How It Works

4 steps to systematize
your decisions

Trade with the same flow every time and emotion has no room to creep in. Fortuna automates everything from entry to review in a 4-step cycle.

Pre-entry through review

STEP 01Pre-entry

AI research

Just enter a symbol and trading style. The AI reads the market structure from objective data and presents 1–3 scenarios.

  • Binance real-time data (candles, order book, fills, funding, OI) + ATR, VWAP, BTC dominance, DXY, and Fear & Greed combined
  • The Strategy Agent picks 1 of 5 strategies (pullback / breakout / range / reversal / wait)
  • Entry, stop, target, trigger conditions, and invalidation criteria derived for each scenario
  • Per-style stop/target standards enforced
STEP 02Pre-entry

Trade execution

Enter the scenario from the AI research like an exchange. An A/B/C/D grade instantly judges whether to enter.

  • Auto-detects chasing entries, unconfirmed candles, BTC direction conflicts, and mid-range entries
  • Risk-%-based auto position sizing + recommended leverage calculation
  • Effective risk-reward shown after deducting 0.12% in fees
  • D-grade means no entry, A-grade means enter as is, B/C come with a caution
STEP 03Post-entry

Trading journal

Save your entered trades and record the actual outcome after exit. Compare the entry evaluation and the actual result on one screen.

  • Market snapshot saved permanently at entry (reproduce the market state from that moment)
  • Enter exit price, realized R, exit reason, mistake tags, and notes
  • AI review: decision evaluation + execution evaluation + next-trade improvements auto-generated
  • Telegram / Discord notification integration
STEP 04Post-entry

Performance analytics

Auto-aggregates closed trades by grade, by mistake, and by time. Understand your trading patterns from objective data.

  • Average R / win rate / trade count stats by grade
  • Cumulative P&L by mistake tag — your most frequent losing pattern auto-highlighted
  • Track improvement over time with a monthly cumulative R chart
  • AI research history integrated too — what you analyzed and how you acted

3-stage AI pipeline

The AI never invents prices. It takes objective data and only interprets it.

STAGE 1

Data collection

Code (deterministic)

Parallel fetch from 12+ sources including Binance Futures, Spot, CoinGecko, Yahoo Finance, and alternative.me. Swings, FVG, OB, POC, ATR, and VWAP auto-calculated.

Code · Deterministic
STAGE 2

Strategy classification

Claude (classifier)

The collected data is passed to the LLM to pick 1 of 5 strategies. It outputs direction (long/short), confidence, and even the reasons rejected strategies were dropped.

LLM · Strategy Agent
STAGE 3

Scenario generation

Claude (main)

Generates 1–3 scenarios only within the chosen strategy's scope. Entry, stop, and target are derived to match objective data levels.

LLM · Scenario Generator

12+ real-time data
sources combined

It's not just looking at the chart. Fortuna collects objective data simultaneously from Binance, CoinGecko, Yahoo Finance, and alternative.me for a comprehensive judgment.

Price

Multi-TF candles

Three timeframes at once (HTF bias / MTF setup / LTF trigger)

Structure

Swings · FVG · Order Block

Auto-detects highs/lows, unfilled gaps, and strong reversal candles

Liquidity

Order Book

100 levels + buy/sell walls + imbalance

Liquidity

Trade flow (aggTrades)

Last 500 market fills, buy vs sell ratio, large trades

Positioning

Funding + 24h trend

Current funding rate + last 3 + trend (rising/falling/flat)

Positioning

Open Interest + change

Current OI + change vs 1h/4h ago

Volatility

ATR · VWAP

Volatility metric + session volume-weighted average price

Top Trader

Top trader long/short ratio

Whale positioning (correcting information asymmetry)

Macro

Spot-Perp Basis

Spot-futures divergence → squeeze signal

Macro

BTC dominance · DXY

BTC market-cap share + dollar index (inverse correlation)

Sentiment

Fear & Greed Index

Overall market sentiment (extremes are contrarian signals)

Time

Trading session (Asia/EU/US)

Volatility time-of-day context

My data is mine

Row-Level Security

Per-user row access is enforced at the PostgreSQL level. Other users' data can't be accessed even if there's a code bug.

Objective data first

We never ask the LLM for prices directly. We pull real data from the public Binance API and the LLM only interprets it. Price hallucination is impossible.

Transparent storage

Every analysis snapshot and trade result is stored in Supabase under your account only. Delete anytime.

Now experience it yourself