OCTOO AI Independent AI Lab

Trade For Me

TFM: An autonomous market-intelligence and decision system.

TFM watches a live market around the clock, makes decisions, tests them in paper trading and records each decision with its outcome. The market is only the test bench: the subject is autonomous decision-making.

Can a system observe, decide, measure the result and get better?

Markets are a tough test for autonomous systems. They never close, feedback is immediate, and every decision has a result you can measure.

A system that holds up there learns something that carries over to other work: acting on incomplete data, against the clock.

How it works

Eight steps in a loop that learns.

TFM decision loop
  1. 01 Observe Live market and on-chain data.
  2. 02 Filter Safety and data-quality checks.
  3. 03 Analyse Market structure across timeframes.
  4. 04 Decide Rule-based strategies make the call.
  5. 05 Paper trade A simulated position.
  6. 06 Manage The exit follows the plan set at entry.
  7. 07 Evaluate The result is stored with the decision.
  8. 08 Learn Strategies tune themselves, every change logged.

Each result feeds the next round.

Honest record

Measured and on the record.

Every decision is stored with the evidence behind it, right next to what actually happened. Losses stay in the record.

What every decision record holds
Evidence
Market structure, safety checks and data freshness at the moment of decision.
Plan
Fixed at entry. The exit follows it.
Outcome
After costs, stored next to the decision that caused it.
Decision record from a SOLLE AI token report. Three strategies each show the signal “watching”. A box below explains why no paper position was opened: the price fell below the point where the setup breaks, and the market cap of $98,552 is below the $100,000 minimum.
A real decision record from SOLLE AI, which is built on TFM. Three strategies looked at one token and kept watching, and the record says why no paper position was opened. Screenshot solle-ai.com/token/5hpc…pump

Design decisions

01 · Rules first

Decisions you can reproduce.

TFM decides with fixed rules, because you can only evaluate what you can reproduce. Language models help with research and explain each decision.

02 · Paper first

Prove it before it counts.

Every decision is simulated and scored.

03 · Learning loop

Strategies that learn from themselves.

The strategies adjust their own settings from past results and log every change. A human sets the limits and watches the direction.

Spec sheet

Market
Solana tokens. Open around the clock and extremely volatile.
Decisions
Three rule-based strategies that read market structure across several timeframes.
Safety checks
Who controls the token, whether liquidity is locked, how concentrated the holders are, risk scores. No evidence, no entry.
Learning
The strategies tune their own settings from past data and log every change. A human sets the limits.
Execution
Paper trading only. Live trading is switched off.
Codebase
About 180,000 lines of Python and 2,900 commits since February 2026.
Built with
AI coding agents. A human approves every irreversible change.

What this is not

  • A trading bot for hire, a fund or a copy-trading service.
  • Financial advice or a signal service.
  • A language model placing trades. Rules decide, models explain.
Next arm · 03 SOLLE AI An AI character that reads markets in public.

Need a decision system that keeps an honest record?

Build with us