Inspiration
I've always been drawn to gold — not as jewelry, but as an asset that holds its value while inflation quietly eats everything else. That conviction led me down a rabbit hole: how does gold relate to oil, the other great commodity that the modern economy runs on? The insight that stopped me cold was that gold and oil usually move together — but in recessions they split apart, with oil collapsing while gold climbs to safety. That divergence is a readable signal about where the economy is heading, and almost nobody trades it systematically. Petrov Capital was born from wanting to turn that one honest insight into something real.
What it does
Petrov Capital is an AI that trades the gold–oil divergence. It continuously reads the gold-to-oil ratio along with inflation and rate data, detects when the two assets break apart, and rotates accordingly — leaning into safe-haven gold when the economy signals trouble, and into oil when it signals growth. Crucially, it explains every move in plain English ("markets are nervous — leaning gold"), so it's not a black box. It's the discipline of a macro fund, running on autopilot.
How we built it
We designed Petrov as a simple four-stage pipeline using proven, available tools. A data layer pulls live gold and oil prices plus macro feeds from public market APIs. A signal engine computes the gold-oil ratio and flags divergence — the moment they break apart. An AI layer uses an LLM to classify the regime (risk-on vs. risk-off) and translate it into plain-English reasoning. Finally, a decision and output layer produces the rotate call and surfaces it on a dashboard with the live signal and its rationale. No new infrastructure — just wiring the right pieces into one disciplined loop.
Challenges we ran into
The hardest part was resisting the urge to overcomplicate. It would have been easy to chase a dozen indicators and build something impressive but unfocused — instead we had to discipline ourselves down to one signal we could actually explain and defend. We also had to be honest about the data: validating whether the gold-oil divergence holds historically (it does — they're positively correlated 80%+ of the time, but split in downturns) meant confronting our own thesis against real numbers rather than just believing it. And as a small team, scoping the idea so it was genuinely buildable, not just a pitch, took real restraint.
Accomplishments that we're proud of
We turned a personal hunch about gold into a coherent, defensible strategy with a clear thesis, a real mechanism, and an honest plan to build it. We're proud that the idea is ours — it came from genuine conviction, not a trend we copied. We're proud that it's narrow enough to actually ship, yet has a real path to grow into a full autonomous fund. And we're proud that we kept it honest: our backtest is clearly labeled as illustrative, and the whole design favors transparency over hype.
What we learned
We learned that leverage doesn't come from a cleverer idea — it comes from knowing something true that others overlook, and then building simply on top of it. We learned the value of narrowing: a small, well-understood problem beats a big vague one every time. We learned how gold and oil actually relate across economic cycles, and how to test a conviction against real data instead of just trusting it. Most of all, we learned that thoughtful execution and clear communication can carry an idea further than raw ambition.
What's next for Petrov Capital
Next, we build the real backtest — pulling actual historical gold and oil data and running the divergence strategy across past cycles to replace our illustrative numbers with verified results. From there, we'd add live data feeds and a working dashboard so the signal updates in real time. The long-term vision is to grow Petrov from a single transparent signal that retail investors can trust into the autonomous engine of a modern macro fund — one that reads the cycle and trades the divergence, end to end.
Log in or sign up for Devpost to join the conversation.