Eclipse Earn combines a smart stop-loss system with real-time predictive modelling, recalculating exposure thresholds as liquidity and volatility shift — rather than waiting for a fixed price level to be breached.
Every automated decision Eclipse Earn makes traces back through three distinct layers. Each layer has a defined input and output, which keeps the system auditable rather than opaque.
Order book data, volume profiles and volatility surfaces are ingested continuously and normalised before reaching any model. Latency between capture and input is measured in milliseconds, which matters when latent volatility detection depends on identifying a shift before it is visible on a standard chart.
Each open position carries a risk score derived from volatility clustering and current liquidity conditions. The score updates as conditions change, distinguishing ordinary price noise from the early stages of an asymmetric risk profile forming against the position.
When a risk score crosses a defined boundary, execution logic acts without requiring manual confirmation. This removes the delay — and the hesitation — that often accompanies a manual decision to cut a losing position.
A fixed stop-loss is set once and ignores everything that happens afterwards. Eclipse Earn's smart stop-loss system instead recalculates its threshold as liquidity depth and volatility clustering change — tightening protection when conditions deteriorate, and giving a position more room when volatility contracts and liquidity remains stable.
Because the adjustment is rule-based rather than discretionary, it removes a specific point of failure in high-stakes environments: the tendency to move, widen or ignore a stop-loss under pressure. The threshold is recalculated by the same logic regardless of how a trader feels about the position in the moment.
Predictive models in trading are often criticised for being a "black box" — a decision is made, but the reasoning behind it is not visible. Eclipse Earn is built around interpretable intermediate steps rather than a single opaque output.
Raw market data is standardised across instruments and timeframes before any model sees it, so a risk score means the same thing whether derived from a liquid index or a thinly traded contract.
Historical and live data are compared against recognised volatility and liquidity regimes. The output is a classification rather than a single opaque number, and each classification is logged with the conditions that triggered it.
Execution parameters are adjusted based on the classified regime and reviewed against subsequent outcomes, with each adjustment traceable back to the data and classification that produced it.
The table below describes operational behaviour, not a projection of returns. Outcomes vary by instrument, position size and prevailing conditions, and the following should be read as a guide to logic, not a guarantee of performance.
| Regime | System Behaviour | Typical Consideration |
|---|---|---|
| High Volatility Contraction | Stop-loss thresholds tighten in proportion to compressed liquidity; execution logic becomes more conservative. | Relevant for reducing exposure ahead of uncertain catalysts. |
| Trend Expansion | Thresholds widen as volatility clustering stabilises, allowing a position room to develop while liquidity remains intact. | Relevant for holding through a directional move without premature exit. |
| Liquidity Fragmentation | Risk scoring weights liquidity depth more heavily than price action alone. | Relevant for less liquid instruments or after-hours sessions. |
Short holding periods mean thresholds need to react within the same session; the system recalculates on intraday liquidity shifts rather than end-of-day data.
Classification logs and risk score history can be reviewed alongside existing models, supporting independent verification rather than blind reliance on a single signal.
For longer-dated positions, wider volatility clustering checks reduce the chance of an early exit triggered by short-term noise.
Historical regime behaviour does not guarantee future performance. Market conditions in practice may combine elements of more than one regime.