Cross-Asset Systematic Strategies — What Actually Works at Retail (2026)

Companion to 01-systematic-trading-doctrine.md. This document covers the empirical case for going beyond single-asset retail FX into a diversified cross-asset book, calibrated to what a sophisticated retail builder can realistically execute on OANDA, IBKR, Saxo, or CME micros.

The honest finding from the FX research: retail-cost single-asset FX (Sharpe 0.2–0.5 net, 7 pairs) is structurally undersized to compound. The literature is unambiguous that cross-asset diversification — even imperfect diversification — is the single biggest lever a systematic builder has. This document quantifies how big the lever actually is, where the lever fails, and what the next-tier retail expansion looks like.

1. The Diversification Math

1.1 Why 58 instruments hits Sharpe 1.2+ and 7 FX pairs doesn't

Moskowitz, Ooi, and Pedersen (2012) ran 12-month time-series momentum on 58 liquid futures and forward contracts — 24 commodity futures, 12 currency pairs, 9 equity indices, 13 bond futures — and reported a gross Sharpe of ~1.31 for the diversified portfolio (Quantpedia summary; the figure is widely cited). Individual asset-class Sharpes were materially lower: roughly 0.35 currency, 0.52 commodity, 0.72 equity-index, with bonds in between. The diversified portfolio Sharpe is roughly 2-3× the average single-asset-class Sharpe — a textbook diversification result.

The math is elementary. If you have n strategies each with Sharpe s and average pairwise correlation ρ̄, the portfolio Sharpe is:

S_portfolio = s × sqrt( n / (1 + (n-1)·ρ̄) )

Plug in MOP's numbers — average single-asset Sharpe ≈ 0.4, average pairwise correlation across 58 TSMOM signals ≈ 0.10:

S = 0.4 × sqrt(58 / (1 + 57·0.10)) = 0.4 × sqrt(58/6.7) ≈ 0.4 × 2.94 ≈ 1.18

That matches the reported figure to within rounding. Now plug in 7 FX pairs with the FX-only signal at retail (Sharpe ≈ 0.23 for Dollar Carry, average pairwise ρ̄ ≈ 0.50–0.65 because every USD pair shares the DXY factor):

S = 0.23 × sqrt(7 / (1 + 6·0.55)) = 0.23 × sqrt(7/4.3) ≈ 0.23 × 1.28 ≈ 0.29

This is exactly why the FX-only book stalls. Seven FX pairs is not seven bets — it is closer to 1.5 effective bets once you account for the dollar factor (Meucci's Effective Number of Bets, 2009). To get to Sharpe 1.0+ you have to attack one or both of (a) average signal quality s and (b) the correlation matrix ρ̄. Within FX you can do (a) only marginally. Across asset classes you collapse ρ̄ from ~0.55 to ~0.10–0.15, which is the diversification windfall.

1.2 Cross-asset correlation map (post-COVID, rolling 2020-2025)

Approximate post-COVID correlations of monthly returns of the headline instruments, sourced from World Gold Council, LSEG/FTSE Russell, Cambridge Associates 2025 Cross-Asset Outlook, and academic correlation reviews:

* Stock-bond correlation has flipped from the long-run negative ~-0.25 of 2000-2020 to positive ~+0.25 over 2022-2025 because inflation is the dominant shock, not growth. Cambridge expects this to soften but not return to the pre-2022 negative regime.

Three takeaways:

1. The cleanest diversifiers vs equities are gold and short-rate-divergent FX, not bonds. Bonds and equities are now in the same risk-on/risk-off bucket when inflation drives the cycle.
2. Bitcoin since the spot-ETF launch (Jan 2024) is a levered NASDAQ. 30-day rolling SPX–BTC correlation hit 0.74 in March 2026 (Phemex, Nasdaq research). It is not a hedge. It is a risk-on instrument with 3-5× equity volatility.
3. WTI is the most genuinely uncorrelated retail instrument in this list. It moves on supply shocks (OPEC, geopolitics) that don't drive equities or bonds. WTI is the highest-marginal-diversification add for an FX or equity book.

