A full-universe walk-forward backtest of the Satis Omnibus strategy under the exact risk caps you actually trade — spot-only, no leverage, ≤10% total exposure, with realistic broker fees. Every number below comes from a real run; nothing is cherry-picked.
Simulated results — and a note on licensing. These are hypothetical, backtested figures on historical data, not a record of live client returns and not a promise of future performance. Satis Omnibus is in beta, and financial-services licensing (NZ FMA) is in progress — we are working toward registration and will operate within it. Nothing here is personalised financial advice.
⚠️ The figures on this page are superseded — read this first. Rigorous re-testing on 9 years of survivorship-bias-free data (walk-forward + out-of-sample + Monte-Carlo) showed the old "+128% / PF 1.28" headline was a per-instrument, ~0.1%-effective-risk envelope — not a portfolio you could actually trade. We've retired it, because honesty matters more than a big number. The validated edges that survived the filter: index+trend blend Sharpe 1.70 · trend sleeve 1.41 · crypto BTC regime-timing ~1.1 · Smart-Dip ~1.0 — all currently proving live in paper before any real capital. The detailed tables below are kept only for transparency; treat them as the old envelope, not a live expectation.
Read the backtest critically — and see the live version. A backtest has real limits, and we'd rather name them than hide them: the universe is today's tradeable instruments, so it can't include names that were delisted along the way (survivorship bias, which flatters results); it covers one period with no out-of-sample holdout; and while exits are now modelled gap-aware (a stop that gaps fills at the bar's open, not the level), any backtest still idealises fills. The honest counterweight is our live forward test — a real-time, out-of-sample paper record of the signals the running system generates, with every trade shown, winners and losers. Trust that over any historical curve.
What it's actually done so far: around break-even. We won't let the headline below be the impression you leave with. The figures on this page are a large-universe historical simulation; the running system's real-money and out-of-sample paper records to date are roughly break-even — not the backtested curve — on a small, early sample. That gap is exactly why we publish every live trade. The backtest shows where the design can work at scale (larger accounts, the shares book, sustained trends); it is not a claim about what your account will do. See the live record →
Backtested with the recommended settings at each account size, on survivorship-bias-free data, using the real broker fee models — Swyftx percentage fees for crypto, IBKR Pro's fixed ~$2.50/order for shares. Risk-adjusted return (Sharpe) is shown; the passive index benchmark is ~0.9.
| Account size | Recommended | CAGR | Sharpe | Max drawdown |
|---|---|---|---|---|
| $10k–$100k | 100% crypto (BTC regime-timing) | ~30% | 0.83 | 34% |
| ~$135k | Shares blend (trend + index + crypto) | ~20% | 1.36 | 21% |
| $250k | Full blend | ~20% | 1.31 | 18% |
| $500k | Full blend | ~22% | 1.42 | 15% |
Trade crypto until ~$135k, then diversify into shares. The validated BTC regime-timing edge (Sharpe ~0.83 / 30% CAGR) carries small-to-mid accounts and beats any shares mix while IBKR's fixed fee still bites. Only at ~$135k does that fee become a rounding error — and the shares trend + index blend takes over at Sharpe ~1.36, climbing to 1.42 at scale with far lower drawdown. Survivorship-free, walk-forward + out-of-sample; proving live in paper, not yet real money.
These are portfolio-constrained, survivorship-bias-free results (walk-forward + out-of-sample), net of modelled broker fees — not the old per-instrument envelope. Hypothetical, currently proving live in paper; not a promise of future performance. See the full risk disclosure and the live record on the forward test.
General guidance, not personalised advice. The dividing line is where IBKR Pro's fixed ~$2.50/order fee stops mattering: trade crypto until ~$135k, then move into the shares trend + index blend. Prove any change in Demo mode first.
| Account size | Book / broker | Max open | Exposure | Market-neutral overlay | Why |
|---|---|---|---|---|---|
| under ~$135k | Crypto (Swyftx) | 2 | 10% | None — crypto spot only | The validated BTC regime-timing edge (Sharpe ~1.1) with percentage fees — and it beats any shares mix while the fixed IBKR fee still bites. |
| ~$135k–$250k | Shares (IBKR Pro) + crypto | ~8 | 10% | Optional (~10–15%) | The fee is now a rounding error; the trend / regime-filtered-dip edge comes through (blend Sharpe ~1.36). A small long/short overlay can be added once comfortable. |
| $250k+ | Full blend | ~20 | 10% | Recommended (~15–25%) | The full trend sleeve runs as designed; blend ~half a passive index (Sharpe ~1.7), plus a market-neutral long/short overlay to cut drawdown further (uncorrelated — the diversification lift). |
Why crypto first? IBKR Pro charges a fixed ~$2.50 per order — on a small book, spread across trades, it dominates returns and keeps shares below crypto right up to ~$135k. Crypto on Swyftx charges a percentage with no fixed floor, and the BTC regime-timing strategy has a genuine edge at any size. Only once positions are large enough (~$135k) does the fee vanish and shares pull ahead.
