Now the Fed has a really large backlog of these securities, and they are starting to unwind that by selling those securities off, effectively removing cash from the economy and slowing down economic activity (and therefore hopefully inflation).

https://www.federalreserve.gov/newsevents/pressreleases/mone...

The Federal Reserve has been working for the past few months agreeing on a plan to reduce the balance sheet. They are just saying that they are still continuing this work.

Outright sales of these securities are another option that they seem reluctant to do at this point.

Either way this has the effect of reducing the amount of money in the system (the opposite of QE where they created money out of thin air to buy the securities in the first place).

The Implementation Note says it only applies as principal is paid, so this is an upper bound.

According to the maturity distribution at https://www.federalreserve.gov/releases/h41/ the Fed has effectively all of its MBS with maturity > 1 year, with the vast majority over 10 years.

AFAIK we don't have any other maturity information that "over 10 years", but I might assume it's 20 years on average because it's uniformly distributed between 10-30 years (AFAIK 30 years is typically the max mortgage time in the US). Dividing $2.7 trillion by 240 months gives $11.25 billion per month.

Which implies an increase of the cap from $17.5B to $35B is a no-op, as rollover would already be maxed a little over $11B. So why bother?

Any bond / Fed experts able to shine a light on what's wrong with my numbers?

You can see it went up substantially in 2020 and 2021.