Research
Occasional writing on markets, models, and the engineering behind systematic trading.
A single backtest is a memory test — it tells you a strategy could fit the past, not that it has an edge. The lab walks forward instead: optimize on what was known, test on what came next, roll, repeat.
The harness has a voice channel — a desk that talks to you, calling out only the events worth interrupting for, so you can step away from the screens and still stay in the loop.
We're training a model to learn a trader's reads — not to forecast prices, but to compress what an experienced trader notices into features the harness can watch across the whole board. A signal to the human, not a replacement for judgment.
What our agents actually do day to day, and the line they don't cross. They carry the surface area; a human keeps the wheel on anything that moves real capital.
The harness splits the work by nature: Python thinks, Rust acts, and one terminal keeps a person on the calls that matter. Each kind of work lives on the side where it belongs.
115 is a personalized trading harness: software that wraps one human trader in a full engineering stack. Here's what we're building, and why.
What the most secretive quantitative firm actually teaches — not its signals, but its posture: hire scientists, respect the data, and treat the market as a research problem.
A mechanical walk through everything that happens between wanting to trade and having traded — the order book, price-time priority, order types, why latency became an edge, and why slippage and impact are costs you cause rather than fees you pay.
A mechanism-first look at what standardization, novation, and daily mark-to-market actually buy you — and why margin, basis, and convergence are the plumbing behind every futures position.
How the US Treasury ladder is issued, why "risk-free" makes it the anchor everything else is priced against, and how the curve and the repo market turn that anchor into a signal and a plumbing system.
How an exchange born to trade butter and eggs became the place the world converts uncertainty into a price — and why the clearinghouse, not the contract, is the real invention.
1971 as the origin of the screen-based market, and what it means for price discovery that the negotiation moved onto competing quotes rather than a single obligated agent.
The oldest venue as a lens on what an auction market actually is, and why a single obligated agent and the closing print still matter in a world that clears electronically.
Before we built anything, we had to be clear about what a market is: a system of humans, machines, and incentives, priced in public and never quite in equilibrium. This is the worldview the rest of the work rests on.