validation and profiles

lib/dynamics/ht_trendmode/HtTrendmode.md

@@ -1,4 +1,4 @@
-# HT_TRENDMODE: Hilbert Transform Trend vs Cycle Mode
+# HT_TRENDMODE: Hilbert Transform Trend vs Cycle Mode
 
 The Hilbert Transform Trend Mode indicator is a binary regime classifier that determines whether price action is dominated by trending behavior (output = 1) or cyclical/mean-reverting behavior (output = 0). It uses the full Ehlers Hilbert Transform pipeline — 4-bar WMA smoothing, Hilbert FIR filters, homodyne discriminator for period estimation, DC phase extraction, and SineWave indicators — then applies four decision criteria to classify the current regime. The implementation follows TA-Lib's Ehlers-faithful algorithm from the February 2002 publication. Output is discrete {0, 1}, making it a direct strategy selector: deploy trend-following logic when mode = 1, and mean-reversion logic when mode = 0.
 
@@ -151,6 +151,44 @@ On each bar (price, isNew):
 | Long run of 1s | Strong, sustained trend |
 | Rapid 0/1 flipping | Transitional/choppy — reduce exposure |
 
+## Performance Profile
+
+### Operation Count (Streaming Mode)
+
+HtTrendmode uses the Hilbert Transform DC Period estimation and compares it against a threshold to output binary trend/cycle mode.
+
+**Post-warmup steady state (per bar):**
+
+| Operation | Count | Cost (cycles) | Subtotal |
+| :--- | :---: | :---: | :---: |
+| Hilbert FIR coefficients × 4 (InPhase, Quad) | 8 | 3 | 24 |
+| Phase accumulator update (ATAN2 equivalent) | 1 | 20 | 20 |
+| Period smoothing (EMA on period estimate) | 2 | 4 | 8 |
+| Trend period threshold comparison | 1 | 1 | 1 |
+| History buffer shifts × 4 | 4 | 1 | 4 |
+| **Total** | **16** | — | **~57 cycles** |
+
+The ATAN2-equivalent phase computation is the dominant cost. For default parameters: ~57 cycles per bar.
+
+### Batch Mode (SIMD Analysis)
+
+| Operation | Vectorizable? | Notes |
+| :--- | :---: | :--- |
+| Hilbert FIR (windowed taps) | Partial | Each tap independent; cross-bar state dependency limits |
+| Period EMA smoothing | **No** | Recursive IIR — sequential |
+| Threshold comparison | Yes | VCMPPD |
+
+The recursive EMA smoothing of the period estimate blocks full vectorization.
+
+### Quality Metrics
+
+| Metric | Score | Notes |
+| :--- | :---: | :--- |
+| **Accuracy** | 7/10 | Phase estimation inherent noise; binary output loses detail |
+| **Timeliness** | 6/10 | Hilbert requires ~32 bar warmup for phase stabilization |
+| **Smoothness** | 10/10 | Binary 0/1 output — maximally smooth |
+| **Noise Rejection** | 7/10 | EMA-smoothed period estimate reduces mode-flip chatter |
+
 ## Resources
 
 - Ehlers, J.F. — "The Instantaneous Trendline" (February 2002)