Benchmarks Should Represent Your Workload: COREX v2
This is your reminder that a benchmark is pointless if it does not represent your own workload. Period.
Some of the big chip manufacturers are making lots of noise about how great their silicon is on various benchmarks and attempting to bludgeon you into belief with chart crime and long words. Fortunately, we have some actual brainiacs calling them out on it too (not me).
So, this is just your friendly reminder that benchmarks are meaningless unless they genuinely reflect your own workload. It really is as simple as that.
Which brings me to COREx, our little financial risk analytics benchmark. The same absolutely applies to this too. Which is why we’re working on version 2.
Financial risk analytics is not a homogenous blob. When I first created COREx, the next best thing was running the unit tests for QuantLib and this was certainly an improvement.
But we can do better.
Specifically, I feel the difference between real time risk and end of day risk needs to be better captured as does the difference between market risk and portfolio risk.
That’s just me though, so before we run headlong into this, I’d love to get some input from you too.
