Redlight helps you reduce drawdowns by switching from risk-on to defensive posture when conditions deteriorate.
We don't predict the future. We separate market regimes into actionable states.
System paused or insufficient data integrity. Safety defaults tailored to preservation.
Low volatility, constructive price action. Historical context favors risk-on exposure.
Deteriorating internals or rising volatility. Caution warranted; partial hedging advised.
High probability of drawdown. Capital preservation becomes the primary objective.
* Research metrics based on backtested performance (2020-2025). How is this measured?
Not investment advice. Past performance isn’t indicative of future results.
* Methodology: Validated on SPY data (2000-2024). "Monotonic Risk Profile" refers to the observation that crash probability increases strictly as the signal degrades (Green < Orange < Red). Lift metrics compare crash frequency on Red days vs baseline unconditional frequency.
Hard-coded safety checks prevent the system from firing on bad data or irrational exuberance.
Gradient boosting models analyze volatility structure and market internals to probability-weight near-term outcomes.
Example System Output
Check Integrity_Sentinel...
>> DATA_OK (Freshness: 0s)
Run Phase_A_Feature_Extraction...
Calculating Composite_Score...
> SYSTEM_STATUS: OPTIMAL
> REGIME_ESTIMATE: CAUTION (Risk 62/100)
Absolutely not. Redlight is an experimental research tool in Alpha. It will produce false signals. It is designed to interpret volatility structure, not to predict the future. Never trade based solely on these signals.
We commit to maintaining Redlight for a minimum of 12 months ensuring Founder value. If the project does not find sustainable product-market fit by then, we may sunset operations. We are building this with you, but we are honest about the experimental nature.
No. We provide data visualization and risk metrics. We are not investment advisors. You are solely responsible for your capital.
The Founder tier is for early believers who want to support the infrastructure costs of running these models. In exchange, you get lifetime access if we succeed, and at least a year of data if we don't.