Are those results real trading performance?

No. It is paper-only, and Jev may choose either a fixed long-only spot paper policy or no trade at all. The simulator enters on the next eligible 1-minute candle and records fees, spread, slippage, exit reason and the source candles, while no-candidate, blocked, no-trade, pending, skipped and unpriceable scans all stay recorded as distinct outcomes.

How much can this cost in API calls?

The private runner allows at most 97 TypeSafe requests per UTC day by default, adjustable through TYPESAFE_DAILY_REQUEST_LIMIT up to 10,000. Each request reserves a daily slot in D1 before it is sent, reservations count even if the call then fails, and SDK retries are disabled so one reservation cannot fan out. Candidate gating usually uses fewer.

What if the scheduled job runs away?

A D1 control row has a pause switch that stops scheduled collection, Jev calls and settlements after the next handler check, and the run history with request counts is queryable. Worker invocations still happen while paused, so you also disable or remove the Cron Trigger. These controls cut runaway usage but cannot enforce a Cloudflare billing cap.

What does Jev actually get to see?

A neutral structured state built from completed candles only: multi-timeframe returns, RSI, MACD, ATR and moving averages, UTC session stats, Bybit funding, mark and index price, open interest, a short liquidation observation and top-20 order book levels. All candidate questions share one request, the key stays server-side in the private runner, and no-candidate scans cost nothing.

How do I run it locally?

Copy .env.example to .env, add TYPESAFE_API_KEY, then npm install and npm run experiment. It writes append-only records to data/ and you can open the read-only dashboard at http://127.0.0.1:3000; restarting is safe because existing boundary IDs are not duplicated.

How is the Cloudflare deployment structured?

A public Worker serves the dashboard and API and owns the Cron Trigger, which invokes a separate private runner through a Service Binding. The runner is placed in aws:eu-north-1 and is the only piece storing TYPESAFE_API_KEY; Cron runs at minutes 1, 16, 31 and 46.

What does the dashboard keep separate?

Five concepts: paper results such as net P&L, expectancy and drawdown; paper decisions with Jev's action, Choice distribution, Noul probability, eligibility, fills and costs; the candidate log including no-candidate and blocked scans; archived forecast diagnostics; and audit records with the exact State and source candles.

CommunityJev Tools & CLIs

Btc Jev Signal

Paper-only BTC experiment on Bybit data where code gates the market state and Jev picks a long-only spot policy or no trade, run on a Cloudflare cron.

See highlights

Open github.com

Stars: 5
Language: TypeScript
Published by: Community
Author: WebGrga
Last verified: Oct 2026

WebGrga/btc-jev-signal

User repository on GitHub

View on GitHub

Experimental multi-horizon BTC signal generator using TypeSafe Jev probabilities and Binance market data.

Forks: 1
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

A public, paper-only BTC research experiment using TypeSafe Jev with Bybit spot and perpetual-futures market data. A public Cloudflare Worker serves the dashboard and schedules a private Stockholm-placed Worker, which collects data, requests Jev decisions, and records simulated outcomes. It does not place orders.

Quoted from the project's README. Read the full README.

Tagscommunity jevbest catalog-import awesomejev

Works forJev for Classification

Used inJev for Software & DevTools

Languages: TypeScript

Highlights

Code checks freshness, volatility, spread, costs and position limits first, and a no-candidate scan makes no API request at all.
The dashboard keeps paper results, paper decisions, the candidate log, archived forecast diagnostics and audit records deliberately separate.
Simulated entries fill on the next eligible 1-minute candle and record fees, spread, slippage, exit reason and the source candles used.
The private runner caps itself at 97 TypeSafe requests per UTC day by default, reserving a slot in D1 before each request goes out.

Watch out

Nothing trades. This is described as a personal software experiment, not financial advice, investment research or a trading service, and the simulator never places or prepares an order. The dashboard itself says small samples are descriptive only and that Jev confidence describes concentration in the returned distribution rather than guaranteed correctness. The usage guard reduces runaway spend but cannot enforce a Cloudflare billing cap, reservations count even when the provider call later fails, and deploying needs a Cloudflare account plus Node 20.19 or newer. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jev-recall

Python memory filter that judges every stored memory against the current request with one calibrated yes/no per memory in a single Jev request, keeping everything above a threshold.