{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "67813042",
   "metadata": {},
   "source": [
    "---\n",
    "title: Short-Term Reversal\n",
    "summary: Buy last week's losers, short last week's winners — the classic weekly mean-reversion effect.\n",
    "tags: [reversal, mean-reversion, short-term]\n",
    "---\n",
    "\n",
    "# Short-Term Reversal\n",
    "\n",
    "Over short horizons, stocks that just fell tend to bounce and stocks that just jumped tend to give\n",
    "some back — liquidity provision and overreaction. We trade the **weekly reversal** signal: long recent\n",
    "losers, short recent winners. It's a high-turnover effect, so it lives or dies on costs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "dde8ab74",
   "metadata": {},
   "outputs": [],
   "source": [
    "try:\n",
    "    import convexpi.lab  # noqa\n",
    "except ImportError:\n",
    "    import subprocess, sys\n",
    "    subprocess.run([sys.executable, \"-m\", \"pip\", \"install\", \"-q\", \"convexpi-lab\"])\n",
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "print(\"ready\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "89261e76",
   "metadata": {},
   "source": [
    "## The idea\n",
    "\n",
    "The `reversal_1w` feature is already sign-flipped so that **high values = recent losers**. So we just\n",
    "go long the top of that signal and short the bottom — buying the losers, shorting the winners."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "eb1f6051",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from convexpi.lab import Strategy\n",
    "\n",
    "def _long_short(signal, frac=0.2):\n",
    "    \"\"\"Dollar-neutral long/short: long the top `frac`, short the bottom `frac`, equal-weighted.\"\"\"\n",
    "    s = np.nan_to_num(np.asarray(signal, dtype=float))\n",
    "    n = len(s); k = max(1, int(n * frac))\n",
    "    order = np.argsort(s)\n",
    "    w = np.zeros(n); w[order[-k:]] = 1.0 / k; w[order[:k]] = -1.0 / k\n",
    "    return w\n",
    "\n",
    "class MyStrategy(Strategy):\n",
    "    \"\"\"Weekly reversal: long recent losers (high reversal_1w), short recent winners.\"\"\"\n",
    "    def on_day(self, day, features, prices, portfolio):\n",
    "        sig = features.get(\"reversal_1w\", np.zeros(len(prices)))\n",
    "        return _long_short(sig, frac=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd1e86e4",
   "metadata": {},
   "source": [
    "## Out-of-sample evaluation\n",
    "\n",
    "Train on the first half of a synthetic market, evaluate on the held-out second half — the same discipline as the leaderboard."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a2a649f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "from convexpi.lab import SyntheticMarket, Grader\n",
    "market = SyntheticMarket(n_stocks=80, n_days=1800, seed=1)\n",
    "report = Grader(market).evaluate(MyStrategy())\n",
    "print(f\"in-sample Sharpe    : {report.is_sharpe:+.2f}\")\n",
    "print(f\"out-of-sample Sharpe: {report.oos_sharpe:+.2f}\")\n",
    "print(f\"overfitting ratio   : {report.overfitting_ratio:+.2f}\")\n",
    "oos = report.oos_result.daily_returns\n",
    "fig, ax = plt.subplots(figsize=(8, 3))\n",
    "ax.plot(np.cumprod(1 + oos)); ax.set_title(\"Out-of-sample equity curve\"); ax.set_ylabel(\"growth of $1\")\n",
    "plt.tight_layout(); plt.show()\n",
    "\n",
    "print(\"Reversal is high-turnover: realistic transaction costs eat much of the gross edge.\")\n",
    "print(\"Try widening the quintile (frac) or trading less often to cut turnover.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f77668c8",
   "metadata": {},
   "source": [
    "## What I'd try next\n",
    "\n",
    "- **Costs dominate** here — estimate turnover and stress-test net-of-cost performance.\n",
    "- Trade only the **most liquid** names where the bounce is real, not a bid-ask artifact.\n",
    "- Blend with a slower signal (momentum) so the two horizons offset each other."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}