Merrill Lynch Email at 2012-2024 Form

I will preface this by saying that, for confidentiality purposes and other reasons, I will not name any specific banks, projects, or persons, whether or not they were mentioned in the question. Any similarity to banks, projects, or persons, living or dead, is purely coincidental.Several years ago, there was a major bank that undertook a project to replace a Java risk system (built in house in less than one year) with a Python risk system (which is still being built in house). Over a number of years, the cost of the Java system totaled less than 20 million British pounds, including hardware costs, software license + maintenance + support costs, developer costs, and operational costs (charge-back). The Java system is able to deliver on-demand results for ad hoc queries (e.g. n-way pivots on any of hundreds of variables) in 5 seconds or less against the real-time risk profile of the entire bank. Furthermore, due to the risk engine architecture, it can handle increased load and/or lower the response time of queries by scaling the hardware (either scaling up or out).In this particular bank, the system that was in place before the current Java system was apparently an overnight process that utilized hundreds or thousands of CPUs from a compute grid. As an overnight process, it provided results only from the previous day's positions, with no ad hoc support, (i.e. no custom pivots, etc.) It may have had some ability to perform intra-day updates, but the results were generally at least a few hours old, if not a day old. (My knowledge of the previous system is minimal, but from what I've seen at other banks, I assume that ad hoc queries were possible on-demand at the cost of a dedicated server and an hour or so of processing. Edit: I have been corrected, that the system did not support any ad hoc queries.)So far, the Python system -- which is not yet usable in production -- has an (estimated) cost much higher (at least an order of magnitude, from what I've heard Edit: supposedly, the entire project now totals over a billion pounds cumulatively) than the cumulative cost of the Java system, and the last time I heard about it, its result times were in the 90 second range. Also, to get it to perform that fast, the bank was forced to code a number of calculation routines in a lower level language (C, I believe), thus losing the potential algorithmic flexibility and dynamic behavior that could be achieved with Python (which was a major reason for selecting Python in the first place; see GS Slang, for example).As an aside, the person who was the driving force behind the Python system described above was coincidentally the same person who was responsible for (among other things) risk technology at an infamously no-longer-existing bank. (Edit: This person departed some time back when "the writing was on the wall.")Lastly, as Abdul mentioned, many of these projects are attempts to emulate the GS Sec/Slang environment. The irony is that as other banks attempt to move toward the GS model, GS is working to reduce their IT exposure that results from a completely proprietary (and rapidly aging) technology stack. (Edit: I was told that this shift began in 2007.)For the sake of full disclosure, I work at Oracle. The opinions and views expressed in this post are my own, and do not necessarily reflect the  opinions or views of my employer.

During the financial crisis, the Fed floated trillions in liquidity to the financial services sector to avoid wholesale failures. Bear was a small firm with no deep pocket global bank behind it. When their inventory losses exceeded their capital they were insolvent and could have been seized by regulators. This would have been a disaster as Bear held vast numbers of customer accounts commingled with their own property as with any margin/securities lending account.The fed and Treasury wanted Bear connected to a huge bank balance sheet so they could continue to float liquidity until markets recovered, avoiding a wholesale liquidation of bear. The company was worthless. $2 a share was a gift. Morgan itself would have failed without the Fed’s ocean of liquidity and easing of regulatory enforcement. I’m sure Dimon didn’t really want bear at the time. It turned out well after years of ZIRP and QE bonanzas.Merrill was a bigger version of the same thing. It could not have survived on its own and a failure would have led to millions of customers finding their stocks and bonds tied up in limbo as general creditors. By putting it into BofA, not a paragon by any means but huge, it bought time.Banks can also put huge amounts of inventory into a held for investment account and not have to mark them to market, avoiding large capital losses. This likely played some role if not a major role.In short, this was the Fed and Treasury using legerdemain to avoid chaos. It was far from the best or right thing to do, but something like this was inevitable at the time.

U.S. politicians demanded Bank of America purchase Merrill Lynch as a bailout. https://sevenpillarsinstitute.or...

That is highly doubtful. The banking giant is known for rescuing risky or folding investments. It’s the name of the game. Most acquisitions lose their name in the transitions under new umbrellas but ML kept it’s distinguished name as, and this is only an opinion, had an untarnished reputation in the finance arena.