1.3 Practical retail floor: how many instruments before diversification kicks in

Empirically, the diversification curve flattens around 10–15 effective bets. With retail-grade pairwise correlation of ~0.20–0.30 between distinct asset classes (TSMOM-signal-on-asset, not raw asset correlation), Meucci's Effective Number of Bets math says you need roughly:

• 5 instruments across 3 asset classes → ~3 effective bets → Sharpe lift ≈ 1.7×
• 10 instruments across 4 asset classes → ~5 effective bets → Sharpe lift ≈ 2.2×
• 20 instruments across 5 asset classes → ~8 effective bets → Sharpe lift ≈ 2.8×
• 58 instruments across 4 asset classes (MOP) → ~10–12 effective bets → Sharpe lift ≈ 3.0–3.4×

The practical retail floor is 10 instruments across at least 4 asset classes. Below that, you are paying full retail costs for marginal diversification. Above ~20 instruments you are hitting infrastructure complexity (data, execution, monitoring) faster than you are buying Sharpe.

2. Asset Classes Accessible at Retail in 2026

2.1 Gold (XAU/USD) and Silver (XAG/USD)

Gold is the highest-value first add for an FX system: deep liquidity, no overnight financing if traded as futures, daily volatility ~1% (similar to FX majors), and decorrelated from both equities and bonds in the current regime. The 0.05 correlation with SPX is approximate — it spikes positive in QE/dovish environments (gold rallies with risk assets when real rates fall) and negative in genuine risk-off (March 2020, October 2022). Retail-net Sharpe for trend on gold standalone: 0.4–0.6 historically (AQR Century of Evidence; commodity sleeve).

2.2 Oil (WTI, Brent)

Oil is the strongest decorrelated add after gold. Watch out: USO ETF has catastrophic contango drag (April 2020: front-month went negative, USO took massive losses on roll). For systematic trend, use Micro WTI futures (MCL) — small notional, retail-accessible, no contango drag for active strategies that don't hold full roll cycles.

2.3 Equity indices (S&P 500, NASDAQ, DAX, FTSE, Nikkei)

Critical: adding SPX + NDX + DOW + Russell does NOT diversify. Pairwise correlations within US equity indices are 0.92–0.97 (S&P-NASDAQ ~0.94, S&P-Russell ~0.90). One US large-cap exposure is one bet. Genuine equity diversification requires geographic breadth: SPX (US) + DAX (Eurozone) + Nikkei (Japan) + Hang Seng or A50 (China) + ASX (Australia). Even then, correlations have risen — DAX-SPX is ~0.75 since 2020, Nikkei-SPX ~0.55. The diversification benefit is real but smaller than the cross-asset benefit.

2.4 Government bonds (US 10Y, German Bund, UK Gilt, Japan JGB)

Two cautions: (a) Bond futures have directional carry that interacts with TSMOM signals — your "trend" on bonds is partly a yield-curve roll trade; size accordingly. (b) Post-2022, bonds are no longer a clean equity hedge. They are a real-rate sensitivity trade. They still diversify a pure FX book (bond-FX correlation ~0.30) but less than they used to diversify equities.

2.5 Industrial metals (Copper, Aluminum)

Copper is "Dr. Copper" — high beta to global growth, partly correlated with equities (~0.40) but with a distinct China-demand factor. Useful add if your book is underweight commodity exposure; otherwise marginal compared to gold + oil.

2.6 Crypto (BTC, ETH)

Honest assessment: post-spot-ETF (Jan 2024), BTC is not a separate edge — it is a levered NASDAQ. SPX-BTC 30-day correlation has run 0.55–0.94 in 2024-2026. Adding BTC to an equity-heavy book ADDS concentration, not diversification. The exception is trend-following on BTC specifically: high volatility and serial trending behavior have historically given decent TSMOM Sharpes (0.7–1.2 gross in backtests through 2024), but post-ETF this is fading. Treat BTC as "high-vol equity proxy with a side-bet on regulatory regime change," not as a diversifier.

3. Documented Cross-Asset Strategies That Work

3.1 Time-Series Momentum (TSMOM) — Moskowitz, Ooi, Pedersen 2012; Hurst, Ooi, Pedersen 2017

Signal: long if past 12-month return > 0, short if < 0. Size to constant volatility per instrument.

The honest retail-net Sharpe estimate for DIY TSMOM across 10-15 instruments at IBKR/CME micro contract costs: 0.45–0.75. That is a 2-3× lift over a 7-pair FX-only TSMOM book.

