Creating vintage loss curves to gauge credit losses on a cohort level. $-$ For a given month, x is the amount of total loans withdrawn. Y is the total of these loans which goes into losses after $4$ months$(90+$ DPD$).$Assuming the loans are given for a tenure of $3-36$ months. There will be some loans going into losses every subsequent month for this cohort. The way to identify this is checking what loans are $90+$ DPD for a particular month from a given disbursal cohort. You will need to find a way to plot this in a waterfall chart.