Cross-Sectional Momentum
Global core
Holds the strongest-trending names (12-month return, skipping the last month), equally weighted, refreshed monthly.
This strategy's universe (121) →
Worst stretch in the backtest: -65.0% — a simulated $1,000 would have dipped to ~$350 before recovering.
Concentration check: 2026 (to 2026-09-30) alone returned +144.6% — more than all the other years of this backtest together (-23.7%). The headline CAGR leans on that one period; see the chart data, year by year.
Backtest window: 2021-10-29 → 2026-09-30 · as of 2026-09-30 (updated quarterly)
This window does not include the COVID crash (Feb–Mar 2020) and the 2008 financial crisis.
Full methodology & limitations
Track records are simulated backtests, not live results. Methodology and current limitations: (1) Universe membership is real point-in-time index constituents - Nasdaq-100 (US) and WIG20 (PL), with dated add/drop history - so backtests see each index as it actually was, with no survivorship from later additions. Prices are adjusted close (total-return). When the backtest runs on the licensed, delisted-capable feed (EODHD), index members that later delisted or renamed are still priced and INCLUDED, closing the survivorship gap on the price side; when it runs on the keyless fallback feed, some delisted names are unpriced and dropped, which can flatter drawdowns. Smaller residuals remain: index changes after the last sourced revision are not yet reflected (Nasdaq-100 membership as pulled on 2026-06-30, WIG20 as of its September 2025 revision), and a few long-delisted tickers may still be unresolved. A name with no real quote for 14 days is treated as untradable - never picked and never held on a frozen price - and a renamed listing is held under its new code with its full history. (2) Costs are charged per market on each rebalance's turnover - 5 bps for US names, 10 bps for EU, 20 bps for Warsaw (GPW) - the conservative end of typical half-spread-plus-fees for liquid names; market impact and FX conversion are not modelled (negligible at retail size, understated for large money). (3) Prices are total-return (dividends reinvested) where the data feed supports it. (4) Sharpe and Sortino use a flat 3%/yr cash (risk-free) rate, approximately the average US T-bill yield over the backtest window; actual cash rates varied 0-5% within it and Polish rates were mostly higher, so ratios are approximate. (5) Cross-market calendar gaps are bridged by carrying the last known price forward, which can modestly understate day-to-day volatility. (6) Rebalances are simulated at the same closing price the signal is computed from (close-to-close execution); a live follower trades at the next session's prices, so simulated results may differ from what next-bar execution would have produced. Past simulated performance does not predict future results.
Hover any metric above — I'll explain what it means.
How you'd carry it out in your broker
Every model portfolio update publishes as a plain list of tickers and weights — the exact same file for every subscriber. Here's the general shape of loading that into four popular brokers; the actual tickers and weights only appear once you unlock this strategy.
This strategy has averaged 0.0 changes in the model portfolio per month, over its last publications.
XTB, Trading 212, IBKR, Revolut
XTB
- Each published ticker is searchable in the XTB app/web platform.
- XTB has no basket/pie import — each ticker's published weight is its share of the model portfolio; whether and how much to trade is your own decision.
Trading 212
- Portfolio → Pies → Create Pie.
- Each published ticker goes in with its published weight % — the Pie then tracks that allocation for you.
Interactive Brokers (IBKR)
- Once unlocked, the model portfolio downloads as a CSV.
- Classic TWS: File → Import Watchlist reads that file — each ticker's published weight is its share of the model portfolio.
Revolut
- Each published ticker is searchable in the Invest tab.
- Revolut has no basket import — each ticker's published weight is its share of the model portfolio; whether and how much to trade is your own decision.
General information about loading a published, identical-for-everyone list into a broker — not a personal recommendation. Whether and how much to trade is your own decision, in your own account; we never trade on your behalf.
Track record
Chart data (year by year)
| Year | Simulated return |
|---|---|
| 2021 (from Oct) | -15.3% |
| 2022 | -49.3% |
| 2023 | +3.4% |
| 2024 | +28.0% |
| 2025 | +34.2% |
| 2026 | +144.6% (to 2026-09-30) |
Dashed line = Equal-weight universe (same window and costs): CAGR 10.0%, max drawdown -32.3%.
Simulated backtest — methodology & limitations above, next to the stats.
This month's model portfolio
Loading…
How it works — in plain words
Once a month this model ranks the universe by their return over the last 12 months — but deliberately SKIPS the most recent month — and buys the top 5 in equal slices. That's it. No forecasts, no vibes: yesterday's persistent winners, re-checked monthly.
Imagine six stocks. Over the last 12 months (ignoring the most recent month) they returned: A +90%, B +55%, C +40%, D +20%, E −5%, F −30%. The model ranks them and buys the top 5 — A, B, C, D, E — in equal 20% slices. It makes no prediction; it simply rides whatever is already climbing. Next month it re-checks: if F has caught fire and E has fizzled, F swaps in and E drops out. Same mechanical ritual every month.
Prices overshoot in the very short run: last month's hottest name tends to snap BACK a little (short-term reversal), while the 2-12 month story tends to keep going (momentum). Two opposite forces on two clocks — so momentum models classically measure 12 months and throw the last one away. Transferable idea: always ask what CLOCK an effect lives on before you copy it.
Geczy & Samonov ran what they called 'the world's longest backtest': momentum showed up in US stock data going back to 1801 — over two CENTURIES of the same human habit.
Live since 2026-07-13 through 2026-11-06: -40.8% (3 mo) — the result of the published portfolios, entering at the next session's close. This is not a backtest.
Update history
| As of | What changed |
|---|---|
| 2026-11-01 | Model update: no material change this month |
| 2026-10-01 | Model update: no material change this month |
| 2026-09-03 | Model update: no material change this month |
| 2026-08-03 | Model update: no material change this month |
| 2026-07-21 | Initial model portfolio published (5 position(s)) |
Grounded in published research
The peer-reviewed findings this strategy is built on:
- Jegadeesh, N. & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance, 48(1).
- Asness, C., Moskowitz, T. & Pedersen, L. (2013). Value and Momentum Everywhere. Journal of Finance, 68(3).
A documented, peer-reviewed effect — cited so you can read the source yourself, not take our word for it. Implementation and results are ours (see methodology); the underlying finding is the literature's.
Understand this strategy
Plain-language explainers behind the mechanics above.
This strategy's rules and every model-portfolio update are written and run by HedgeYourOwn's founder.