RTTADocumentation Built by Goblin Reactor ↗

MatchedFlowConformalSignal

Incremental, causal technical analysis documentation

Summary

MatchedFlowConformalSignal is RTTA's streaming implementation of: Intraday OHLCV matched-flow signal with conformal-style rolling error bands and target sizing diagnostics.

Update API

result = rtta.MatchedFlowConformalSignal().update(open, high, low, close, volume)

The update(...) call consumes one observation using open, high, low, close, volume. advance(...) uses the same inputs when the caller wants to update state without materializing a Python return value.

Theory Of Operation

MatchedFlowConformalSignal is a composite intraday research signal. It forms a horizon return forecast from multi-scale momentum, signed dollar-flow participation, VWAP displacement, and abnormal activity; then it scales that forecast by a rolling empirical error quantile. The longer research note linked below discusses the paper lineage and the conformal-style calibration caveats.

Recurrence

Let \(z_t = (open_t, high_t, low_t, close_t, volume_t)\) denote the observation consumed by one update(...) call and let \(\theta\) denote constructor parameters such as window lengths, thresholds, and smoothing constants.

\[r^{(k)}_t=\log(close_t/close_{t-k}), \qquad m_t=0.20r^{(3)}_t+0.35r^{(6)}_t+0.45r^{(12)}_t\]
\[DV_t=close_t\max(volume_t,0), \qquad relvol_t=\frac{DV_t}{normal\_dollar\_volume_t}\]
\[a_t=\operatorname{sgn}(r^{(1)}_t) \frac{close_t\,\max(volume_t,0)}{\operatorname{scale}_t}, \qquad p_t=\operatorname{sgn}(r^{(1)}_t) \frac{close_t\,\max(volume_t,0)}{normal\_dollar\_volume_t}\]

Here \(\operatorname{scale}_t\) is market capitalization when supplied and the normal dollar-volume baseline otherwise.

\[flow_t=\tanh\left(\frac{\sum_{i\in W^{12}_t}a_i}{\alpha_{norm}} +0.5\frac{\sum_{i\in W^{6}_t}p_i}{6}\right), \qquad vwap\_gap_t=\frac{close_t}{VWAP_t}-1\]
\[\widehat{r}_{t+h}= \frac{0.35m_t+0.001flow_t+0.05vwap\_gap_t +0.0005\tanh((relvol_t-1)/2)} {1+25\max(high_t-low_t,0)/close_t}\]
\[\mathcal{E}_t=\{|r^{(h)}_i-\widehat{r}^{(h)}_i|:\ i+h\le t\}, \qquad radius_t=\max(Q_{\tau}(\mathcal{E}_t), cost)\]
\[score_t=\frac{\widehat{r}_{t+h}}{radius_t+cost}\]

update(...) returns a result struct with fields prediction, radius, score, signal, target_fraction, alpha_flow, participation, flow_score, momentum, volatility, vwap_gap, rel_dollar_volume, max_trade_dollars, realized_error.

Implementation Notes

The recurrence is implemented in src/rtta/indicator.cpp in class MatchedFlowConformalSignal.

Reference