Ethan Lawinsky

I turn objectives into finished, tested solutions. Efficient by default — never at the cost of the standard.

see the tools

Trade Chart

Live interactive demo

Catalyst

I needed to replay futures markets tick by tick — fast enough to build real intuition for how price moves — with custom metrics computed in-step and a way to track simulated trading performance against them.

Result

A browser-based futures replay simulator and live monitor covering 15 CME contracts. It pulls 1-second data, volume-splices each contract into a continuous price series, and replays it at adjustable speed while computing rolling percentile-rank metrics (volume, range) in-step. You place simulated orders during playback with realistic slippage and per-contract sizing, so performance is measured under the conditions you'd actually face. A live mode streams the same metrics and threshold alerts on the current session in real time.

EDA Tool

Catalyst

I needed to test whether a strategy's adaptation process actually produces an edge over time — refitting filters across rolling periods — while structurally preventing the statistical traps that quietly ruin most strategy research: overfitting and lookahead bias.

Result

A browser-based exploratory analysis tool whose Procedure Mode runs rolling walk-forward validation. You commit a set of in-sample filters, which locks them into an immutable version; the tool steps forward through time, refits on each new training window, and reveals out-of-sample results only once — locked on view — so the holdout can never be quietly turned back into a training set. An optional auto-optimizer searches candidate filters on in-sample data alone and replays the winners on out-of-sample automatically, never peeking at the holdout. And when a version changes, the old one keeps running in parallel ("shadow tracking") to prove whether the change was a real gain or just noise-fitting. The discipline is enforced by construction, not willpower.

Armed Existence

Catalyst

Build a strategy game with real depth out of a deliberately small set of elements — depth that emerges from player choices rather than piled-on systems — structured around two interlocking core loops.

Result

A deterministic real-time strategy game in Godot 4, split into two handed-off loops: a strategic layer for building an economy, managing energy, and researching, and a tactical layer where the army you prepared fights. Complexity is emergent rather than authored — one fuel ladder, a single resource-flow graph, and a few building types interact to create deep decisions, while ~350 tunable values live in JSON so the whole game balances as data, not code. The simulation is deterministic by construction (fixed-timestep, seeded), so a fresh boot produces byte-identical state every run — multiplayer-ready without a rewrite. Stage 1 (economy and base-building) is built; the tactical combat layer is specified and in progress.

Kalshi Bot

Catalyst

On Kalshi's "mention markets" — where you wager on whether a specific word will be said during a live sports broadcast — the entire edge goes to whoever reacts first. I wanted to be that fastest mover.

Result

A Python tool that listens to a live broadcast on a laptop, transcribing audio in real time against a set of trigger words mapped to specific Kalshi mention markets. The instant a trigger word is spoken, it moves to fill "Yes" across as much of the corresponding market as it can — capturing the outsized return that goes to the earliest mover, before the rest of the market reprices.

Accomplishments

Objectives I've owned end to end — define the goal, figure out what it actually takes, and build the most efficient solution that meets it.

Law-Firm Client Intake & E-Signature Automation

Objective Let a law firm's reps collect client information and automatically generate, send, and track signature documents — at the lowest possible cost.

Solution Built an end-to-end intake-to-signature pipeline. A shared link (or rep-entered form) captures client data into a tracking sheet; scripts auto-populate the firm's PDFs, send them for e-signature, and feed every document's live status into a single dashboard the firm's controller manages.

Built with Google Forms & Sheets, Apps Script, Resend, SignatureAPI.

Value-Add Rental Acquisition & Repositioning

Objective Find, create, and capture value in a small multifamily property using conventional financing.

Solution Built a cash-on-cash return model for 2–4 unit buildings and screened ~30 properties against defined "deal" criteria. Identified one where converting a 3BR/1BA unit to 3BR/2BA unlocked value; acquired it, managed the renovation contractors, and built a tenant-vetting and lease system.

Result Placed tenants on 24-month leases and sold the stabilized, fully-leased property.

Contact

GitHub ↗