[abalos]

It would be cool to start seeing more of this from other companies. Also, I wonder if anyone has done an extensive cost-benefit analysis of using OCaml instead of a more common language. It seems like the training and cost of always needing to build your own frameworks would be exceptionally high.

[rwmj]

Why "exceptionally" high? The programmers you are hiring are (or should be!) bright, they learn new things fast, and OCaml is not very difficult to learn.

At any software company you have to learn new things. Every company has its internal development stack and custom software, unless the company literally isn't a software company at all or is doing something so unoriginal and dull that you wouldn't want to work there.

[user5994461]

>>> and [any language] is not very difficult to learn.

I'm all for training and learning, but pretending that anything can be picked up quickly and easily is just wrong, especially in a trading environment.

Any typo or bug is going to cost millions of dollars, it's very hostile to newcomers. They will be struggling enough to learn about finance and the company.

[rwmj]

> Any typo or bug is going to cost millions of dollars

Good job they're using a language and other techniques which ensure that bugs are caught early, often at compile time.

[dmitriid]

yeah... no. only certain tupes of bugs.

a compiler can’t catch a bug if you calculate interest rates incorrectly, for example.

[frou_dh]

Minsky is on record as saying they have a maniacal bent for code review & code readability at Jane Street.

[ayushgp]

That needs to be written in unit tests and also needs to be code reviewed. Also such bugs can be brought into the system by experienced devs as well.

[steve_musk]

That is what unit tests are for.

[smt88]

Building your own libraries is often extremely costly and not cost effective.

Building your own frameworks usually saves some time at the beginning, but the costs usually outweigh the benefits later on.

Sorry to be pedantic, but the distinction between libraries and frameworks is important, in my opinion.

[gaius]

It seems like the training and cost of always needing to build your own frameworks would be exceptionally high

It is purely a matter of volume - are you doing enough business to make it worthwhile?

The last company I worked for had in-house database, programming language, IDE for that language, version control system, job scheduler... At the time they started doing that there was nothing available that did what they needed, and even now, their stack does things that I'm not sure I've seen outside, tho' 10 years on it's less clear that it would be worth starting a project like that from scratch now.

Remember, in business, there are no concepts of "cheap" or "expensive", there is only "worth it" or "not".