Closed ijyliu closed 4 months ago
Probalby not goint to predict changes
may do changes
could do differences and levels of variables
I believe this is actually only necessary for Altman Z and any financial ratios
Specifically,
['Altman_Z', 'Ratio_A', 'Ratio_B', 'Ratio_C', 'Ratio_D', 'Ratio_E', 'grossProfitRatio', 'ebitdaratio', 'operatingIncomeRatio', 'incomeBeforeTaxRatio', 'netIncomeRatio']
Reasoning: for things like total assets, debt, etc. the cash flow and other variables already capture quarterly changes
These can be added to all the models, not just the changes model
added code to difference to file Create Combined All Data - Fixed Quarter Dates.ipynb
see section "Changes in Financial Variables"
integrated into dataset and variable index
If we are going to predict upgrades/downgrades in credit rating, it seems likely that the changes in financial statement variables will be more useful than their levels.
Each earning call would be accompanied by data representing the change in variables from the prior quarter.
If time allows, other lags (1 year back in particular to remove seasonality) might be useful as well.