feat(028-strategy-signals-contrib): add 9 core strategies (momentum, earnings_momentum, value, volatility, multifactor, mean_reversion, moving_average, support_resistance, sector_rotation)
This commit is contained in:
Clayton Brown
parent
dde8fffa15
commit
1b1710b509
|
|
@ -0,0 +1,39 @@
|
|||
"""Shared data helpers for strategy modules."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import logging
|
||||
from typing import Any
|
||||
|
||||
import pandas as pd
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
def get_ohlcv(ticker: str, date: str, context: dict[str, Any] | None = None) -> pd.DataFrame | None:
|
||||
"""Return OHLCV DataFrame up to *date*, or None on failure.
|
||||
|
||||
Uses context["ohlcv"] if provided, otherwise fetches via load_ohlcv.
|
||||
"""
|
||||
if context and "ohlcv" in context:
|
||||
return context["ohlcv"]
|
||||
try:
|
||||
from tradingagents.dataflows.stockstats_utils import load_ohlcv
|
||||
df = load_ohlcv(ticker, date)
|
||||
return df if not df.empty else None
|
||||
except Exception:
|
||||
logger.debug("Failed to load OHLCV for %s@%s", ticker, date, exc_info=True)
|
||||
return None
|
||||
|
||||
|
||||
def get_info(ticker: str, context: dict[str, Any] | None = None) -> dict[str, Any] | None:
|
||||
"""Return yfinance .info dict, or None on failure."""
|
||||
if context and "info" in context:
|
||||
return context["info"]
|
||||
try:
|
||||
import yfinance as yf
|
||||
from tradingagents.dataflows.stockstats_utils import yf_retry
|
||||
return yf_retry(lambda: yf.Ticker(ticker.upper()).info) or None
|
||||
except Exception:
|
||||
logger.debug("Failed to load info for %s", ticker, exc_info=True)
|
||||
return None
|
||||
|
|
@ -0,0 +1,44 @@
|
|||
"""Earnings Momentum strategy signal (§3.2 — Earnings Momentum / SUE).
|
||||
|
||||
Computes Standardized Unexpected Earnings (SUE) from the most recent
|
||||
earnings surprise relative to trailing EPS standard deviation.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.2
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_info
|
||||
|
||||
|
||||
class EarningsMomentumStrategy(BaseStrategy):
|
||||
name = "Earnings Momentum (§3.2)"
|
||||
roles = ["fundamentals", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
info = get_info(ticker, context)
|
||||
if not info:
|
||||
return None
|
||||
|
||||
trailing_eps = info.get("trailingEps")
|
||||
forward_eps = info.get("forwardEps")
|
||||
if trailing_eps is None or forward_eps is None or trailing_eps == 0:
|
||||
return None
|
||||
|
||||
# SUE proxy: (forward - trailing) / |trailing|
|
||||
sue = (forward_eps - trailing_eps) / abs(trailing_eps)
|
||||
strength = max(-1.0, min(1.0, sue))
|
||||
direction = "bullish" if strength > 0.05 else ("bearish" if strength < -0.05 else "neutral")
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"SUE proxy (fwd-trail)/|trail|: {sue:+.2f} (trail={trailing_eps}, fwd={forward_eps})",
|
||||
)
|
||||
|
|
@ -0,0 +1,54 @@
|
|||
"""Mean Reversion strategy signal (§3.9 — Short-Term Reversal / Mean Reversion).
|
||||
|
||||
Z-score of current price vs rolling mean to detect overbought/oversold.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.9
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv
|
||||
|
||||
|
||||
class MeanReversionStrategy(BaseStrategy):
|
||||
name = "Mean Reversion (§3.9)"
|
||||
roles = ["market", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
df = get_ohlcv(ticker, date, context)
|
||||
if df is None or len(df) < 60:
|
||||
return None
|
||||
|
||||
close = df["Close"].values[-60:]
|
||||
mean = float(np.mean(close))
|
||||
std = float(np.std(close))
|
||||
if std == 0:
|
||||
return None
|
||||
|
||||
z = (close[-1] - mean) / std
|
||||
# Mean reversion: high z → bearish (expect revert down), low z → bullish
|
||||
strength = max(-1.0, min(1.0, -z / 3.0))
|
||||
if z > 1.5:
|
||||
direction = "bearish"
|
||||
label = "overbought"
|
||||
elif z < -1.5:
|
||||
direction = "bullish"
|
||||
label = "oversold"
|
||||
else:
|
||||
direction = "neutral"
|
||||
label = "fair"
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"Z-score: {z:+.2f} ({label}), 60d mean={mean:.2f}, price={close[-1]:.2f}",
|
||||
)
|
||||
|
|
@ -0,0 +1,43 @@
|
|||
"""Momentum strategy signal (§3.1 — Cross-Sectional Momentum).
