A trading terminal whose most common answer is WAIT
Atlas reads market structure — swings, breaks, supply and demand zones — across 26 CME futures, scores every setup, and only says BUY or SELL when the structure backs it. The rules are code, the code is backtested in CI, and the trades are paper-filled in real contract specs before anything touches real money.
- Coverage
- 26 CME futures · 1-minute to weekly
- Cadence
- 30-second marks · 5-minute rescan
- Validation
- 2-year sweeps, in- and out-of-sample
Founder — Dmitry Kotov. Status: Live terminal · paper accounts.
The industry is paid on volume, so it never says no
Every retail platform makes money when a trade happens. The result is an interface engineered to produce them: indicators that always print something, alerts that always fire, signals with no memory of having been wrong. The user gets constant encouragement and no discipline, and the losses are treated as their personal failing rather than the product's design.
The tools that do apply discipline are institutional — a quant desk, a research process, an execution stack, none of which fits on one person's screen. In between sits everyone who is serious about the work and has no way to hold themselves to a rule they cannot test.
What the platform does
One pipeline, from raw price to a filled paper trade. Every stage is deterministic and readable — the point is that a person can check it, not trust it.
A market-structure engine, not indicators
Swings, breaks of structure and character, supply and demand zones, fair-value gaps, liquidity sweeps, and volume climaxes — computed from the 1-minute chart to the weekly and drawn where they happened, with the numbers attached.
Ranked verdicts, including the honest one
26 CME futures scored 0–100 on location, how fresh the structural break is, and whether the weekly agrees. BUY or SELL only when the structure supports it; otherwise the board says WAIT, and the automated side touches nothing.
Event risk priced off prediction markets
A risk radar tracks nine geopolitical event chains on Polymarket and links each to the commodity it actually moves — a shipping-lane scare to crude, a weather market to coffee. Live probability changes, not headlines after the fact.
Paper trading in real contracts
Fills are simulated in genuine Micro contracts with commissions, slippage, gap fills, and whole-contract sizing. If one percent of equity does not buy a single contract, the trade is skipped — the same constraint the real account would impose.
The discipline is the product
None of these are features in the usual sense. They are constraints, and they are the reason the output is worth anything.
Rules as readable code
Each strategy is a block of text on the page — universe, entry gate, sizing, stop, target, re-entry rules. Nothing is a black box, which means nothing gets quietly loosened after a losing month.
Validated before it ships
A two-year simulator sweeps parameters autonomously in CI and splits the period in half. A configuration goes live only if it survives both halves — not if it looked good on the whole sample.
Findings that failed are written down
Rules that died in testing stay documented as dead, with the reason. A research process that only records its wins is not a research process.
Paper before capital
Every strategy runs on its own paper account against live data first. The path from an idea to real risk is a sequence of checks, not a decision made on a good week.
Why this generalises past commodities
The engine does not know what it is looking at. Swings, breaks, and zones are properties of a price series, so the same pipeline that ranks gold and crude applies to any liquid instrument with a clean feed — the universe is a configuration file, not an architecture.
The event layer generalises the same way. Linking a prediction market to the instrument it moves is a mapping, and every new pair widens the surface without changing the machinery underneath. That is what turns a personal trading system into infrastructure other people can be sold access to.
Questions investors ask first
- What does Atlas Trader do?
- It reads market structure — swings, breaks of structure and character, supply and demand zones, fair-value gaps, liquidity sweeps — across 26 CME futures from the 1-minute chart to the weekly, scores every setup out of 100, and says BUY or SELL only when the structure supports it. Otherwise the answer is WAIT.
- Why does it say WAIT so often?
- Because most of the time there is nothing there. Retail platforms earn when a trade happens, so their tools are built to always print a signal. Scoring on location, how fresh the structural break is, and whether the weekly agrees means most instruments most days do not qualify — and the automated side touches nothing.
- How are the strategies validated before they go live?
- A two-year simulator sweeps parameters autonomously in CI and splits the period in half. A configuration ships only if it survives both halves, not if it looked good across the whole sample. Rules that died in testing stay documented as dead, with the reason.
- Does it trade real money?
- Each strategy runs on its own paper account against live data, with fills simulated in genuine Micro contract specifications — commissions, slippage, gap fills, and whole-contract sizing. If one percent of equity does not buy a single contract, the trade is skipped, which is the same constraint a real account would impose.
Talk to the founder
For investors, partners, and operators who want the detail behind the summary — unit economics, architecture, and where the platform is going next.