3.2 Cross-Sectional Momentum across futures — Asness, Moskowitz, Pedersen 2013

Signal: rank all instruments in a class by 12-month return; long top tercile, short bottom tercile, vol-targeted.

Reported Sharpes (AMP 2013, "Value and Momentum Everywhere"):

• Country-equity index momentum: ~0.72
• Currency momentum: ~0.35
• Commodity momentum: ~0.52
• Bond momentum: ~0.40
• Combined cross-asset momentum: ~1.0 gross
• Value + Momentum 50/50 combined across asset classes: ~1.6–1.8 gross

The value-momentum negative correlation is the key result: across asset classes, value and momentum are -0.40 to -0.60 correlated, so the 50/50 blend captures both factors with much smaller drawdowns than either alone.

Retail-net Sharpe estimate: 0.5–0.9 for the blend, assuming monthly rebalance and 10–15 instruments. The bottleneck is short legs — many retail brokers don't allow shorting individual commodity ETFs without margin upgrades, but futures and CFDs handle shorts cleanly.

3.3 Carry — Koijen, Moskowitz, Pedersen, Vrugt 2018

Signal: long high-carry instruments, short low-carry, across asset classes. Carry is defined as "expected return if price doesn't change" — for bonds it's yield, for currencies it's interest-rate differential, for commodities it's the futures basis (contango/backwardation), for equities it's dividend yield + buyback yield.

Reported Sharpes (KMP 2018):

• Currency carry: ~0.40
• Bond carry: ~0.85
• Commodity carry: ~0.55
• Equity carry: ~0.70
• US Treasury slope carry: ~1.10
• Diversified carry across all asset classes: ~1.20 gross
• Range: 0.37 (call options) to 1.80 (put options)

Retail implementation is harder than TSMOM because measuring carry on each asset class requires different data: futures curves for commodities, OIS rates for currencies, term spreads for bonds. But it's doable with IBKR data plus FRED.

Retail-net estimate: 0.5–0.8 Sharpe for a 4-asset carry book.

3.4 Risk Parity / All Weather — Dalio (Bridgewater)

Signal: no signal — strategic allocation by inverse volatility, rebalanced quarterly. Designed to be robust across four economic regimes (rising growth, falling growth, rising inflation, falling inflation).

DIY retail version (canonical):

• 30% equity (VTI or split US/international)
• 40% long-dated bonds (TLT)
• 15% intermediate bonds (IEF)
• 7.5% gold (GLD)
• 7.5% diversified commodities (DBC or PDBC)

Historical Sharpe of the basic All Weather (1996–2022): ~0.6–0.8 unlevered (~6% return, ~8% vol). Bridgewater's leveraged institutional version reportedly ~1.0 Sharpe. The retail ALLW ETF (State Street SPDR Bridgewater All Weather, launched March 2025) has delivered higher risk-adjusted return than 60/40 since launch but the live sample is short.

Critical weakness exposed 2022: stock-bond correlation went positive, so risk parity took its worst drawdown ever (~-25%). The "All Weather" framing implies four regimes are equally probable; in practice the dominant inflation shock of 2022 broke the model. Bridgewater partially recovered in 2023-2024.

Retail-net Sharpe estimate: 0.5–0.7 for a DIY 5-ETF All Weather. It's a baseline allocation, not alpha — but it diversifies an FX trading book powerfully because their correlation is near zero.

3.5 Trend Following on Commodities Only — AQR / Man AHL

The commodity sleeve of TSMOM historically delivered the second-strongest single-asset-class Sharpe (~0.52 from MOP, higher in some replications). Commodities are the most genuinely uncorrelated asset class — physical supply and demand drives them, not central bank policy.

Retail-net estimate for a 6-8 commodity trend book (gold, silver, copper, WTI, natural gas, corn, soybeans, sugar) via micro futures: 0.4–0.7 Sharpe, with high return-on-margin given low overnight margin on micros.

The catch: 2011-2019 was a near-decade of poor trend-following returns ("trend winter") as low volatility and central-bank-induced reversals destroyed the autocorrelation. 2022 reversed that — SG Trend Index +24% — but the 10-year drawdown of trend funds is a real failure mode that retail traders must underwrite.