|
||||
|
||||
Computes 12-1 month price momentum: cumulative return over months [-12, -1]
|
||||
skipping the most recent month to avoid short-term reversal.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.1
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
import pandas as pd
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv
|
||||
|
||||
|
||||
class MomentumStrategy(BaseStrategy):
|
||||
name = "Momentum (§3.1)"
|
||||
roles = ["market", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
df = get_ohlcv(ticker, date, context)
|
||||
if df is None or len(df) < 252:
|
||||
return None
|
||||
|
||||
close = df["Close"].values
|
||||
# 12-1 month momentum: return from 252 days ago to 21 days ago
|
||||
ret = (close[-21] - close[-252]) / close[-252]
|
||||
|
||||
strength = max(-1.0, min(1.0, ret)) # clamp
|
||||
direction = "bullish" if strength > 0.05 else ("bearish" if strength < -0.05 else "neutral")
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"12-1 month return: {ret:+.2%}",
|
||||
)
|
||||
|
|
@ -0,0 +1,55 @@
|
|||
"""Moving Average strategy signal (§3.11-3.13 — Moving Average Crossovers).
|
||||
|
||||
SMA crossover signals: 50/200 golden cross / death cross.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.11-3.13
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv
|
||||
|
||||
|
||||
class MovingAverageStrategy(BaseStrategy):
|
||||
name = "Moving Average (§3.11-3.13)"
|
||||
roles = ["market", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
df = get_ohlcv(ticker, date, context)
|
||||
if df is None or len(df) < 200:
|
||||
return None
|
||||
|
||||
close = df["Close"].values
|
||||
sma50 = float(np.mean(close[-50:]))
|
||||
sma200 = float(np.mean(close[-200:]))
|
||||
|
||||
if sma200 == 0:
|
||||
return None
|
||||
|
||||
spread = (sma50 - sma200) / sma200
|
||||
strength = max(-1.0, min(1.0, spread * 5))
|
||||
|
||||
if sma50 > sma200:
|
||||
direction = "bullish"
|
||||
label = "golden cross" if spread > 0.02 else "SMA50 > SMA200"
|
||||
elif sma50 < sma200:
|
||||
direction = "bearish"
|
||||
label = "death cross" if spread < -0.02 else "SMA50 < SMA200"
|
||||
else:
|
||||
direction = "neutral"
|
||||
label = "converged"
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"{label}: SMA50={sma50:.2f}, SMA200={sma200:.2f}, spread={spread:+.2%}",
|
||||
)
|
||||
|
|
@ -0,0 +1,73 @@
|
|||
"""Multifactor strategy signal (§3.6 — Multifactor Models).
|
||||
|
||||
Combined momentum + value + quality + low-vol composite.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.6
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv, get_info
|
||||
|
||||
|
||||
class MultifactorStrategy(BaseStrategy):
|
||||
name = "Multifactor (§3.6)"
|
||||
roles = ["researcher", "risk"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
df = get_ohlcv(ticker, date, context)
|
||||
info = get_info(ticker, context)
|
||||
if df is None or len(df) < 252 or not info:
|
||||
return None
|
||||
|
||||
factors: list[float] = []
|
||||
details: list[str] = []
|
||||
close = df["Close"].values
|
||||
|
||||
# Momentum factor: 12-1 month return
|
||||
if len(close) >= 252:
|
||||
mom = (close[-21] - close[-252]) / close[-252]
|
||||
factors.append(max(-1.0, min(1.0, mom)))
|
||||
details.append(f"mom={mom:+.2%}")
|
||||
|
||||
# Value factor: inverse PE
|
||||
pe = info.get("trailingPE")
|
||||
if pe and pe > 0:
|
||||
val = min(1.0 / pe / 0.15, 1.0) * 2 - 1
|
||||
factors.append(max(-1.0, min(1.0, val)))
|
||||
details.append(f"val_pe={pe:.1f}")
|
||||
|
||||
# Quality factor: ROE
|
||||
roe = info.get("returnOnEquity")
|
||||
if roe is not None:
|
||||
factors.append(max(-1.0, min(1.0, roe * 2)))
|
||||
details.append(f"roe={roe:.2%}")
|
||||
|
||||
# Low-vol factor
|
||||
if len(close) >= 63:
|
||||
vol = float(np.std(np.diff(np.log(close[-63:]))) * np.sqrt(252))
|
||||
lv = max(-1.0, min(1.0, (0.30 - vol) / 0.30))
|
||||
factors.append(lv)
|
||||
details.append(f"vol={vol:.1%}")
|
||||
|
||||
if not factors:
|
||||
return None
|
||||
|
||||
strength = round(sum(factors) / len(factors), 4)
|
||||
strength = max(-1.0, min(1.0, strength))
|
||||
direction = "bullish" if strength > 0.05 else ("bearish" if strength < -0.05 else "neutral")
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=strength,
|
||||
direction=direction,
|
||||
detail=f"{len(factors)}-factor composite: {', '.join(details)}",
|
||||
)
|
||||
|
|
@ -0,0 +1,72 @@
|
|||
"""Sector Rotation strategy signal (§4.1 — Sector Rotation).
