Adaptive Market Intelligence for Non-Stationary, Regime-Switching Markets
Status: Research-grade prototype in private testing.
Day-trader oriented. Multi-timeframe. Designed for non-stationary, regime-shifting markets.
Bunny Quant is a quantitative research and trading support engine designed for non-stationary, regime-switching financial markets. It is built from the ground up to answer a simple but difficult question:
“If the market is constantly changing its behavior, how can we still extract reliable, actionable signals?”
Most classic indicators and many machine learning models implicitly assume some form of stationarity, ergodicity, or slow drift. In practice, markets exhibit:
Bunny Quant treats markets as a complex adaptive system, combining tools from fractals, distribution diagnostics, cycle detection, multi-agent reasoning, and ML-based filters into a single engine focused on real-time decision support.
Traditional technical indicators (RSI, MACD, moving averages…) and naive ML models share several limitations when used as standalone decision tools:
As a result, long-horizon ML prediction often looks good in backtest but collapses when regimes shift or new volatility structures appear. Bunny Quant does not try to predict a single “future price” far into the future; instead, it focuses on:
Bunny Quant is inspired by research in evolutionary computation, collective intelligence, and complex adaptive systems. Instead of assuming a fixed law of motion, it:
Short-term price action only makes sense inside a longer-term context. Bunny Quant explicitly integrates:
Bunny Quant is not a “one-click autopilot”. It is a signal engine:
The focus is on interpretable, regime-aware signals rather than opaque long-horizon prediction.
The current prototype of Bunny Quant integrates several internal modules (all research-grade, implemented from scratch):
Tools for analyzing fractal structure, roughness, and phase transitions in price series across timeframes, so that “trend vs chop vs squeeze” is treated as a measurable state rather than a vague impression.
Real-time checks on return distributions (tails, skew, kurtosis, clustering) to detect when the market deviates from usual behavior and enters abnormal risk zones.
Algorithms that track cycles, structural breaks, and regime persistence, helping traders align strategies with the current phase instead of fighting the tape.
A library of signals — from classic indicators (for intuition) to ML-driven features — filtered through regime-aware logic instead of naive fixed thresholds.
Hooks for integrating sentiment analysis and LLM-based narrative extraction, turning price-only views into richer representations of what the market is “talking about”.
Lightweight risk engines to contextualize signals with volatility, drawdown profiles, and regime-dependent risk estimates, especially important for intraday traders.
If you are interested in early access, collaboration, or fund-level deployment, please reach out: drnamlabs@gmail.com.
The live app will be hosted at:
app.drnamlabs.com
For now, the platform is under active development and internal testing. Once the public or invite-only beta is available, access instructions will appear here.