0x20 How it works

From signal to decision

Market data → indicators → AI interpretation → signal → risk check → position sizing → execution → performance analysis. Eight stages, each with a defined input and output.

  1. 0x01

    Market data

    Price, volume, ETF flows, bond yields and macro indicators, collected and normalised.

  2. 0x02

    Indicators

    MACD, RSI, ATR, ADX and trend strength, computed across several timeframes.

  3. 0x03

    AI interpretation

    Models read the indicators together and say, in plain language, where they agree and where they conflict.

  4. 0x04

    Signal

    A graded research signal — or no signal at all when the evidence is mixed.

  5. 0x05

    Risk check

    Volatility regime, liquidity and existing exposure are tested before anything proceeds.

  6. 0x06

    Position sizing

    Size follows risk, not conviction. Wider ranges mean smaller positions.

  7. 0x07

    Human decision & execution

    A person decides. Entries and exits are recorded with the full indicator state at the time.

  8. 0x08

    Performance analysis

    Every outcome — win or loss — is reviewed and fed back into the research.

0x21 Stage by stage

What happens at each step

  1. 0x01

    Market data

    Everything starts with observed data. Inputs from different markets arrive on different clocks — crypto never closes, ETFs trade exchange hours, yields update through the bond session — so they are aligned to common timeframes before anything is computed from them.

    In
    Raw prices, volumes, flows, yields, macro releases
    Out
    Clean, time-aligned series per instrument
  2. 0x02

    Indicators

    Standard technical indicators are calculated on each series, on more than one timeframe. The same instrument can look bullish on the hourly chart and bearish on the four-hour chart; both readings are kept, because the disagreement is itself information.

    In
    Time-aligned series
    Out
    Momentum, volatility and trend-strength readings per timeframe
  3. 0x03

    AI interpretation

    Statistical and machine-learning components weigh the indicator set alongside cross-asset context. The result is a written interpretation, not a price target. When the inputs disagree, the interpretation is “conflicting signals” — it is not rounded up to a call.

    In
    Indicator readings, cross-asset context, sentiment
    Out
    A plain-language interpretation and a confidence grade
  4. 0x04

    Signal

    A signal states a direction and a tier. The tier describes how strongly the inputs agreed at the time. It is not a probability of profit, and most market states produce no signal.

    In
    Interpretation and confidence grade
    Out
    Direction and tier, or nothing
  5. 0x05

    Risk check

    A signal is examined against the conditions it would be acted on in: how wide the recent range is, how deep the market is, and what the portfolio already holds. A good-looking signal in a bad regime can be rejected here.

    In
    Signal, volatility regime, liquidity, current exposure
    Out
    Pass, reduce or reject
  6. 0x06

    Position sizing

    Size is derived from how far the instrument typically moves, so that an ordinary adverse move is survivable. Leverage multiplies both sides of the outcome and is treated as a risk input, never as a way to make a small edge look large.

    In
    Passed signal, ATR, account risk limits
    Out
    Position size and risk per trade
  7. 0x07

    Human decision & execution

    The system produces research; a human decides whether to act on it. When a trade is taken, the entry, exit and the indicator state at entry are written to the record so the decision can be reconstructed later.

    In
    Sized, risk-checked signal
    Out
    A recorded trade — or a recorded decision not to trade
  8. 0x08

    Performance analysis

    Closed trades are reviewed for what the system saw, what it missed, and how far price moved for and against the position. Losing trades are published in the journal alongside winning ones.

    In
    Closed trade and its recorded state
    Out
    Journal entry, review notes, changes to the research
0x22 The human step

The system informs. A person decides.

Nothing in the pipeline removes the need for judgment. The output of the research is a description of conditions and a graded signal; whether to act on it is a decision made by a person who can see things the system cannot.

Read the methodology →

See the system architecture →