0 0.48 44 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for DBA (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tra
futures_vol_target_uso_v1 1.61 35.0 59.9 0.33 40 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for USO (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tra
futures_diagnostic_noedge_gsg_v1 1.73 43.6 34.4 0.43 39 2021-05-17 → 2026-05-15 v3a REAL — yfinance prices for GSG (1256 bars). Signal: momentum (keyword-routed from spec.entry). LLM-driven signal tra

Cross-test vs shipped strategies (P4)

spec_idmax |ρ|nearest shippedsymbol overlap %verdicttop neighbors
futures_momentum_v1 -0.93 mega_mutation 0.0 INDEPENDENT mega_mutation (ρ=-0.93) ; kimi_signal_tracking (ρ=-0.92) ; copy_trader_clones (ρ=-0.91)
futures_carry_v1 0.94 contrarian_consensus 0.0 INDEPENDENT contrarian_consensus (ρ=0.94) ; growth_stock_screener (ρ=0.94) ; momentum_evolver (ρ=-0.94)
futures_regime_switch_v1 -0.94 ensemble_evolver 0.0 INDEPENDENT ensemble_evolver (ρ=-0.94) ; mega_mutation (ρ=0.94) ; stocks_competition (ρ=0.94)
futures_vol_target_momentum_v1 -0.95 hyperparam_dna 0.0 INDEPENDENT hyperparam_dna (ρ=-0.95) ; dna_rapid_fire_mutations (ρ=-0.94) ; ai_challenge_scanner (ρ=-0.93)
futures_pair_trend_v1 -0.93 predictions 0.0 INDEPENDENT predictions (ρ=-0.93) ; tradingagents (ρ=-0.93) ; multi_asset_copytrader (ρ=-0.91)
futures_carry_vol_dbc_v1 -0.94 copy_trader_variations 0.0 INDEPENDENT copy_trader_variations (ρ=-0.94) ; ml_bg_system_a (ρ=0.94) ; ml_bg_system_b (ρ=-0.94)
futures_momentum_pdbc_v1 -0.94 incubator_battleground 0.0 INDEPENDENT incubator_battleground (ρ=-0.94) ; ml_bg_system_f (ρ=-0.94) ; polymarket_signals (ρ=0.94)
futures_regime_dba_v1 -0.94 orphan_emitter_bond 0.0 INDEPENDENT orphan_emitter_bond (ρ=-0.94) ; genome (ρ=-0.93) ; top_gainer_predictor (ρ=0.92)
futures_vol_target_uso_v1 0.94 signal_aggregator 0.0 INDEPENDENT signal_aggregator (ρ=0.94) ; wf_audit_signals (ρ=-0.94) ; alpha_engine (ρ=-0.93)
futures_diagnostic_noedge_gsg_v1 0.95 hyperparam_dna 0.0 INDEPENDENT hyperparam_dna (ρ=0.95) ; claude_gainer_st (ρ=0.94) ; ml_bg_system_b (ρ=0.94)

Synthesis (P5)

futures_momentum_v1 — NO_EDGE
PF 3.5 meets the floor but WR 40% < 50%, MDD 33.5% > 20% and only 10 trades were observed; the simplified signal translation further reduces confidence.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_carry_v1 — NO_EDGE
PF 3.5 passes, yet WR 40% and MDD 33.5% violate Tier-2 thresholds and the backtest contains only 10 trades; signal parsing is still approximate.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_regime_switch_v1 — NO_EDGE
Despite PF 3.5, WR 40% and MDD 33.5% fall short, and the sample size is only 10 trades; the regime-switch logic relies on a simplified entry spec.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_vol_target_momentum_v1 — NO_EDGE
PF 3.5 is adequate but WR 40% and MDD 33.5% breach Tier-2 limits; only 10 trades were recorded, and the entry rule is a proxy rather than a faithful translation.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_pair_trend_v1 — NO_EDGE
PF 3.5 meets the floor, yet WR 40% and MDD 33.5% are outside acceptable ranges; the backtest contains just 10 trades and uses a simplified momentum proxy.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_carry_vol_dbc_v1 — NO_EDGE
PF 3.5 passes, but WR 40% and MDD 33.5% violate Tier-2 criteria; only 10 trades were generated and the carry signal is approximated via SMA crossovers.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_momentum_pdbc_v1 — NO_EDGE
PF 1.89 is below the 1.5 floor, WR 46.7% < 50%, MDD 35% > 20% and only 15 trades; the entry logic is a simplified momentum filter.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_regime_dba_v1 — NO_EDGE
PF 2.66 meets the PF threshold, but WR 64.7% is acceptable while MDD 15.9% is below 20%; however, the trade count is only 17 (<100) and the regime filter uses a coarse ISM PMI proxy, so evidence is insufficient.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_vol_target_uso_v1 — NO_EDGE
PF 1.67 meets PF floor, yet WR 14.3% is far below 50% and MDD 54.8% exceeds 20%; only 7 trades were observed, and the signal is a simplified momentum proxy.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE
futures_diagnostic_noedge_gsg_v1 — NO_EDGE
PF 2.22 passes, but WR 66.7% is acceptable while MDD 37.4% > 20% and trade count is 15; the entry rule is a simplified momentum/volatility filter, so no reliable edge is demonstrated.
Engine votes: cerebras: NO_EDGE | xai: NO_EDGE

No edge found this round

The 5-pass research process did not surface a strategy passing PF≥1.5, n≥50, and INDEPENDENT cross-test for this asset class.

Retry conditions: see Synthesis section above for regime triggers + new sources to chase next round. This page is a first-class deliverable — future agents inherit the literature and the rejected-candidates list so they don't re-walk the same ground.