quantforge-miner / docs /Model77.md
DevWizard-Vandan
Execute Phase 13 updates: Purge ts_skewness and ts_kurtosis, define explicit safe list of operators, and prevent constant/no-trades expressions
1076f55
|
Raw
History Blame Contribute Delete
1.71 kB
# Analyst Revision Momentum (Conceptual β€” PV Proxies Only)
Analyst consensus upgrades predict future price drift. When a large number of analysts revise
estimates upward, institutions position ahead of the revision β€” creating volume and price
signatures that are detectable in Price-Volume data before the news is public.
**PV proxies for analyst revision dynamics:**
- **Unusual volume as leading indicator:**
Institutional accumulation ahead of upgrades produces above-average volume.
Proxy: `volume / adv20` β€” sustained elevation over 5–10 days precedes upgrades.
- **Price gap vs. VWAP:**
Institutional buying pushes the close above the VWAP, creating a positive spread.
Proxy: `(close - vwap) / (vwap + 1e-6)` β€” positive values indicate institutional demand.
- **Return median deviation as momentum proxy:**
Strong upward revision momentum shows up as mean-vs-median divergence.
Proxy: `ts_mean(returns, 5) - ts_median(returns, 20)` β€” positive = short-term drift.
- **Momentum persistence after analyst cluster:**
Stocks that have already drifted up on low volatility tend to continue.
Proxy: `ts_mean(returns, 21) / (ts_std_dev(returns, 21) + 1e-6)` β€” Sharpe ratio of returns.
**Expression building blocks (PV only β€” safe operators):**
```
rank((close - vwap) / (vwap + 1e-6)) # VWAP deviation rank
rank(ts_mean(volume / adv20, 10)) # sustained volume elevation
ts_mean(returns, 5) - ts_median(returns, 20) # mean vs median momentum
rank(ts_mean(returns, 21)) / (rank(ts_std_dev(returns, 21)) + 1e-6) # normalized drift
signed_power(rank((close - vwap) / (vwap + 1e-6)), 0.5) # nonlinear VWAP signal
```