|
||||
|
||||
Compares ticker's sector performance to broad market using relative strength.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §4.1
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import logging
|
||||
from typing import Any
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv, get_info
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# Sector ETF proxies
|
||||
_SECTOR_ETFS: dict[str, str] = {
|
||||
"Technology": "XLK",
|
||||
"Healthcare": "XLV",
|
||||
"Financial Services": "XLF",
|
||||
"Financials": "XLF",
|
||||
"Consumer Cyclical": "XLY",
|
||||
"Consumer Defensive": "XLP",
|
||||
"Energy": "XLE",
|
||||
"Industrials": "XLI",
|
||||
"Basic Materials": "XLB",
|
||||
"Utilities": "XLU",
|
||||
"Real Estate": "XLRE",
|
||||
"Communication Services": "XLC",
|
||||
}
|
||||
|
||||
|
||||
class SectorRotationStrategy(BaseStrategy):
|
||||
name = "Sector Rotation (§4.1)"
|
||||
roles = ["market", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
info = get_info(ticker, context)
|
||||
if not info:
|
||||
return None
|
||||
|
||||
sector = info.get("sector", "")
|
||||
etf = _SECTOR_ETFS.get(sector)
|
||||
if not etf:
|
||||
return None
|
||||
|
||||
sector_df = get_ohlcv(etf, date)
|
||||
spy_df = get_ohlcv("SPY", date)
|
||||
if sector_df is None or spy_df is None or len(sector_df) < 63 or len(spy_df) < 63:
|
||||
return None
|
||||
|
||||
# 3-month relative strength: sector ETF vs SPY
|
||||
sec_ret = (sector_df["Close"].values[-1] - sector_df["Close"].values[-63]) / sector_df["Close"].values[-63]
|
||||
spy_ret = (spy_df["Close"].values[-1] - spy_df["Close"].values[-63]) / spy_df["Close"].values[-63]
|
||||
rel = sec_ret - spy_ret
|
||||
|
||||
strength = max(-1.0, min(1.0, rel * 5))
|
||||
direction = "bullish" if strength > 0.1 else ("bearish" if strength < -0.1 else "neutral")
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"{sector} ({etf}) 63d relative strength vs SPY: {rel:+.2%}",
|
||||
)
|
||||
|
|
@ -0,0 +1,55 @@
|
|||
"""Support/Resistance strategy signal (§3.14 — Support and Resistance).
|
||||
|
||||
Identifies local min/max price levels and current proximity.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.14
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv
|
||||
|
||||
|
||||
class SupportResistanceStrategy(BaseStrategy):
|
||||
name = "Support/Resistance (§3.14)"
|
||||
roles = ["market", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
df = get_ohlcv(ticker, date, context)
|
||||
if df is None or len(df) < 60:
|
||||
return None
|
||||
|
||||
close = df["Close"].values[-60:]
|
||||
price = float(close[-1])
|
||||
high = float(np.max(close))
|
||||
low = float(np.min(close))
|
||||
rng = high - low
|
||||
if rng == 0:
|
||||
return None
|
||||
|
||||
# Position within range: 0 = at support, 1 = at resistance
|
||||
pos = (price - low) / rng
|
||||
|
||||
# Near resistance → bearish (expect pullback), near support → bullish
|
||||
strength = max(-1.0, min(1.0, (0.5 - pos) * 2))
|
||||
if pos > 0.85:
|
||||
direction, label = "bearish", "near resistance"
|
||||
elif pos < 0.15:
|
||||
direction, label = "bullish", "near support"
|
||||
else:
|
||||
direction, label = "neutral", "mid-range"
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"{label}: price={price:.2f}, support={low:.2f}, resistance={high:.2f}, range_pos={pos:.0%}",
|
||||
)
|
||||
|
|
@ -0,0 +1,62 @@
|
|||
"""Value strategy signal (§3.3 — Value).
