{
  "id": 1317,
  "title": "Some questions regarding dataset",
  "url": "/competitions/benchmark-bond-trade-price-challenge/discussion/1317",
  "author_name": "",
  "post_date": "2012-01-30T08:30:46.670Z",
  "votes": null,
  "comment_count": 2,
  "views": 2053,
  "content": "<p>Hello,</p>\r\n<p>I was trying to model the data, and have several questions:</p>\r\n<p>(1) curve_based_price: A fair price estimate based on implied hazard and funding curves of the issuer of the bond.</p>\r\n<p>Is this the estimated price of the bond at the time of maturity? Can you illustrate with an example?</p>\r\n<p>(2) We are given coupon rate and time to maturity for the trade price for only one trade (in a given row). I would imagine that the bond prices during the last 10 trades will also have been affected by these parameters. Will not having this data result in\r\n a model which may be good for competetion purposes, but not accurate in real life?</p>\r\n<p>(3) I would imagine that the data consists of bonds issued by several companies, and hence will exhibit different behavior for different values of coupons, maturity times, etc. There appears to be no direct way of grouping the data based on this particular\r\n fact. &nbsp;</p>\r\n<p>(4) Does &quot;customer buy&quot; mean that a non-dealer bought the bond from a dealer? What is the trade type if &quot;customer buys from customer (who sells)&quot;?&nbsp;</p>\r\n<p>(5) If past trading data is NA, is it the case that the data was taken at the time markets opened for trading? Missing data (how?)? Just curious to know what could be the reasons behind this.</p>\r\n<p>(6) What are the reasons behind &quot;reporting delays&quot;? For e.g. is it correlated to the type of trade, or maybe trade between dealers have smaller reporting delays?&nbsp;</p>\r\n<p>(7) Is this data taken from a week's worth of trade? Month? What is the sampling window?&nbsp;</p>",
  "messages": [
    {
      "id": "8345",
      "postDate": "01/30/2012 08:30:46",
      "content": "<p>Hello,</p>\r\n<p>I was trying to model the data, and have several questions:</p>\r\n<p>(1) curve_based_price: A fair price estimate based on implied hazard and funding curves of the issuer of the bond.</p>\r\n<p>Is this the estimated price of the bond at the time of maturity? Can you illustrate with an example?</p>\r\n<p>(2) We are given coupon rate and time to maturity for the trade price for only one trade (in a given row). I would imagine that the bond prices during the last 10 trades will also have been affected by these parameters. Will not having this data result in\r\n a model which may be good for competetion purposes, but not accurate in real life?</p>\r\n<p>(3) I would imagine that the data consists of bonds issued by several companies, and hence will exhibit different behavior for different values of coupons, maturity times, etc. There appears to be no direct way of grouping the data based on this particular\r\n fact. &nbsp;</p>\r\n<p>(4) Does &quot;customer buy&quot; mean that a non-dealer bought the bond from a dealer? What is the trade type if &quot;customer buys from customer (who sells)&quot;?&nbsp;</p>\r\n<p>(5) If past trading data is NA, is it the case that the data was taken at the time markets opened for trading? Missing data (how?)? Just curious to know what could be the reasons behind this.</p>\r\n<p>(6) What are the reasons behind &quot;reporting delays&quot;? For e.g. is it correlated to the type of trade, or maybe trade between dealers have smaller reporting delays?&nbsp;</p>\r\n<p>(7) Is this data taken from a week's worth of trade? Month? What is the sampling window?&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "8350",
      "postDate": "01/30/2012 18:56:42",
      "content": "<p>Questions #1, #5, #6 are identical to what I asked in:&nbsp;<a href=\"https://www.kaggle.com/c/benchmark-bond-trade-price-challenge/forums/t/1315/clarifications-of-received-time-diff-reporting-delay-and-curve-based-price\">&quot;Clarifications of received_time_diff,\r\n reporting_delay and curve_based_price&quot;</a></p>\r\n<p>&nbsp;</p>\r\n<p>#2: you *do*&nbsp;have the historical trade prices of the last 10 trades, that's <strong>\r\nt</strong><span><strong>rade_price_last{1-10}</strong></span></p>\r\n<p>#7 They're really trying to obscure the&nbsp;actual dates and date range involved. The spirit of\r\n<a href=\"http://www.kaggle.com/c/benchmark-bond-trade-price-challenge/details/Rules\">\r\nrules 3 and 9</a> seems to discourage trying to reverse-engineer that. Their thesis seems to be that this is not needed for prediction of trade_price. The admins can comment more.</p>\r\n<p>#3 Yes I've also been trying to group/cluster by bond type, just to understand which bond is which (not to break rule 3), and how\r\n<strong>current_coupon</strong> changes over time. I await the answer to that and #6, #7.</p>\r\n<p>Stephen</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "8374",
