Tactical and Strategic

I've always thought that my relative skills advantage in my career was the fact that I was bilingual in English and nerd- that I was able to navigate between the MBAs and the quants- again relatively speaking, in such a way that it plays to my strength to be in that niche. Given that you need both in modern financial institutions, there is definitely a need to coordinate between the two such that they do not talk past each other.

There is another dimension I didn't really think about until now. And that is even within the quants (economists/statisticians/etc) there is a strategic/tactical divide. And it is not always easy to hop from one to the other. This is important because, again, there is a need for coordination between these two levels of aggregation.

Broadly speaking, the tactical side involve bottom-up statistical tools such as logistics regression and strategic side involves top-down tools such as time series regression. While the divide is fuzzy at the margin (a roll-rate estimation could be done top down or bottom up), the econometrics are sufficiently different between these two levels of aggregation that it is hard to develop expertise at both levels.

But I think there is more to that. You have to approach tactics and strategies with different mindsets. Running loan-level regression model by and large are operations stuff, and that involves a very narrow focus- you really have to get your hands dirty with the data. Conversely, when you run macro-level forecast models, your validation methodology is not really via things like hold-out samples, but that it conforms to varying sets of expectations. So the approach is basically that of breadth vs. depth.

Let's take the military analogy further. The tactical statistician would be like cavalry captains who have very deep knowledge of the battle terrains while the strategic economist would be like the generals in the tent plotting out troop movements. It helps if the generals have enough background as infantry so that they not march the troops into a ditch, but also helps if the cavalry captain have enough of a "big picture" view such that you know when to hold a hill at all costs or whether to beat a tactical retreat.

I think that with the increasing sophistication of econometric tools and the complexity of financial organizations, it becomes just that much harder to coordinate between the generals and captains.