This issue became apparent while resolving #7. In financial statement forms there are certain fields that cannot take positive values — e.g. cost of goods sold, 2120, — and those that can be both positive and negative — e.g. profit (loss), 2110. Fields in the first group are enclosed in parentheses, which indicates negative values, and it is expected that in such a field a value is entered without the minus sign. However, it looks like occasionally financials statements still prepend a value with '-' prefix, leading to errors.
To harmonize, we had a choice of making all such values positive or negative, and in our original processing we made them positive. This seemed an intuitive way out since it implied that in order to calculate, say, gross profit one would subtract cost of goods sold from profit. But what we missed was that making everything positive is inconsistent with the existence of the signed variables, like profit (loss): we had a situation where negative values of signed variables were represented as negative numbers and negative values of negative-only variables -- as positive numbers.
This is resolved in #8 (as it was not possible to separate fixing adjustments procedures from this issue): now, all the negative-only variables' values are included in the dataset with minus sign.
This issue became apparent while resolving #7. In financial statement forms there are certain fields that cannot take positive values — e.g. cost of goods sold, 2120, — and those that can be both positive and negative — e.g. profit (loss), 2110. Fields in the first group are enclosed in parentheses, which indicates negative values, and it is expected that in such a field a value is entered without the minus sign. However, it looks like occasionally financials statements still prepend a value with '-' prefix, leading to errors.
To harmonize, we had a choice of making all such values positive or negative, and in our original processing we made them positive. This seemed an intuitive way out since it implied that in order to calculate, say, gross profit one would subtract cost of goods sold from profit. But what we missed was that making everything positive is inconsistent with the existence of the signed variables, like profit (loss): we had a situation where negative values of signed variables were represented as negative numbers and negative values of negative-only variables -- as positive numbers.
This is resolved in #8 (as it was not possible to separate fixing adjustments procedures from this issue): now, all the negative-only variables' values are included in the dataset with minus sign.