      "postDate": "01/31/2012 16:31:07",
      "content": "<p>1). Curve_based_price is the estimated price of the bond at the time of the trade. &nbsp;Please look in the data section or in responses to other questions for more detail.</p>\r\n<p>2). Time to maturity is from the time of the trade. &nbsp;It can be easilty calculated for each previous trade if you want. &nbsp;Coupon is normally a constant for a bond, so it will remain the same for each of the trades. &nbsp;Some bonds have coupons that change slightly,\r\n but there are few of them in the data set and the coupon changes infrequently enough that you can ignore this possiblity.</p>\r\n<p>3). Data does consist of different bonds issued by different companies. &nbsp;The aim of this competition was to deternine general characteristics of bond trades, not specific rules per bond or issuer.</p>\r\n<p>4). Customers cannot trade with each other in this market.</p>\r\n<p>5). Past trades as NaN happen when there are not enough previous trades in the time period that was sampled. &nbsp;No other data is padded in.</p>\r\n<p>6). This might have no correlation, but we decided to include it on the chance that it does. &nbsp;Reporting is normally required to be within 15 minutes.</p>\r\n<p>7). This data is taken from more than a month, but less than a year of data. &nbsp;It is a fraction of all of the trades in that period.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 8350,
      "author_name": "smcinerney",
      "author_url": "",
      "post_date": "01/30/2012 18:56:42",
      "content": "<p>Questions #1, #5, #6 are identical to what I asked in:&nbsp;<a href=\"https://www.kaggle.com/c/benchmark-bond-trade-price-challenge/forums/t/1315/clarifications-of-received-time-diff-reporting-delay-and-curve-based-price\">&quot;Clarifications of received_time_diff,\r\n reporting_delay and curve_based_price&quot;</a></p>\r\n<p>&nbsp;</p>\r\n<p>#2: you *do*&nbsp;have the historical trade prices of the last 10 trades, that's <strong>\r\nt</strong><span><strong>rade_price_last{1-10}</strong></span></p>\r\n<p>#7 They're really trying to obscure the&nbsp;actual dates and date range involved. The spirit of\r\n<a href=\"http://www.kaggle.com/c/benchmark-bond-trade-price-challenge/details/Rules\">\r\nrules 3 and 9</a> seems to discourage trying to reverse-engineer that. Their thesis seems to be that this is not needed for prediction of trade_price. The admins can comment more.</p>\r\n<p>#3 Yes I've also been trying to group/cluster by bond type, just to understand which bond is which (not to break rule 3), and how\r\n<strong>current_coupon</strong> changes over time. I await the answer to that and #6, #7.</p>\r\n<p>Stephen</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 8374,
      "author_name": "danglaser",
      "author_url": "",
      "post_date": "01/31/2012 16:31:07",
      "content": "<p>1). Curve_based_price is the estimated price of the bond at the time of the trade. &nbsp;Please look in the data section or in responses to other questions for more detail.</p>\r\n<p>2). Time to maturity is from the time of the trade. &nbsp;It can be easilty calculated for each previous trade if you want. &nbsp;Coupon is normally a constant for a bond, so it will remain the same for each of the trades. &nbsp;Some bonds have coupons that change slightly,\r\n but there are few of them in the data set and the coupon changes infrequently enough that you can ignore this possiblity.</p>\r\n<p>3). Data does consist of different bonds issued by different companies. &nbsp;The aim of this competition was to deternine general characteristics of bond trades, not specific rules per bond or issuer.</p>\r\n<p>4). Customers cannot trade with each other in this market.</p>\r\n<p>5). Past trades as NaN happen when there are not enough previous trades in the time period that was sampled. &nbsp;No other data is padded in.</p>\r\n<p>6). This might have no correlation, but we decided to include it on the chance that it does. &nbsp;Reporting is normally required to be within 15 minutes.</p>\r\n<p>7). This data is taken from more than a month, but less than a year of data. &nbsp;It is a fraction of all of the trades in that period.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "8345": "",
    "8350": "",
    "8374": ""
  },
  "source": "meta"
}