4. Specific Retail-Feasible Portfolios

4.1 Portfolio A — "Mini All-Weather (Strategic Risk Parity)"

Instruments (5): VTI, VXUS, TLT, GLD, PDBC (or via IBKR equivalents)

Rebalance: quarterly to fixed weights (e.g., 25/15/30/15/15) OR monthly to inverse-vol weights.

Expected gross Sharpe: 0.7–0.9 unlevered; 1.0–1.2 with modest 1.5× leverage via IBKR portfolio margin.

Retail-net Sharpe: 0.5–0.7 after ETF expenses (avg ~0.20%) and any IBKR margin interest.

Key failure modes:

• Inflation regime (stock-bond positive correlation broke the model in 2022; -20% drawdown).
• Long-duration bond risk (TLT can drop 30% in a rate-hike cycle).
• No alpha component — just diversified beta. Beats most retail traders but doesn't compound at hedge-fund rates.

Why a builder runs this: it's the world's most boring add to an active FX system, but it's the cleanest way to add 4 truly different return drivers. Run it as a "core" while your FX book runs as "satellite."

4.2 Portfolio B — "Diversified TSMOM (15-instrument)"

Instruments (15):

• FX (5): EUR/USD, GBP/USD, USD/JPY, AUD/USD, USD/CAD (existing FX book)
• Equity index (3): MES, MNQ, MFS (Micro DAX) — or SPY, QQQ, EWG
• Commodity (4): MGC (gold), MCL (WTI), MHG (copper), MSI (silver)
• Bonds (2): ZN (10Y note), FGBL (Bund)
• Crypto (1): MBT (Micro BTC) — optional, treat as high-vol equity

Signal: 12-month TSMOM (sign of 252-day return) with monthly rebalance and 10% volatility target per instrument; portfolio vol-scaled to 10% annualized.

Expected gross Sharpe: 0.9–1.2 (MOP-style universe, smaller-N adjustment).

Retail-net Sharpe: 0.5–0.8 after IBKR commissions (~$1–2 per trade per instrument), spreads, and slippage on micros.

Capital requirement: ~$25k–50k minimum to hold all 15 instruments at meaningful size on micros without margin breach during drawdowns. $100k+ is more realistic for proper sizing.

Key failure modes:

• "Trend winter" (2011–2019 type periods of low volatility / frequent reversals).
• Bond TSMOM signal becomes a duration bet when rates trend, magnifies in 2022-type shocks.
• Equity index TSMOM tracks the broad market — adds modest signal but high correlation among the 3 equity instruments. Could reduce equity sleeve to 2 if simplifying.

Infrastructure needs:

• Multi-asset data feed (IBKR or independent like Norgate, ~$30-50/mo).
• Single execution layer (IBKR API handles all 15).
• Per-asset cost model (different spreads, commissions, swap rates).
• Volatility forecasting per instrument (EWMA or GARCH).
• Correlation matrix re-estimated monthly.

4.3 Portfolio C — "Carry + Trend Hybrid (10-instrument)"

Carry sleeve (5 FX pairs): existing Dollar Carry book — long high-rate, short low-rate G10 currencies, monthly rebalance.

Trend sleeve (5 non-FX): TSMOM on MGC (gold), MCL (WTI), MES (S&P 500), ZN (10Y notes), MHG (copper).

Combined: 50% capital to each sleeve, vol-targeted to 8-10%.

Expected gross Sharpe: 0.8–1.1 (carry and trend are -0.10 to -0.30 correlated across asset classes — they hedge each other in crises).

Retail-net Sharpe: 0.5–0.8.

Key failure modes:

• Carry blow-up (2008 AUD/JPY -22% in 6 weeks; 2024 yen carry unwind).
• Trend whipsaw (2018-2019 commodity sideways markets).
• Crisis correlation spike (carry and trend can BOTH be wrong if a black swan flips signs on a Friday).

Why this works: this is the closest retail approximation of the AQR / Man AHL multi-strategy approach. Carry and trend are the two most-replicated cross-asset factors. They are negatively correlated, so combined they smooth the equity curve materially.

4.4 Portfolio D — "Value + Momentum Everywhere (institutional-lite)"

Instruments: 8 country ETFs (US, UK, Germany, Japan, Australia, Canada, Switzerland, EM) + 8 currency pairs + 4 bond ETFs + 6 commodity ETFs/futures = ~26 instruments.

