#!/usr/bin/env python3 """ **Expanded TSMOM / Managed Futures** — 24-asset universe. Same academic signal as ``run_tsmom_managed_futures.py`` (Moskowitz, Ooi, Pedersen 2012): 12-minus-1-month blended 3/6/12m signal → long/short → vol-normalized sizing. Extended universe (6 buckets, 24 ETF proxies): Equities (5): SPY, EFA, EEM, IWM, EWJ Bonds (4): TLT, IEF, SHY, EMB (EM debt) FX (5): UUP (USD), FXE (EUR), FXY (JPY), FXA (AUD), FXB (GBP) Commodities (4): GLD, SLV, DBC, USO Real assets (3): VNQ (US REIT), PDBC (commodity), CPER (copper) Rates/Inflation (3): TIP (inflation), FLOT (float-rate), HYG (HY bonds) Why this helps weak years: 2015: short EEM/commodity, long USD — all trended strongly 2019: long bonds (TLT +14%), short USD (faded) — misses if 8-asset book is short bonds 2022: short TLT/EEM/SPY, long UUP/CPER/USO — classic managed-futures year Example:: cd /Users/robzingale/trading_bot && PYTHONUNBUFFERED=1 \\ .venv/bin/python RenTech/strategy_stack/run_tsmom_expanded.py \\ --start 2011-01-03 --end 2025-12-31 --capital 100000 """ from __future__ import annotations import argparse import json import sys from pathlib import Path import numpy as np import pandas as pd import yfinance as yf _REPO = Path(__file__).resolve().parents[2] LOGS = _REPO / "RenTech" / "data" / "logs" UNIVERSE: dict[str, str] = { # Equities "SPY": "US large-cap equities", "EFA": "Intl developed equities", "EEM": "Emerging market equities", "IWM": "US small-cap equities", "EWJ": "Japan equities", # Bonds "TLT": "US 20yr Treasury", "IEF": "US 7-10yr Treasury", "SHY": "US 1-3yr Treasury", "EMB": "EM sovereign USD bonds", # FX (via ETFs — UUP is USD long; others are quoted vs USD so move opposite USD) "UUP": "US Dollar basket (DXY proxy)", "FXE": "Euro (EUR/USD proxy)", "FXY": "Japanese Yen (JPY/USD proxy)", "FXA": "Australian Dollar (AUD/USD proxy)", "FXB": "British Pound (GBP/USD proxy)", # Commodities "GLD": "Gold", "SLV": "Silver", "DBC": "Broad commodity index", "USO": "Crude oil", # Real assets / inflation "VNQ": "US REITs", "TIP": "US TIPS (inflation-linked Treasuries)", "HYG": "US High Yield bonds", "PDBC": "Diversified commodity (tax-efficient DBC)", "CPER": "Copper ETF", "IAU": "Gold (iShares, longer history backup)", } SIGNAL_LOOKBACKS = [ ( 63, 5), (126, 10), (252, 21), ] VOL_WINDOW = 63 TARGET_ASSET_VOL = 0.15 PORT_VOL_TARGET = 0.12 MAX_GROSS_LEVER = 3.0 def download_prices(tickers: list[str], start: str, end: str) -> pd.DataFrame: raw = yf.download(tickers, start=start, end=end, auto_adjust=True, progress=False) if isinstance(raw.columns, pd.MultiIndex): close = raw["Close"].copy() else: close = raw[["Close"]].copy() if "Close" in raw.columns else raw.copy() close.index = pd.to_datetime(close.index).tz_localize(None) close = close.sort_index().ffill(limit=5) # Drop tickers with < 2 years of history close = close.loc[:, close.notna().sum() >= 500] return close def run_tsmom_expanded( start: str, end: str, capital: float, *, out_prefix: Path, verbose: bool = True, ) -> dict: fetch_start = (pd.Timestamp(start) - pd.DateOffset(years=2)).strftime("%Y-%m-%d") fetch_end = end or pd.Timestamp.today().strftime("%Y-%m-%d") prices = download_prices(list(UNIVERSE.keys()), fetch_start, fetch_end) # IAU is backup for GLD — drop if GLD present if "GLD" in prices.columns and "IAU" in prices.columns: prices = prices.drop(columns=["IAU"]) # PDBC/CPER may have short history — keep if available, drop otherwise tickers = [t for t in prices.columns if t in UNIVERSE] prices = prices[tickers] rets = prices.pct_change() rebal_dates = prices.resample("BME").last().index max_lookback = max(lb for lb, _ in SIGNAL_LOOKBACKS) weights = pd.DataFrame(0.0, index=prices.index, columns=tickers) for rd in rebal_dates: if rd not in prices.index: prior = prices.index[prices.index <= rd] if not len(prior): continue rd = prior[-1] i_now = prices.index.get_loc(rd) if i_now < max_lookback + VOL_WINDOW: continue signal_votes = pd.Series(0.0, index=tickers) valid_lb = 0 for lb, skip in SIGNAL_LOOKBACKS: i_past, i_skip = i_now - lb, i_now - skip if i_past < 0 or i_skip < 0: continue p_now = prices.iloc[i_skip] p_past = prices.iloc[i_past] r_sig = (p_now / p_past.replace(0, np.nan) - 1.0).fillna(0.0) denom = r_sig.abs().mean() + 1e-8 signal_votes += r_sig / denom valid_lb += 1 if valid_lb == 0: continue signal_votes /= valid_lb direction = np.sign(signal_votes) vol_data = rets.iloc[max(0, i_now - VOL_WINDOW): i_now] realized_vol = vol_data.std(ddof=1) * np.sqrt(252) realized_vol = realized_vol.replace(0, np.nan).fillna(0.20) asset_weight = direction * (TARGET_ASSET_VOL / realized_vol) gross_vol_est = (asset_weight.abs() * realized_vol).sum() if gross_vol_est > 0: port_scale = PORT_VOL_TARGET / gross_vol_est gross = asset_weight.abs().sum() if gross > 0: port_scale = min(port_scale, MAX_GROSS_LEVER / gross) asset_weight *= port_scale asset_weight = asset_weight.fillna(0.0) next_rdates = rebal_dates[rebal_dates > rd] next_rd = next_rdates[0] if len(next_rdates) else prices.index[-1] mask = (prices.index > rd) & (prices.index <= next_rd) weights.loc[mask] = asset_weight.values weights = weights.shift(1).fillna(0.0) t0, t1 = pd.Timestamp(start), pd.Timestamp(fetch_end) mask = (prices.index >= t0) & (prices.index <= t1) w = weights.loc[mask] r = rets.loc[mask].fillna(0.0) portfolio_ret = (w * r).sum(axis=1) cap = float(capital) eq = cap * (1.0 + portfolio_ret).cumprod() n = len(portfolio_ret) years = n / 252.0 end_eq = float(eq.iloc[-1]) total_ret = end_eq / cap - 1.0 cagr = (end_eq / cap) ** (1.0 / years) - 1.0 if years > 0 else float("nan") dd = eq / eq.cummax() - 1.0 max_dd = float(dd.min()) sd = float(portfolio_ret.std(ddof=1)) sharpe = float(portfolio_ret.mean() / sd * np.sqrt(252)) if sd > 1e-12 else float("nan") vol_ann = sd * np.sqrt(252) spy_r = rets["SPY"].loc[mask].fillna(0.0) if "SPY" in rets.columns else pd.Series(0.0, index=r.index) rho_spy = float(pd.DataFrame({"p": portfolio_ret, "spy": spy_r}).dropna().corr().iloc[0, 1]) yr_rows = [] eq_cur = cap for yr, g in portfolio_ret.groupby(portfolio_ret.index.year): ret_y = float((1 + g).prod() - 1) * 100 end_y = eq_cur * (1 + ret_y / 100) yr_rows.append({"year": int(yr), "return_pct": round(ret_y, 2), "pnl_usd": round(end_y - eq_cur, 0)}) eq_cur = end_y yr_df = pd.DataFrame(yr_rows) # Asset-level attribution for weak years attribution: dict[str, dict] = {} for check_yr in [2015, 2016, 2018, 2019, 2022]: wy = w[w.index.year == check_yr] ry = r[r.index.year == check_yr] if wy.empty: continue attribution[str(check_yr)] = { t: round(float(((wy[t] * ry[t]) * cap).sum()), 0) for t in tickers if t in wy.columns } gross_long = (w.clip(lower=0) * cap).sum(axis=1) gross_short = ((-w).clip(lower=0) * cap).sum(axis=1) margin_usd = gross_long * 0.15 + gross_short * 0.30 daily_df = pd.DataFrame({ "date": portfolio_ret.index.strftime("%Y-%m-%d"), "daily_ret": portfolio_ret.values, "daily_pnl_usd": (portfolio_ret * cap).values, "equity_usd": eq.values, "gross_long_notional_usd": gross_long.values, "gross_short_notional_usd":gross_short.values, "margin_usd": margin_usd.values, }) out_prefix = Path(out_prefix) out_prefix.parent.mkdir(parents=True, exist_ok=True) daily_df.to_csv(f"{out_prefix}_daily.csv", index=False) yr_df.to_csv(f"{out_prefix}_yearly.csv", index=False) meta = { "strategy": "tsmom_expanded", "universe": tickers, "n_assets": len(tickers), "start": str(portfolio_ret.index.min().date()), "end": str(portfolio_ret.index.max().date()), "n_sessions": n, "capital": cap, "ending_equity_usd": round(end_eq, 2), "total_return_pct": round(total_ret * 100, 4), "cagr_pct": round(cagr * 100, 4), "sharpe": round(sharpe, 4), "max_dd_pct": round(max_dd * 100, 4), "vol_ann_pct": round(vol_ann * 100, 4), "rho_spy": round(rho_spy, 4), "weak_year_attribution": attribution, "yearly_csv": f"{out_prefix}_yearly.csv", "daily_csv": f"{out_prefix}_daily.csv", } with open(f"{out_prefix}_meta.json", "w") as fh: json.dump(meta, fh, indent=2) if verbose: print(f"=== TSMOM Expanded ({len(tickers)} assets) ===") print(f"Window: {meta['start']} → {meta['end']} ({n} sessions)") print(f"Return {total_ret*100:.1f}% CAGR {cagr*100:.1f}% Sharpe {sharpe:.2f} MaxDD {max_dd*100:.1f}% ρ(SPY) {rho_spy:.2f}") print("\nYearly returns:") print(yr_df.to_string(index=False)) print("\nAttribution in weak years:") for yr_str, attr in attribution.items(): top = sorted(attr.items(), key=lambda x: -abs(x[1]))[:5] top_str = " ".join(f"{t}:{v:+,.0f}" for t, v in top) yr_total = sum(attr.values()) print(f" {yr_str}: total ${yr_total:+,.0f} | {top_str}") return meta def main() -> None: ap = argparse.ArgumentParser(description=__doc__.split("\n\n")[0]) ap.add_argument("--start", default="2011-01-03") ap.add_argument("--end", default="2025-12-31") ap.add_argument("--capital", type=float, default=100_000.0) ap.add_argument("--out-prefix", type=Path, default=LOGS / "tsmom_expanded") args = ap.parse_args() run_tsmom_expanded( start=args.start, end=args.end, capital=float(args.capital), out_prefix=args.out_prefix, verbose=True, ) if __name__ == "__main__": main()