Claude_Memecoin_Analyst.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyOpUKSM/kq4ssDNKG30r4sq",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mshumer/gpt-investor/blob/main/Claude_Memecoin_Analyst.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Created as a fun experiment by [Matt Shumer](https://twitter.com/mattshumer_).\n",
        "\n",
        "Claude-Memecoin-Analyst is an educational/informational tool. It should not be considered as financial advice or a substitute for professional investment guidance."
      ],
      "metadata": {
        "id": "blJPzkbXbV7w"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Pick your coins here\n",
        "coins = [\"Dogecoin\", \"Shiba Inu\", \"Floki\"]\n",
        "\n",
        "# Put your API keys here\n",
        "ANTHROPIC_API_KEY = \"YOUR KEY HERE\"  # Replace with your Anthropic API key\n",
        "SERP_API_KEY = \"YOUR KEY HERE\"  # Replace with your SERP API key"
      ],
      "metadata": {
        "id": "_B95nsZeuMfC"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install newspaper3k"
      ],
      "metadata": {
        "id": "zafyWxLDsRFl"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import requests\n",
        "from bs4 import BeautifulSoup\n",
        "import newspaper\n",
        "from newspaper import Article\n",
        "from datetime import datetime\n",
        "import ast\n",
        "\n",
        "def get_search_results(search_term):\n",
        "    url = f\"https://serpapi.com/search.json?q={search_term}&api_key={SERP_API_KEY}\"\n",
        "    response = requests.get(url)\n",
        "    data = response.json()\n",
        "    return data['organic_results']\n",
        "\n",
        "def select_relevant_urls(search_results):\n",
        "    system_prompt = \"You are a crypto analyst assistant. From the given search results, select the URLs that seem most relevant and informative for analyzing the sentiment of the coin.\"\n",
        "    search_results_text = \"\\n\".join([f\"{i+1}. {result['link']}\" for i, result in enumerate(search_results)])\n",
        "    messages = [\n",
        "        {\"role\": \"user\", \"content\": f\"Search Results:\\n{search_results_text}\\n\\nPlease select the numbers of the URLs that seem most relevant and informative for analyzing the sentiment of the coin. Respond with the numbers in a Python-parseable list, separated by commas.\"},\n",
        "    ]\n",
        "    headers = {\n",
        "        \"x-api-key\": ANTHROPIC_API_KEY,\n",
        "        \"anthropic-version\": \"2023-06-01\",\n",
        "        \"content-type\": \"application/json\"\n",
        "    }\n",
        "    data = {\n",
        "        \"model\": 'claude-3-haiku-20240307',\n",
        "        \"max_tokens\": 200,\n",
        "        \"temperature\": 0.5,\n",
        "        \"system\": system_prompt,\n",
        "        \"messages\": messages,\n",
        "    }\n",
        "    response = requests.post(\"https://api.anthropic.com/v1/messages\", headers=headers, json=data)\n",
        "    response_text = response.json()['content'][0]['text']\n",
        "    numbers = ast.literal_eval(response_text)\n",
        "    relevant_indices = [int(num) - 1 for num in numbers]\n",
        "    relevant_urls = [search_results[i]['link'] for i in relevant_indices]\n",
        "    return relevant_urls\n",
        "\n",
        "def get_article_text(url):\n",
        "    article = Article(url)\n",
        "    article.download()\n",
        "    article.parse()\n",
        "    return article.text\n",
        "\n",
        "def get_sentiment_analysis(coin, article_texts):\n",
        "    system_prompt = \"You are a crypto sentiment analysis assistant. Analyze the sentiment of the given news articles for the coin and provide a summary of the overall sentiment. Be measured and discerning. You are a skeptical investor.\"\n",
