Speed Is Alpha: Why Research Velocity Quietly Compounds Into an Edge

Take two quants. Same talent, same data, same access to the same markets, the same quality of ideas. The only difference between them is mundane: one can test a hypothesis and see honest results in two minutes; the other waits twenty, fighting with tooling, re-fetching data, babysitting jobs. Run that difference forward six months, and the two are not slightly apart. They're in different leagues.

This is one of the most underappreciated truths in systematic trading: research velocity is a compounding edge. Not a nice-to-have, not a productivity tip — an actual, durable competitive advantage that determines who finds robust strategies and who's perpetually one backtest behind. Speed, it turns out, is its own kind of alpha.

Edges are found by iteration, not insight

There's a romantic myth that great strategies come from a flash of insight — the brilliant idea that just works. Reality is far more iterative. A real edge is found by having an idea, testing it, watching it mostly fail, understanding why it failed, adjusting, testing again, and repeating until something robust emerges from the wreckage of dozens of attempts.

The raw material of research isn't ideas — most quants have more ideas than they can test. The raw material is iterations: complete loops of hypothesize, test, learn, adjust. And the number of iterations you can run is set almost entirely by how long a single loop takes. Cut the loop time in half and you double your iterations. Double your iterations and you find robust edges roughly twice as fast — or find ones the slower quant never reaches at all, because they ran out of patience three attempts before the breakthrough.

What slow research actually costs

The cost of slow tooling isn't just the wasted minutes, though those add up brutally. The deeper cost is what it does to your thinking.

When testing an idea is fast, you stay in flow. You have a thought, you test it, the result is back before the thought has left your head, and the result immediately suggests the next thought. The research is a fluid conversation between you and the market. When testing is slow, that conversation shatters. You have a thought, you kick off a job, you context-switch to something else because you can't sit and wait, and twenty minutes later you come back to a result you've half-forgotten the reasoning for. The flow is gone, and with it the compounding chain of "this result suggests that test suggests this refinement" that is where real discovery happens.

Slow tooling doesn't just make you do research more slowly. It makes you do worse research, because it taxes the very iterative momentum that good research depends on. And worst of all, it makes you stop early — you run three variations instead of thirty, you grab the first thing that looks decent, because the friction of continuing is too high. The robust edge that was four iterations away never gets found.

Where the time actually hides

The slow parts of research are rarely the interesting parts. Nobody's iteration loop is bottlenecked on thinking. It's bottlenecked on plumbing: re-fetching the same data for the hundredth time, waiting on a job queue, fighting an environment, manually wiring up a parameter test, re-running everything from scratch because the last result wasn't cached.

This is the good news, because plumbing is solvable. When the data you tested yesterday is cached and reused instead of re-fetched, when a sweep of a thousand parameter combinations costs little more than a single backtest, when an idea can go from typed to tested in the time it takes to read the result — the loop collapses from twenty minutes to two. And that collapse isn't a 10x convenience. It's a 10x in the one quantity that compounds into edge: iterations per unit of your finite attention.

The compounding, made concrete

Play it forward. The fast quant runs, say, thirty meaningful tests a day. The slow one runs three. Over a quarter, that's the difference between roughly two thousand iterations and two hundred. The fast quant has explored ten times more of the strategy space, killed ten times more bad ideas, and refined ten times more promising ones. They haven't just found edges faster — they've found edges the slow quant will never find, because those edges lived in the parts of the space the slow quant never had the throughput to reach.

This is why two equally talented people end up in different leagues. It was never about talent. It was about how many times each of them got to spin the loop, and the loop time was set by the tools.

Speed as a strategic choice

The lesson for a fund is to treat research velocity as a first-class strategic concern, not an afterthought. The instinct is to optimize for the quality of any single backtest — and quality matters enormously, as anyone who's been burned by a dishonest one knows. But quality without speed is a sports car in traffic. The real target is honest results, fast: backtests you can trust, produced quickly enough that you can run hundreds of them and stay in flow while you do.

When you have that, research stops being a series of scheduled batch jobs and becomes a fluid, compounding conversation with the market. You test more, learn more, and find the robust edges that only reveal themselves to whoever can afford to keep looking. In a competition where everyone has similar talent and similar data, the one who can iterate fastest quietly pulls ahead — and stays ahead.

Speed isn't the opposite of rigor. Done right, it's the multiplier on it. And over a long enough horizon, the multiplier is the edge.

---

Test an idea in minutes, sweep a thousand parameters for the cost of one, and stay in flow — research velocity, by design. Start researching →