Research
Papers with figures built from real data.
What shipped in Terminal, Finance, Labs and ArcDelta.
18 entries→02ArcDelta history, venues, latency and completeness.
live→03Returns, open interest, funding and basis across perpetuals.
every Monday→04What each model does, how it is reviewed and where it stops.
6 models→Liquidations now show on the correct side on every exchange, and the API reports what share of liquidations is visible on each exchange.
The service puts less load on the server: option data takes noticeably less space and container logs no longer grow without limit.
alpha.trevms.com is back as an open section for alphas: conditions after which an event happens noticeably more often than usual. In Terminal alphas are a toggle above the chart.
Labs is rebuilt around six models, each checked on real data and showing its own calibration. The liquidation map in Terminal loads more reliably.
The ASCII scene set grew from 136 to 304, spread across Terminal, ArcDelta, Finance, Labs, Research and the company site.
The home page, company pages, product hubs, account, docs and Finance have a new look and an ASCII scene for every feature.
The ASCII scenes on the site, in Terminal and in Finance look the same but use noticeably less CPU and GPU.
Estimates the price levels where long and short liquidations are likely to cluster.
Simulates thousands of price paths that follow current volatility and the time of day, and shows how likely a chosen level is to be touched.
Shows how much a position can lose over an hour, a day or three days, at the close and at the worst point within the period.
Estimates the probability that a leveraged position is liquidated within an hour to a week, and the leverage that keeps that risk low.
Uses the order books of five venues to show what a market order of a chosen size really costs and which venue is cheapest.
Checks whether a backtest result is real or just the best of many lucky parameter tries.
The changelog, a live ArcDelta data panel, weekly market reports and Labs model cards are now published here.
Papers with figures built from real data.
Data, model, test, publish; failures loop back.
Branches fork, collect commits and merge back.
A market report every week, page by page.
Data, window, metric and limits for every model.
Uptime of every venue feed, gaps marked.
How much of the expected data actually arrived.
New research delivered to your inbox.
Plain definitions of market terms.
Notes link the work they build on.
Same seed, same data, same curve.
Every note is checked before it is published.
Each snapshot is hashed and kept.
A dated record of notes, data and fixes.
Every estimate ships with its interval.
Daily returns by asset from the weekly report.