        "    combined_text = \"\\n\\n\".join(article_texts)\n",
        "    messages = [\n",
        "        {\"role\": \"user\", \"content\": f\"Coin: {coin}\\n\\nNews Articles:\\n{combined_text}\\n\\nPlease analyze the sentiment of the news articles for the coin and provide a summary of the overall sentiment. Be measured and discerning. You are a skeptical investor.\"},\n",
        "    ]\n",
        "    headers = {\n",
        "        \"x-api-key\": ANTHROPIC_API_KEY,\n",
        "        \"anthropic-version\": \"2023-06-01\",\n",
        "        \"content-type\": \"application/json\"\n",
        "    }\n",
        "    data = {\n",
        "        \"model\": 'claude-3-haiku-20240307',\n",
        "        \"max_tokens\": 2000,\n",
        "        \"temperature\": 0.5,\n",
        "        \"system\": system_prompt,\n",
        "        \"messages\": messages,\n",
        "    }\n",
        "    response = requests.post(\"https://api.anthropic.com/v1/messages\", headers=headers, json=data)\n",
        "    sentiment_analysis = response.json()['content'][0]['text']\n",
        "    return sentiment_analysis\n",
        "\n",
        "def predict_price_movement(coin, sentiment_analysis):\n",
        "    system_prompt = \"You are a crypto price prediction assistant. Based on the sentiment analysis, predict whether the price of the coin will go up or down. Provide a brief explanation for your prediction.\"\n",
        "    messages = [\n",
        "        {\"role\": \"user\", \"content\": f\"Coin: {coin}\\n\\nSentiment Analysis:\\n{sentiment_analysis}\\n\\nBased on the sentiment analysis, predict whether the price of the coin will go up or down. Provide a brief explanation for your prediction.\"},\n",
        "    ]\n",
        "    headers = {\n",
        "        \"x-api-key\": ANTHROPIC_API_KEY,\n",
        "        \"anthropic-version\": \"2023-06-01\",\n",
        "        \"content-type\": \"application/json\"\n",
        "    }\n",
        "    data = {\n",
        "        \"model\": 'claude-3-haiku-20240307',\n",
        "        \"max_tokens\": 450,\n",
        "        \"temperature\": 0.5,\n",
        "        \"system\": system_prompt,\n",
        "        \"messages\": messages,\n",
        "    }\n",
        "    response = requests.post(\"https://api.anthropic.com/v1/messages\", headers=headers, json=data)\n",
        "    prediction = response.json()['content'][0]['text']\n",
        "    return prediction\n",
        "\n",
        "def rank_coins(coin_data):\n",
        "    system_prompt = \"You are a crypto investment advisor. Based on the sentiment analysis and price predictions for each coin, rank the coins from most attractive to least attractive for investment. Provide a brief rationale for your ranking.\"\n",
        "    coin_data_text = \"\\n\\n\".join([f\"Coin: {coin}\\nSentiment Analysis:\\n{data['sentiment']}\\nPrice Prediction:\\n{data['prediction']}\" for coin, data in coin_data.items()])\n",
        "    messages = [\n",
        "        {\"role\": \"user\", \"content\": f\"Coin Data:\\n{coin_data_text}\\n\\nBased on the sentiment analysis and price predictions for each coin, rank the coins from most attractive to least attractive for investment. Provide a brief rationale for your ranking.\"},\n",
        "    ]\n",
        "    headers = {\n",
        "        \"x-api-key\": ANTHROPIC_API_KEY,\n",
        "        \"anthropic-version\": \"2023-06-01\",\n",
        "        \"content-type\": \"application/json\"\n",
        "    }\n",
        "    data = {\n",
        "        \"model\": 'claude-3-haiku-20240307',\n",
        "        \"max_tokens\": 1000,\n",
        "        \"temperature\": 0.5,\n",
        "        \"system\": system_prompt,\n",
        "        \"messages\": messages,\n",
        "    }\n",
        "    response = requests.post(\"https://api.anthropic.com/v1/messages\", headers=headers, json=data)\n",