Signal: Asness-Moskowitz-Pedersen 50/50 value-momentum blend. Value = book-to-price (equities), PPP deviation (FX), term spread (bonds), spot-vs-5Y-average (commodities). Momentum = 12-month return ex-most-recent-month.

Expected gross Sharpe: 1.4–1.8 (per AMP 2013 paper).

Retail-net Sharpe: 0.7–1.0 — this is the upper bound of what's realistically achievable at retail without prime brokerage.

Why most retail builders shouldn't do this: complexity. Value signals are different for every asset class. Monthly rebalance across 26 instruments means many small trades. Costs compound. Only consider this if you have >$200k capital and are willing to invest 6-12 months in research/build.

4.5 Portfolio E — "The Conservative Add" (highest ROI move)

Instruments (3 added to existing FX book):

• MGC (Micro Gold) — single trend-following position, 12-month TSMOM
• MCL (Micro WTI) — single trend-following position, 12-month TSMOM
• TLT or ZN — single trend-following position on long-duration bonds

Capital: marginal — micros require ~$500–1500 overnight margin each.

Expected lift to existing 7-pair FX book: Sharpe from ~0.25 to ~0.45–0.55 (rough estimate). This is the highest marginal-Sharpe-per-effort addition because:

• Each instrument is nearly uncorrelated with FX (gold-USD ~-0.30, oil-FX ~0.10, bonds-FX ~0.30 but the trend signals decorrelate further).
• Each requires only one signal to maintain (12-month TSMOM).
• Each has standardized retail access via CME micros.
• No new data feeds needed if you already have IBKR.

This is the recommended Phase 2 expansion path. See Section 5.

5. Phase 2 Expansion Plan for a Retail FX System

Starting point: 7-pair FX system, Sharpe 0.2–0.5 net, running on OANDA or IBKR with Python execution.

Phase 2A (0–3 months): Add 3 Macro Diversifiers — gold + oil + 10Y bonds

Tier: Lowest effort, highest Sharpe lift per unit work.

Effort:

• 2-4 weeks of research/backtest per instrument (12-month TSMOM is well-documented).
• Migrate execution from OANDA-only to IBKR (most OANDA accounts don't trade CME micros).
• Add per-asset volatility scaling and cost model.
• Walk-forward 3 months on paper, then live with 0.1× size.

Broker requirements: IBKR Pro with futures permissions (US/EU residents both can get this). Approx $10k minimum effective capital to size positions meaningfully.

Infrastructure: extend existing Python pipeline to handle futures contract rolls (each instrument rolls quarterly). Tools like ib_insync handle this in ~200 lines.

Expected outcome: Sharpe 0.45–0.60 net, max drawdown 15-20%, materially smoother equity curve.

Phase 2B (3–9 months): Add 4-5 More — international equities, copper, silver, second bond

Tier: Medium effort. Largely incremental once Phase 2A infrastructure is in place.

Effort:

• Each new instrument adds ~1 week of validation work.
• Equity ETFs (EWG, EWJ, EWU) can run via OANDA or IBKR with simpler execution than futures.
• Add geographic equity exposure to capture the within-equity diversification (which is meaningful even though correlations are 0.5–0.8).

Expected outcome: 12-15 instrument book, Sharpe 0.55–0.75 net, drawdowns 12-18%.

Phase 2C (9–18 months): Add a Second Signal — Carry alongside Trend

Tier: High effort. Materially different code path.

Effort:

• Implement Koijen et al. carry across 4 asset classes (FX rates already in your FX book; equity dividend yields from IBKR; commodity basis from CME data; bond term spreads from FRED).
• Carry and trend are -0.10 to -0.30 correlated, so 50/50 blend dramatically reduces drawdowns.
• 6 months walk-forward minimum before allocating real capital.

Expected outcome: Sharpe 0.65–0.85 net. The expected drawdown improvement is bigger than the Sharpe improvement — carry-trend blend's worst year is meaningfully better than either alone.

Phase 2D (18+ months): Consider Value or Cross-Sectional Momentum

Only if (a) you've validated Phase 2C, (b) capital > $150k, (c) you have the bandwidth to maintain value signals (different per asset class).

