Let's assume I found the holy grail of low-latency trading (which I didn't). For educational purposes, what would be the first strategy I would direct my trading code?
Being very fast within a single datacenter is not as valuable as having the fastest line between two datacenters. So being able to write a very fast program wouldn't be the holy grail of trading anyway (it would be to instantaneously transport information between e.g. NJ and Chicago using quantum entanglement or something.)
That said, if you found an imaginary way to write a program that has zero latency, I would point it toward capturing arbitrage opportunities available within a single datacenter. One example is that on exchanges that list both futures and options on those futures, pure arbitrage opportunities can be available to whoever is fastest to capture them (i.e. violations of put-call parity and variations on that).