|
||||
|
||||
Composite value score from Book/Market, Earnings/Price, and CashFlow/Price.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.3
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_info
|
||||
|
||||
|
||||
class ValueStrategy(BaseStrategy):
|
||||
name = "Value (§3.3)"
|
||||
roles = ["fundamentals", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
info = get_info(ticker, context)
|
||||
if not info:
|
||||
return None
|
||||
|
||||
scores: list[float] = []
|
||||
|
||||
# Book/Market (inverse of P/B)
|
||||
pb = info.get("priceToBook")
|
||||
if pb and pb > 0:
|
||||
bm = 1.0 / pb
|
||||
scores.append(min(bm, 3.0) / 3.0) # normalize: BM=3 → 1.0
|
||||
|
||||
# Earnings/Price (inverse of trailing PE)
|
||||
pe = info.get("trailingPE")
|
||||
if pe and pe > 0:
|
||||
ep = 1.0 / pe
|
||||
scores.append(min(ep, 0.15) / 0.15)
|
||||
|
||||
# Free Cash Flow yield proxy
|
||||
mcap = info.get("marketCap")
|
||||
fcf = info.get("freeCashflow")
|
||||
if mcap and fcf and mcap > 0:
|
||||
cfy = fcf / mcap
|
||||
scores.append(max(-1.0, min(cfy / 0.10, 1.0)))
|
||||
|
||||
if not scores:
|
||||
return None
|
||||
|
||||
composite = sum(scores) / len(scores)
|
||||
# Map [0,1] → [-1,1]: high value = bullish
|
||||
strength = max(-1.0, min(1.0, composite * 2 - 1))
|
||||
direction = "bullish" if strength > 0.1 else ("bearish" if strength < -0.1 else "neutral")
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"Composite value score: {composite:.2f} from {len(scores)} factors",
|
||||
)
|
||||
|
|
@ -0,0 +1,44 @@
|
|||
"""Volatility strategy signal (§3.4 — Volatility / Low-Vol Anomaly).
|
||||
|
||||
Computes realized volatility ranking and flags the low-volatility anomaly.
|
||||
|
||||
Reference:
|
||||
Kakushadze & Serur, "151 Trading Strategies", §3.4
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .base import BaseStrategy, StrategySignal
|
||||
from ._data import get_ohlcv
|
||||
|
||||
|
||||
class VolatilityStrategy(BaseStrategy):
|
||||
name = "Volatility (§3.4)"
|
||||
roles = ["risk", "market", "researcher"]
|
||||
|
||||
def compute(self, ticker: str, date: str, context: dict[str, Any] | None = None) -> StrategySignal | None:
|
||||
df = get_ohlcv(ticker, date, context)
|
||||
if df is None or len(df) < 63:
|
||||
return None
|
||||
|
||||
close = df["Close"].values[-63:]
|
||||
returns = np.diff(np.log(close))
|
||||
vol = float(np.std(returns) * np.sqrt(252))
|
||||
|
||||
# Low-vol anomaly: lower vol → mildly bullish signal
|
||||
# Map vol: 0.10→+0.5, 0.30→0, 0.60→-1.0
|
||||
strength = max(-1.0, min(1.0, (0.30 - vol) / 0.30))
|
||||
direction = "bullish" if strength > 0.1 else ("bearish" if strength < -0.1 else "neutral")
|
||||
|
||||
return StrategySignal(
|
||||
name=self.name,
|
||||
ticker=ticker,
|
||||
date=date,
|
||||
signal_strength=round(strength, 4),
|
||||
direction=direction,
|
||||
detail=f"Realized vol (63d annualized): {vol:.1%}, low-vol anomaly {'active' if vol < 0.25 else 'inactive'}",
|
||||
)
|
||||
Loading…
Reference in New Issue