I visited a top floor of Goldman Sachs in the early 90s. They were using J, a dialect of APL. They still had those dumb little Bloomberg cube terminals on their desks, with the angry-fruit-salad colors scheme.

Nowadays it's in a window, but it looks the same. I don't know if they still use J, or K, but anyway not in real-time. That happens on FPGAs in a rack where traders once roamed.

But while APL cannot be said to be very popular, the most popular "time-series database" used in finance is published by APL fananics.

No they don't, GS strats have overwhelmingly used Slang/SecDB for many years now.

I have had no contact with GS for 25 years, but KDB is unaccountably popular in other places on the Street.

The specialized time-series database lost much of its reason to exist when disk seek times went to zero. Nowadays you stage onto an SSD, and spool to rotating storage later, if at all.

They vastly exaggerate their popularity, helped by the lack of publicly available information. The number of quants writing q is dwarfed by those writing Python, Scala or even Slang.

First Derivatives’ revenue last FY was ~200m and their pre-tax profit is ~17m. To say that they are unaccountably popular on the Street is quite an overstatement.

Last time I went interviewing, I found places who admitted using it. $200M revenue is nothing to sneeze at, for strictly optional infrastructural software.

I guess first derivatives just flushed 100 million bucks down the toilet then. Maybe you should send Conlon an email.

Or maybe people should learn not to extrapolate from anecdotes and outliers, as if what a single company does means anything for a huge sector...