Set your password
This link works once. Use 12 or more characters.
Sign in
New to Aurik? Find the engine that fits you →Turn on two-factor
Scan with Google Authenticator or Authy, then type the 6-digit code.
Reading your account…
Pending orders appear here with how close they are and how that strategy has done historically (needs bridge v1.2).
The ring shows how big an average win is next to an average loss for that strategy, from its record (tester and research, 2020–2026). These are breakout engines: they win less often, but each win pays for several losses. A record, not a forecast for this trade.
Waiting for the book from the bridge.
rule summary, no AIWaiting for candle history from the bridge (v1.1).
- Waiting for candles.
No trade is open right now. When Aurik Core or the Swarm opens one, it appears here and on the chart with its live result.
Each dot is a past day. Green = gold closed up, copper = down. Gold ring = today.
Latest headlines
Rule: 1.5x on setups the models call "strong", normal size otherwise (weak setups still win big often enough to keep). Nothing is traded differently: this recalculates your real trades. Every verdict and result is stored permanently.
| Time IST | Type | Engine | Symbol | Lots | Price | Result | Comment |
|---|
Reading the fleet…
Every Mahakal trade the live EA opens is shown to Claude with a chart image, the market context and the module's own record. Claude says BIG (1.5x), NORMAL, SMALL (0.5x) or SKIP. When the trade closes we compare Mahakal's real result with "if we had listened".
- Daily trend agrees
- 4-hour trend agrees
- 1-hour trend agrees
- London or New York session
- Room to the next level ≥ 2R
- Volatility not wild
- Day or week level
A+ = 6-7 checks, A = 5, B = 4, C = 3 or fewer.
Every 30 min, Mon-Fri 12:30-21:30 IST, the strongest Claude model gets the live D1/H4/H1/M15 candles and levels and answers BUY, SELL or NONE. Every call is kept, graded on the 1-minute path with costs, next to a random-side placebo with the same stop and target.
Open paper trades · live
All calls
Every model learns only from data that is actually collected. This page shows how much there is, and what each model can do with it today.
The knowledge index measures how much live evidence the models have: live market days (40 points), closed live trades (30), minute-level price data (20) and candle history (10). It grows with data, not with profit, and it is not a forecast.
One market day counts when the bridge reported on a weekday. The countdown assumes the bridge keeps reporting every market day. Milestones open checks, not profits: a model only goes live after it beats simple fixed sizing on data it has never seen.
Every new setup of both engines is scored the moment it is placed. The v1 models read the market: trend and volatility on 5-minute to daily charts, where price sits against the session and yesterday's levels, the hour, the side, and the strategy's recent form. They say may fail, normal or strong.
They were checked year by year on data they had never seen, with the rules fixed before the test: the Swarm model improved return per unit of drawdown in 8 of 10 years, the Core model in 5 of 5. Right now they only log and get graded. They do not touch the EA.
Next step: after 30 graded live setups, if "may fail" setups keep doing worse than "strong" ones, the models can suggest smaller size on weak setups and larger on strong ones. That reaches the EA only with your approval.
Numbers from the research runs (tester reports, placebo and monkey tests). They do not change on this page; they are the baseline the live engines are compared against.