        "    ranking = response.json()['content'][0]['text']\n",
        "    return ranking\n",
        "\n",
        "def determine_investment(coin_data, ranking):\n",
        "    system_prompt = \"You are a crypto investment advisor. Based on the sentiment analysis, price predictions, and ranking of the coins, provide an overall determination of which coin(s) to invest in. Explain your reasoning.\"\n",
        "    coin_data_text = \"\\n\\n\".join([f\"Coin: {coin}\\nSentiment Analysis:\\n{data['sentiment']}\\nPrice Prediction:\\n{data['prediction']}\" for coin, data in coin_data.items()])\n",
        "    messages = [\n",
        "        {\"role\": \"user\", \"content\": f\"Coin Data:\\n{coin_data_text}\\n\\nRanking:\\n{ranking}\\n\\nBased on the sentiment analysis, price predictions, and ranking of the coins, provide an overall determination of which coin(s) to invest in. Explain your reasoning.\"},\n",
        "    ]\n",
        "    headers = {\n",
        "        \"x-api-key\": ANTHROPIC_API_KEY,\n",
        "        \"anthropic-version\": \"2023-06-01\",\n",
        "        \"content-type\": \"application/json\"\n",
        "    }\n",
        "    data = {\n",
        "        \"model\": 'claude-3-opus-20240229',\n",
        "        \"max_tokens\": 1000,\n",
        "        \"temperature\": 0.5,\n",
        "        \"system\": system_prompt,\n",
        "        \"messages\": messages,\n",
        "    }\n",
        "    response = requests.post(\"https://api.anthropic.com/v1/messages\", headers=headers, json=data)\n",
        "    investment_decision = response.json()['content'][0]['text']\n",
        "    return investment_decision\n",
        "\n",
        "# Get today's date\n",
        "today = datetime.now().strftime(\"%Y-%m-%d\")\n",
        "print('Analyzing the following coins:', coins)\n",
        "\n",
        "coin_data = {}\n",
        "\n",
        "# Perform analysis for each coin\n",
        "for coin in coins:\n",
        "    print(f\"\\nAnalyzing {coin} for {today}...\")\n",
        "\n",
        "    # Generate search query\n",
        "    search_query = f\"{coin} news {today}\"\n",
        "\n",
        "    # Get search results\n",
        "    search_results = get_search_results(search_query)\n",
        "\n",
        "    # Select relevant URLs\n",
        "    relevant_urls = select_relevant_urls(search_results)\n",
        "\n",
        "    # Get article text from relevant URLs\n",
        "    article_texts = []\n",
        "    for url in relevant_urls:\n",
        "        try:\n",
        "            text = get_article_text(url)\n",
        "            article_texts.append(text)\n",
        "        except:\n",
        "            pass\n",
        "\n",
        "    # Get sentiment analysis\n",
        "    sentiment_analysis = get_sentiment_analysis(coin, article_texts)\n",
        "    print(f\"\\nSentiment Analysis for {coin}:\")\n",
        "    print(sentiment_analysis)\n",
        "\n",
        "    # Predict price movement\n",
        "    prediction = predict_price_movement(coin, sentiment_analysis)\n",
        "    print(f\"\\nPrice Prediction for {coin}:\")\n",
        "    print(prediction)\n",
        "\n",
        "    coin_data[coin] = {\n",
        "        'sentiment': sentiment_analysis,\n",
        "        'prediction': prediction\n",
        "    }\n",
        "\n",
        "    print('\\n\\n')\n",
        "\n",
        "# Rank the coins\n",
        "ranking = rank_coins(coin_data)\n",
        "print(\"Ranking of Coins:\")\n",
        "print(ranking)\n",
        "\n",
        "# Determine which coin(s) to invest in\n",
        "investment_decision = determine_investment(coin_data, ranking)\n",
        "print(\"\\nInvestment Decision:\")\n",
        "print(investment_decision)"
      ],
      "metadata": {
        "id": "UXts518eXhus"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}