Correct portfolio quality filters. Update portfolio analysis.

README.md

@@ -112,21 +112,49 @@ The project uses a trading date observation window, which sets 4 portfolio rebal
 
 First, update the stock database:
    ```bash
-   stocksense --update
+   stocksense update
    ```
 
 ### Model Training
 
-Train the model for a given trade date:
+Train the model for a specific trade date:
    ```bash
-   stocksense --train --trade-date YYYY-MM-DD
+   stocksense train --trade-date YYYY-MM-DD
    ```
+   For example:
+   ```bash
+   stocksense train --trade-date 2024-03-01
+   ```
+   Note: Trade date must be the 1st of March, June, September, or December.
+
+
+Generate an investment portfolio:
+   ```bash
+   stocksense portfolio --trade-date YYYY-MM-DD [--weighting STRATEGY] [--n-stocks N]
+   ```
+
+   Options:
+   - `--weighting`: Portfolio weighting strategy (`market_cap` or `equal`). Default: `market_cap`
+   - `--n-stocks`: Number of stocks to include in portfolio. Default: 30
 
-Score stocks for a given trade date:
+   Example:
    ```bash
-   stocksense --score --trade-date YYYY-MM-DD
+   # Generate a market-cap weighted portfolio with 25 stocks
+   stocksense portfolio --trade-date 2024-03-01 --weighting market_cap --n-stocks 25
+
+   # Generate an equal-weighted portfolio with default 30 stocks
+   stocksense portfolio --trade-date 2024-03-01 --weighting equal
    ```
 
+#### Command Reference
+
+```bash
+stocksense --help              # Show available commands
+stocksense update --help       # Show help for update command
+stocksense train --help        # Show help for train command
+stocksense portfolio --help    # Show help for portfolio command
+```
+
 In order to evaluate for the last trading date, don't specify a trade date.
 
 ### Streamlit App

notebooks/portfolio_analysis.ipynb

@@ -0,0 +1,78 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from pathlib import Path\n",
+    "\n",
+    "import polars as pl\n",
+    "\n",
+    "DATE = \"2023-12-01\"\n",
+    "REPORT_DIR = Path(\"../reports/scores\")\n",
+    "PORTFOLIO_DIR = Path(\"../reports/portfolios\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Portfolio: ../reports/portfolios/portfolio_2023-06-01.xlsx\n",
+      "Average return: 32.73% (80.0)\n",
+      "\n",
+      "Portfolio: ../reports/portfolios/portfolio_2023-03-01.xlsx\n",
+      "Average return: 59.51% (100.0)\n",
+      "\n",
+      "Portfolio: ../reports/portfolios/portfolio_2023-12-01.xlsx\n",
+      "Average return: 21.08% (80.0)\n"
+     ]
+    }
+   ],
+   "source": [
+    "for portfolio in PORTFOLIO_DIR.glob(\"portfolio_*.xlsx\"):\n",
+    "    df = pl.read_excel(portfolio, sheet_name=\"Full Portfolio\")\n",
+    "    top = df.head(5)\n",
+    "\n",
+    "    if \"fwd_return_4Q\" not in df.columns:\n",
+    "        continue\n",
+    "\n",
+    "    top_freturn = top.select(pl.col(\"fwd_return_4Q\")).mean().item()\n",
+    "    top_hits = top.select(pl.col(\"fwd_return_4Q\") > 0).sum().item()\n",
+    "\n",
+    "    top_hitrate = (top_hits / len(top)) * 100\n",
+    "\n",
+    "    print(f\"\\nPortfolio: {portfolio}\")\n",
+    "    print(f\"Average return: {top_freturn:.2f}% ({top_hitrate:.1f})\")"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.0"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

stocksense/model/portfolio.py

@@ -98,10 +98,9 @@ def _filter_candidates(self, df: pl.DataFrame) -> pl.DataFrame:
 
         quality_filters = (
             (pl.col("pe") > 0)
-            & (pl.col("ev_ebitda") < 50)
             & (pl.col("saleq_yoy") > -20)
             & (pl.col("fcf_yoy") > -50)
-            & (pl.col("price_mom") > -20)
+            & (pl.col("price_mom") > -25)
         )
         return df.filter(quality_filters)