How fast do quantitative researchers code?

A lot of quants are less impressive than you’d think in their programming abilities. My experience is they code things that are good enough just very quickly but the ideas are usually extremely simple. So the answer is fast, but not godlike if that makes sense. 

It depends on the complexity of your models. Generalizing somewhat, slower strategies (and also those with less assets) tend to be simpler and require less coding (and signals can be as little as one line), whereas higher frequency/breath strategies could potentially get more complex.

It again highly varies, to actually code up a very simple model where you know exactly what needs to be done and how it needs to be done (this rarely happens) could take 10 minutes. But do you need to transform/clean the dataset you're working on? Run some summary stats and plot the data so you understand it? Understand its relationship to other variables? Try a couple of variations (or more) of similar ideas? I'd say in a more fast-paced environment, a particular signal can be worked on for several days up to a couple of weeks. This would be intermittently working, plus getting feedback from others, refining, etc. In a slower, less demanding environment you can potentially work on the same model for several months. It really depends, but I think those broad generalizations are reasonable.

In general, quant researchers need to be functional in coding- not great.

I’ve seen strategies go from idea to implementation in as little as two weeks or as long as a year. Sometimes developers are involved to transition code from research to an implementation state.