A .m file would actually allow you to store native semantic data, but I can see why you'd want to use a SQL database. For the example you provided, it should be easy to use a few customized DatabaseLink commands to export the data into your schema. I suggested stringification because I didn't know if you wanted to store Wolfram Language-specific objects such as GeoObjects or entities, although you could actually use SQLExpr for that.

Here's an example:

dataset = 
 Dataset[{
    <|"date" -> DateObject["Aug 5 2015"],
    "portfolio" -> 1044,
    "percentMaturedEarly" -> "30"|>, 
    <|"date" -> DateObject["Aug 6 2015"],
    "portfolio" -> 2775,
    "percentMaturedEarly" -> "25"
|>}]

SQLInsert[myconnection, "MYTABLE", {"date", "portfolio", 
  "percentMaturedEarly"}, {SQLDateTime[
     DateList[#[["date"]]]], #[["portfolio"]], #[["percentMaturedEarly"]]} & /@ Normal[dataset]]

I don't really use SQL myself, but depending on the format of your data, you should be able to use DatabaseLink and SQLInsert. If you're making use of semantic data or entities, you could store them stringified and then use ToExpression to read them back in.