{
  "id": 286012,
  "title": "How to recreate the evaluation metric?",
  "url": "/competitions/g-research-crypto-forecasting/discussion/286012",
  "author_name": "",
  "post_date": "2021-11-07T14:56:08.094695700Z",
  "votes": 10,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hello, guys.</p>\n<p>Based on the overview page, we know that the evaluation metric is the correlation with weights.(<a href=\"https://en.wikipedia.org/wiki/Pearson_correlation_coefficient)\" target=\"_blank\">https://en.wikipedia.org/wiki/Pearson_correlation_coefficient)</a>. I found the definition of the weighted correlation in the provided wikipage. </p>\n<p><img src=\"https://wikimedia.org/api/rest_v1/media/math/render/svg/c0c9277e02db423861aa4ad28c3a66dbdfa8850d\" alt=\"Weighted Mean\"></p>\n<p><img src=\"https://wikimedia.org/api/rest_v1/media/math/render/svg/3dc4f106a16942bd22b5c8721a0a567816dce50d\" alt=\"Weighted Cov\"></p>\n<p><img src=\"https://wikimedia.org/api/rest_v1/media/math/render/svg/3d703455daf32e17a67de117ef93433125980424\" alt=\"Weighted Corr\"></p>\n<p>Does anyone successfully recreate the metric? Are we taking an global weighted average of the predicted residual returns? I thought that we first compute the correlation of each assets and output weighted correlation, but that seems to be incorrect. I will appreciate any help!</p>\n<p>Thank you!</p>",
  "messages": [
    {
      "id": "1574445",
      "postDate": "11/07/2021 14:56:08",
      "content": "<p>Hello, guys.</p>\n<p>Based on the overview page, we know that the evaluation metric is the correlation with weights.(<a href=\"https://en.wikipedia.org/wiki/Pearson_correlation_coefficient)\" target=\"_blank\">https://en.wikipedia.org/wiki/Pearson_correlation_coefficient)</a>. I found the definition of the weighted correlation in the provided wikipage. </p>\n<p><img src=\"https://wikimedia.org/api/rest_v1/media/math/render/svg/c0c9277e02db423861aa4ad28c3a66dbdfa8850d\" alt=\"Weighted Mean\"></p>\n<p><img src=\"https://wikimedia.org/api/rest_v1/media/math/render/svg/3dc4f106a16942bd22b5c8721a0a567816dce50d\" alt=\"Weighted Cov\"></p>\n<p><img src=\"https://wikimedia.org/api/rest_v1/media/math/render/svg/3d703455daf32e17a67de117ef93433125980424\" alt=\"Weighted Corr\"></p>\n<p>Does anyone successfully recreate the metric? Are we taking an global weighted average of the predicted residual returns? I thought that we first compute the correlation of each assets and output weighted correlation, but that seems to be incorrect. I will appreciate any help!</p>\n<p>Thank you!</p>",
      "rawMarkdown": "Hello, guys.\n\nBased on the overview page, we know that the evaluation metric is the correlation with weights.(https://en.wikipedia.org/wiki/Pearson_correlation_coefficient). I found the definition of the weighted correlation in the provided wikipage. \n\n![Weighted Mean](https://wikimedia.org/api/rest_v1/media/math/render/svg/c0c9277e02db423861aa4ad28c3a66dbdfa8850d)\n\n![Weighted Cov](https://wikimedia.org/api/rest_v1/media/math/render/svg/3dc4f106a16942bd22b5c8721a0a567816dce50d)\n\n\n![Weighted Corr](https://wikimedia.org/api/rest_v1/media/math/render/svg/3d703455daf32e17a67de117ef93433125980424)\n\nDoes anyone successfully recreate the metric? Are we taking an global weighted average of the predicted residual returns? I thought that we first compute the correlation of each assets and output weighted correlation, but that seems to be incorrect. I will appreciate any help!\n\nThank you!",
      "votes": null
    },
    {
      "id": "1647522",
      "postDate": "01/12/2022 17:06:43",
      "content": "<p>Have a look at these links:<br>\n    - <a href=\"https://www.kaggle.com/c/g-research-crypto-forecasting/overview/evaluation\" target=\"_blank\">https://www.kaggle.com/c/g-research-crypto-forecasting/overview/evaluation</a><br>\n    - <a href=\"https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/286778\" target=\"_blank\">https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/286778</a><br>\n    - <strong>Clarification:</strong> <strong><a href=\"https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/291845\" target=\"_blank\">https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/291845</a></strong><br>\n        - Notebook: <a href=\"https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition\" target=\"_blank\">https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition</a><br>\n        - API:  <a href=\"https://www.kaggle.com/sohier/detailed-api-introduction\" target=\"_blank\">https://www.kaggle.com/sohier/detailed-api-introduction</a></p>",
      "rawMarkdown": "Have a look at these links:\n    - https://www.kaggle.com/c/g-research-crypto-forecasting/overview/evaluation\n    - https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/286778\n    - **Clarification:** **https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/291845**\n        - Notebook: https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition\n        - API:  https://www.kaggle.com/sohier/detailed-api-introduction",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1647522,
      "author_name": "neomatrix369",
      "author_url": "",
      "post_date": "01/12/2022 17:06:43",
      "content": "<p>Have a look at these links:<br>\n    - <a href=\"https://www.kaggle.com/c/g-research-crypto-forecasting/overview/evaluation\" target=\"_blank\">https://www.kaggle.com/c/g-research-crypto-forecasting/overview/evaluation</a><br>\n    - <a href=\"https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/286778\" target=\"_blank\">https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/286778</a><br>\n    - <strong>Clarification:</strong> <strong><a href=\"https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/291845\" target=\"_blank\">https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/291845</a></strong><br>\n        - Notebook: <a href=\"https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition\" target=\"_blank\">https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition</a><br>\n        - API:  <a href=\"https://www.kaggle.com/sohier/detailed-api-introduction\" target=\"_blank\">https://www.kaggle.com/sohier/detailed-api-introduction</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1574445": "Hello, guys.\n\nBased on the overview page, we know that the evaluation metric is the correlation with weights.(https://en.wikipedia.org/wiki/Pearson_correlation_coefficient). I found the definition of the weighted correlation in the provided wikipage. \n\n![Weighted Mean](https://wikimedia.org/api/rest_v1/media/math/render/svg/c0c9277e02db423861aa4ad28c3a66dbdfa8850d)\n\n![Weighted Cov](https://wikimedia.org/api/rest_v1/media/math/render/svg/3dc4f106a16942bd22b5c8721a0a567816dce50d)\n\n\n![Weighted Corr](https://wikimedia.org/api/rest_v1/media/math/render/svg/3d703455daf32e17a67de117ef93433125980424)\n\nDoes anyone successfully recreate the metric? Are we taking an global weighted average of the predicted residual returns? I thought that we first compute the correlation of each assets and output weighted correlation, but that seems to be incorrect. I will appreciate any help!\n\nThank you!",
    "1647522": "Have a look at these links:\n    - https://www.kaggle.com/c/g-research-crypto-forecasting/overview/evaluation\n    - https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/286778\n    - **Clarification:** **https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/291845**\n        - Notebook: https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition\n        - API:  https://www.kaggle.com/sohier/detailed-api-introduction"
  },
  "source": "meta"
}