Skip: micro-second execution, market-making, options strategies. These require infrastructure outside retail reach.

Honest realism check

Even after a full 18-month Phase 2 buildout, a retail systematic cross-asset book is most likely to land at Sharpe 0.6–0.9 net, with 12-18% volatility and 15-25% max drawdowns. That's enough to compound real money over 5-10 years but it is NOT a "retire-in-three-years" outcome. Anyone presenting cross-asset systematic trading at retail as a higher-Sharpe layup is either describing institutional implementation or selling something.

6. What Does NOT Diversify

6.1 Adding more USD pairs

Adding USD/CHF, NZD/USD, USD/SEK to your existing EUR/USD, GBP/USD, USD/JPY, AUD/USD, USD/CAD does not diversify. They all share the DXY factor. Pairwise correlations within USD pairs range 0.40 (USD/JPY vs EUR/USD) to 0.85 (EUR/USD vs GBP/USD). The effective number of bets in a 10-pair USD book is approximately 2.5, not 10. To genuinely diversify within FX you need to add cross-pairs (EUR/JPY, GBP/JPY, AUD/NZD, EUR/CHF) which still share regional factors but at least decouple from pure dollar exposure. Even then, you're adding marginal — not transformational — diversification.

6.2 Adding S&P 500 + NASDAQ + Dow + Russell 2000

US large-cap equity indices are 0.90–0.97 pairwise correlated. Adding all four to your book adds approximately 1.05 effective bets, not 4. Worse, in crisis they correlate even higher (March 2020, the four indices were essentially identical). Genuine equity diversification requires geographic breadth: at minimum SPX + DAX + Nikkei. Even better: SPX + EM (EEM or IEMG) which has ~0.65 correlation with US large-cap and runs on different macro drivers.

6.3 Gold as an equity hedge

The "gold is a crisis hedge" narrative is partially true but more nuanced than retail dashboards suggest:

• 2008 crisis: gold initially DROPPED with equities (Aug-Oct 2008, gold -30%) before rallying through 2011 as QE drove real rates down. Not a clean hedge in the actual moment.
• 2020 COVID: gold rallied with equities post-Fed pivot (March 23 onward) — both were beneficiaries of money printing.
• 2022 inflation shock: gold underperformed both stocks and bonds for most of the year (gold -3%, SPX -19%, TLT -31%) — not a hedge.
• 2024-2025 rally: gold rallied alongside equities (both at record highs together) — gold-equity correlation turned positive.

Gold is a real-rates and liquidity-conditions trade more than a crisis hedge. It diversifies a portfolio meaningfully (correlation near zero on average) but it does not specifically protect against equity drawdowns. The instruments that actually hedge equity drawdowns are long volatility (VIX futures, but these have negative carry) and long-dated Treasuries (which only work in deflationary recessions, not inflationary ones).

6.4 Crypto as a "diversifier"

Pre-2020, BTC was ~zero correlated with everything — a genuine diversifier. Post-2020, and especially post-spot-ETF (January 2024), BTC has become a levered NASDAQ:

• 30-day rolling SPX-BTC correlation: 0.55–0.94 in 2024-2026.
• BTC standard deviation: 3-5× SPX.
• BTC behavior in equity drawdowns (Aug 2024, April 2025): drops at higher beta than NDX.

The "BTC adds diversification" thesis is empirically dead since the ETF launch. BTC may still offer trend-following alpha (high serial autocorrelation in volatile assets) but it does NOT provide portfolio diversification anymore.

6.5 Multiple bond maturities

Adding TLT (20Y+), IEF (7-10Y), and SHY (1-3Y) does provide some diversification because the yield curve doesn't move in parallel. But correlations are still 0.75–0.95. One bond duration exposure plus possibly one shorter-maturity for risk management is enough.

6.6 Multiple commodity ETFs (DBA, DBC, USCI, PDBC)

These are all broad commodity indices with 0.85+ pairwise correlation. Pick one. If you want diversified commodity exposure, hold individual commodity futures (gold + oil + copper + ag) rather than multiple broad-basket ETFs.

