A complete, no-marketing walkthrough of the pipeline behind every cycle: how data becomes a signal, how a signal becomes a trade, and how a trade becomes a transparent line in your portfolio.
Three independent AI engines run on fixed 2H, 6H, and 12H cycles. Each cycle, models score the market, a strict quality filter rejects almost everything, and only high-confidence setups are executed with hard stops and pre-sized risk. You only pay a fee — 2% — on cycles that actually earn profit.
Every engine, on every cycle, executes the same disciplined sequence.
Each engine continuously pulls real-time price action, orderbook depth, funding rates, open interest, on-chain flows, and the volatility surface across major centralized venues. Data is normalized, deduplicated, and timestamped before any model ever sees it.
We treat data quality as a first-class problem. Bad ticks, exchange outages, and stale feeds are detected and quarantined automatically — a model is only as honest as the inputs it receives.
Models evaluate dozens of features per candidate setup — trend structure, momentum, volatility regime, liquidity, microstructure, and cross-asset context — and output a calibrated confidence score with directional bias.
Confidence is not a marketing word here. Each engine has a target hit-rate and expected payoff, and the score is calibrated against historical out-of-sample outcomes so a 70% score actually means roughly 70%.
Only setups above the engine's confidence threshold survive. Most cycles end with zero trades — and that is the point. Forcing a trade is the single fastest way to lose money in crypto.
Thresholds are tuned per engine. The 2H engine is more reactive and accepts a higher trade frequency at smaller targets; the 12H engine waits for fewer, cleaner setups with larger expected moves.
Each accepted trade enters with a hard stop, a defined target, and a position size that is risk-budgeted to a fixed fraction of capital — typically 0.5% to 1% of the engine's allocation per trade.
There is no averaging down, no widening of stops, and no 'hoping it comes back'. Once a trade is live, the only outcomes are: target hit, stop hit, or time-based close at cycle end.
Cycles close cleanly. Realized profit and loss is written to your portfolio with a full audit trail: entry, exit, size, fees, and timestamps. The 2% performance fee is computed only on net positive cycles.
If a cycle ends flat or negative, no fee is charged on that cycle. You only pay when the engine actually makes you money.
Outcomes feed back into model retraining on a scheduled cadence. Engines adapt to new market regimes — bull, bear, chop, high-vol — instead of memorizing the last one.
Retraining is governed: every candidate model must beat the live model on a held-out, walk-forward backtest before it is allowed to take real risk.
Same pipeline, different timeframes, different risk profiles. Pick one, mix several.
Best for investors who want frequent, smaller realized cycles with tighter risk per trade.
Balanced middle ground — fewer trades than the 2H engine, larger expected moves per cycle.
Patient, highly selective. Many cycles result in no trade. When it does trade, it expects a larger move.
Every trade has a predefined maximum loss. No exception, no override.
We do not chase trade count. A cycle with zero trades is a valid, often correct outcome.
Engines run on fixed cycles. Decisions are scheduled, not emotional.
2% performance fee on profits only. If you don't earn, we don't earn.
What actually happens between one cycle tick and the next.
More detail on infrastructure and security: see the Technology page.
Targets shown for each engine are simulated, derived from historical and walk-forward backtests. Crypto markets are volatile and losses are possible at any time. Past or simulated performance is not indicative of future results.
Read the full risk disclosureIt varies by cycle and regime. The 2H engine may take several trades per day; the 12H engine often takes none for stretches. Inactivity is a feature, not a bug.
Yes. Engines are independent and can be allocated to in any combination. Many investors split allocation across all three to diversify cadence and risk.
Only on net profitable cycles. If a cycle closes flat or negative, no performance fee is taken on that cycle.
Open positions are managed by their hard stops and targets. If infrastructure detects an anomaly, circuit breakers close exposure conservatively.
No. Each investor's allocation is tracked individually and trades are reported per account.
Explore live cycle targets, cadence, and minimum allocation for each engine.