The market-neutral overlay (and shorting). At larger accounts, a dollar-neutral long/short momentum sleeve — long the strongest names, short the weakest, ~0 net market exposure — is a validated diversifier: uncorrelated to the long-only book, it cushions drawdowns and earns when markets fall (a blend roughly halves drawdown). It is not a standalone edge, and it needs an IBKR margin account to short — an opt-in that steps off the spot-only default, so it's off by default and never applies to the core long-only book or to crypto. It runs as a paper long/short engine first (proving it forward on the forward test) before any live, large-account decision.
The sweet-spot profile (a $100,000 account, concentrated positions, IBKR Pro fees, min-quality 2) running the recommended config over 7.4 months (5 Jan – 17 Aug 2026) across 649 spot instruments.
Why the win rate is below 50%. The strategy is designed to lose small and win big: stop-losses cut losers fast (roughly half of all trades, 0% win rate by design), while the ratcheting trailing exit lets winners run (about a third of trades, ~98.6% win rate). The winners are large enough to more than pay for the frequent small losses — that's the positive expectancy.
Everything in this expander is a slice of the discredited $100k IBKR Pro run (14,168 trades, all modes, concentrated positions). Every dollar figure nets back to the retired +$128,600 headline — it is not a live expectation. The validated, portfolio-level results are the tables at the top of this page; the full per-strategy evidence lives on the Research Lab.
How trades ended. The trailing floor captures large moves; the stop-loss keeps every loss capped.
DIP buys pullbacks in an uptrend; POSITION enters confirmed continuation; STRICT requires all core conditions. All three run side by side.
Signals are scored 1–3 stars by indicator confluence. The edge is broad — lower tiers still carry a real edge. On IBKR Lite (free commission) it paid to take every signal; on IBKR Pro, where each order costs a fixed fee, the recommended config sets min-quality 2 to skip the weakest 1-star DIP signals whose edge doesn't clear the commission.
A real edge survives heavy costs. IBKR Pro's fixed per-order commission is the stress test — it's the harshest fee structure available to you, applied to a high-turnover strategy.
We retired the "+128%" figure: honest re-testing showed it was a per-instrument, ~0.1%-effective-risk envelope, not a tradeable portfolio. The real, validated edges are shown above and on the Research Lab.
The median trade is a small loss by design; a minority of large winners drives the positive expectancy.
The five most profitable instruments by net P&L. Of the instruments with a meaningful sample, 276 were profitable and 98 were not — the edge is broad, not concentrated in a handful of names.
Per-instrument P&L is small in absolute terms because each position is capped at ~$250 (2.5% of a $10k account). The full per-instrument breakdown is available inside the app's backtest simulator.
How the backtest was run and what assumptions were made.
Walk-forward on 4H bars with a daily higher-timeframe gate, across 649 spot instruments (crypto, US & global equities, ETFs, spot index ETFs, and spot commodity ETFs; forex excluded by the recommended config). 100% spot — no futures, no leverage. 200-bar warmup excluded from trading. Period: 5 Jan – 17 Aug 2026 (7.4 months). Modelled on IBKR Pro fees.
Forex is excluded — on this spot strategy every FX pair was net-negative after IBKR's per-order minimum, so dropping forex improved returns, costs, win rate and drawdown.
Each position is capped at total exposure ÷ max-open-trades of the account (≤10% total exposure). Spot only, no leverage. Note: the ≤2% figure is the stop width (2% of price), not 2% of the account — effective account risk per trade is ≈0.1% (a ~5% position with a ~2% stop). The backtest applies the per-trade sizing/stop caps but, unlike the live engine, does not cap concurrent open positions — see the sizing note above.
Cost model: Swyftx/XRPL for crypto and Interactive Brokers (Pro) for everything else, including the fixed ~$2.50 per-order minimum, spreads and FX conversion. On the $100k profile, IBKR Pro fees consumed ~22% of gross profit — a heavy but survivable drag once positions are concentrated.
Three modes run together: DIP (pullback-in-uptrend), POSITION (confirmed continuation) and STRICT (all core conditions). All signals are taken (no quality filter, no Fib-confluence requirement — both reduced returns in testing).
Exits: hard stop-loss, a ratcheting trailing floor (the profit engine), or a 6-bar time-stop for capital efficiency.
Daily-EMA HTF trend gate. Circuit breaker pauses trading after 7 consecutive losses (tripped 288× this run, 1,045 trades skipped). Reward-to-risk 1.5×.
These controls are part of the result — the figures reflect a live risk-management layer, not raw signal performance.
Past performance does not guarantee future results. These are hypothetical, backtested results on historical data. Real trading involves slippage, spread and liquidity constraints not fully captured here, and results vary by broker, timing and account. Satis Omnibus is in beta and pursuing NZ FMA licensing; nothing here is personalised financial advice. Never trade more than you can afford to lose.
Start with a free account. Run the same backtest yourself, on any account size.