7. Recommendation

If a builder has already shipped a working retail FX system today, the 3 most-leveraged cross-asset additions are:

1. Micro Gold (MGC) with 12-month TSMOM

Why: Highest marginal Sharpe-per-effort move. Gold runs on real-rates + liquidity factors that are largely orthogonal to G10 FX rate-differential carry. Add one instrument, one signal, one position-sizing rule — and Sharpe lift of roughly +0.10–0.15 over an FX-only book. CME micro contract requires only ~$1k overnight margin, available on IBKR, no new infrastructure beyond futures-roll handling.

2. Micro WTI Crude (MCL) with 12-month TSMOM

Why: The most genuinely uncorrelated retail instrument in the universe. Oil moves on supply shocks (OPEC, geopolitics, inventory data) that don't drive any FX major. Pairwise correlation with the FX book ~0.05–0.15. Adding it after gold compounds the diversification — you go from ~2-3 effective bets (FX-only) to ~5-6 effective bets, the bulk of the diversification windfall. Same retail access path as MGC; trivial incremental infrastructure.

3. The Carry signal extended to bonds and commodities (Koijen-Pedersen carry across asset classes)

Why: This is the highest-leverage signal extension, not just an instrument extension. You already have Dollar Carry working in FX (Lustig-Roussanov-Verdelhan). Extending the same conceptual signal — "go long high-yield, short low-yield" — to bond term spreads and commodity futures bases adds an entirely new return driver with -0.20 to -0.30 correlation to your existing FX carry. The Koijen 2018 paper documents 1.20 gross Sharpe for the diversified carry portfolio. Retail-net 0.5–0.7 is plausible. Effort: 2-3 months of research/build, mainly because the data sources differ per asset class.

Why these three and not 15 instruments at once: marginal Sharpe per unit-of-effort. Each of these three adds ~0.10–0.20 Sharpe to a 7-pair FX book at ~2-4 weeks of engineering work each. That's a 2-3× total Sharpe lift in 3 months. Adding the full MOP 58-instrument universe would be ~3× the Sharpe lift but ~10× the engineering complexity, plus data costs, plus monitoring overhead. The first three additions give you 80% of the diversification benefit for 20% of the effort — the canonical 80/20.

Honest expectation after these three additions: Sharpe 0.5–0.7 net, max drawdown 15-20%, CAGR 5-10% net of costs at 8-12% volatility target. Real money over 5+ years; not a get-rich quickly path. If a builder expects more from retail systematic trading, they are either lying to themselves or have not yet been on the wrong side of a 2008, 2015, or 2022 regime shock.

Sources

• Moskowitz, T., Ooi, Y., Pedersen, L. (2012). Time Series Momentum. Journal of Financial Economics 104(2): 228–250.
• Hurst, B., Ooi, Y., Pedersen, L. (2017). A Century of Evidence on Trend-Following Investing. AQR.
• Asness, C., Moskowitz, T., Pedersen, L. (2013). Value and Momentum Everywhere. Journal of Finance 68(3): 929–985.
• Koijen, R., Moskowitz, T., Pedersen, L., Vrugt, E. (2018). Carry. Journal of Financial Economics 127(2): 197–225.
• Babu, A., Levine, A., Ooi, Y., Pedersen, L., Stamelos, E. (2019). Trends Everywhere. AQR.
• Meucci, A. (2009). Managing Diversification. Risk. (Effective Number of Bets.)
• Cambridge Associates. 2025 Outlook: Cross Asset.
• World Gold Council. Gold's optimal portfolio weight in a higher correlated environment.
• LSEG / FTSE Russell. Digital Assets: Evolution and Correlations.
• Bridgewater All Weather ETF (ALLW). State Street SPDR fact sheet.
• Faber, M. Cambria Global Asset Allocation (GAA).
• Interactive Brokers. Futures & FOPs Margin Requirements; Micro Futures Comparison; CFDs.
• CME Group. Micro E-mini Equity Index Futures FAQ.
• SocGen. SG Trend Index Methodology.
• BarclayHedge. SG Trend Index.
• Picture Perfect Portfolios. DBMF vs KMLM vs CTA Comparison.
• Phemex / Nasdaq research. Bitcoin-S&P 500 Correlation Analysis.
• Lustig, H., Roussanov, N., Verdelhan, A. (2014). Countercyclical Currency Risk Premia. Journal of Financial Economics.
• Della Corte, P., Riddiough, S., Sarno, L. (2016). Currency Premia and Global Imbalances. Review of Financial Studies.