> However, the chart is flawed because it changes its calculation method mid-chart once it turns negative. The smaller positive numbers show how much the Fed was sending every week (around $2 billion on average) but when it turns negative, it switches to a cumulative calculation and thus falls off like a cliff.
i agree its crazy but i thought tfa did a good job explaining why the original chart ended up looking the way it did. regime change in a dataset can expose bad assumptions about relationships between balance sheet items you thought could be summed, etc.
Why on earth would anyone make a chart like that?