{
  "id": 587226,
  "title": "Submission Evaluation - Other techniques can be used than Pearson Correlation",
  "url": "/competitions/drw-crypto-market-prediction/discussion/587226",
  "author_name": "",
  "post_date": "2025-06-30T08:17:53.213682300Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi</p>\n<p>Can't we use other problem evaluation techniques as we do seem to have high complex, non-linear relationship between our final predictors and label have a sound RSME - 0.33 but still the pearson correlation is really low - 0.006 indicating that we have high influence of final predictors on target label though not direct linear relationship?</p>\n<p>Thanks</p>",
  "messages": [
    {
      "id": "3236289",
      "postDate": "06/30/2025 08:17:53",
      "content": "<p>Hi</p>\n<p>Can't we use other problem evaluation techniques as we do seem to have high complex, non-linear relationship between our final predictors and label have a sound RSME - 0.33 but still the pearson correlation is really low - 0.006 indicating that we have high influence of final predictors on target label though not direct linear relationship?</p>\n<p>Thanks</p>",
      "rawMarkdown": "Hi\n\nCan't we use other problem evaluation techniques as we do seem to have high complex, non-linear relationship between our final predictors and label have a sound RSME - 0.33 but still the pearson correlation is really low - 0.006 indicating that we have high influence of final predictors on target label though not direct linear relationship?\n\nThanks",
      "votes": null
    },
    {
      "id": "3237175",
      "postDate": "07/01/2025 00:37:02",
      "content": "<p>Just my 2 cents, but - Pearson is popular because many trading strategies simply scale positions up or down based on signal strength - so it directly shows how well your predictions will work in practice! Plus, linear relationships are much easier to manage risk-wise and fit nicely into other, sometimes legacy, portfolio tools.</p>",
      "rawMarkdown": "Just my 2 cents, but - Pearson is popular because many trading strategies simply scale positions up or down based on signal strength - so it directly shows how well your predictions will work in practice! Plus, linear relationships are much easier to manage risk-wise and fit nicely into other, sometimes legacy, portfolio tools.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3237175,
      "author_name": "taylorsamarel",
      "author_url": "",
      "post_date": "07/01/2025 00:37:02",
      "content": "<p>Just my 2 cents, but - Pearson is popular because many trading strategies simply scale positions up or down based on signal strength - so it directly shows how well your predictions will work in practice! Plus, linear relationships are much easier to manage risk-wise and fit nicely into other, sometimes legacy, portfolio tools.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3236289": "Hi\n\nCan't we use other problem evaluation techniques as we do seem to have high complex, non-linear relationship between our final predictors and label have a sound RSME - 0.33 but still the pearson correlation is really low - 0.006 indicating that we have high influence of final predictors on target label though not direct linear relationship?\n\nThanks",
    "3237175": "Just my 2 cents, but - Pearson is popular because many trading strategies simply scale positions up or down based on signal strength - so it directly shows how well your predictions will work in practice! Plus, linear relationships are much easier to manage risk-wise and fit nicely into other, sometimes legacy, portfolio tools."
  },
  "source": "meta"
}