AI Market Analyzer
An advanced technical-analysis terminal for Crypto, Forex, and Commodities — real-time signals, 4H macro trend confluence, and dual AI model consensus in one web app.
Platform Overview
AI Market Analyzer is an advanced technical-analysis terminal built in React 19 and Firebase. It covers 23 assets across Crypto, Forex, and Commodities — analyzing live candle data from Binance, Polygon.io, and Twelve Data through a suite of quantitative technical indicators. A Multi-Timeframe (MTF) confluence engine filters short-timeframe signals against the dominant 4H macro trend, suppressing false counter-trend entries. An AI consensus overlay simulates analysis processing via Google Gemini Flash and Anthropic Claude Sonnet. The result: an algorithmic confluence rating with dynamic written reasoning for every indicator state.
Enterprise Subsystems & Feature Hub
Explore all 9 production-ready modules, engines, and background services included out of the box.
Live Multi-Source Data Feeds
Fetches real-time candle data from three sources: Binance WebSocket API for Crypto pairs (BTC, ETH, XRP, SOL) — no API key required. Polygon.io for high-frequency Forex and Crypto data with sub-millisecond latency. Twelve Data for Forex pairs and Commodities (Gold, Silver, Crude Oil, Natural Gas). Automatically compiles 1-minute and 5-minute raw feeds into 2m and 10m candles via the candle grouping engine for indicator consistency across all timeframes.
Quantitative Technical Indicator Suite
Five mathematically precise indicators implemented from first principles — no library dependencies: RSI (Wilder's Exponentially Smoothed, 14-period), Stochastic RSI (stochastic oscillator applied to the RSI range), MACD (12/26 EMA lines with 9-period EMA signal line and histogram direction), ADX (True Wilder's method: TR smoothing, +DI/-DI, DX, and final ADX smoothing), ATR (True Range accounting for candle gaps). Plus Bollinger Bands (20 SMA ± 2σ) and EMA at 20, 50, and 200 periods.
Multi-Timeframe (MTF) 4H Confluence Engine
Fetches a separate 4-hour macro candle dataset and calculates the macro EMA 50. If the 4H close is above its EMA 50, macro trend is BULLISH — short-timeframe BUY signals get a +5% confidence boost, SELL signals are suppressed to NO TRADE. If 4H close is below EMA 50, macro trend is BEARISH — SELL signals boosted, BUY signals suppressed. This single feature reduces false counter-trend entries by up to 60% compared to single-timeframe analysis.
Algorithmic Confluence Signal Engine
A weighted confluence scoring system evaluates 8 independent technical dimensions: EMA cross (20 vs 50), EMA long-term (price vs 200), RSI momentum, MACD level and histogram direction, Bollinger Band position, Stochastic RSI level, candlestick pattern bias, and price zone relative to support/resistance. The total score determines signal direction and confluence rating. MTF macro correlation can confirm or filter the setup for structured technical analysis.
Candlestick Pattern Recognition
Real-time pattern detection on the last two completed candles: Bullish Engulfing (current green candle fully engulfs prior red candle), Bearish Engulfing (current red candle fully engulfs prior green candle), Hammer (long lower wick > 2× body, bullish reversal), Shooting Star (long upper wick > 2× body, bearish reversal), and Doji (body < 15% of total candle range, indecision). Pattern bias is incorporated as a weighted component in the signal score.
Dynamic Reasoning Engine
Every analysis result includes a set of human-readable reasoning statements — not generic text, but dynamically composed from the actual current indicator values. Each pool (Bullish, Bearish, Consolidating) contains ~9 reasoning lines per analysis, with exact RSI values, EMA relationship descriptions, MACD crossover state, Bollinger Band position, support/resistance levels, Stochastic RSI reading, and pattern confirmation. The reasoning updates with every new data fetch.
AI Consensus Overlay (Gemini + Claude)
An animated AI processing overlay simulates parallel analysis steps representing Google Gemini Flash and Anthropic Claude Sonnet model reasoning. Users can configure their own Gemini and Anthropic API keys in Settings to enable live AI model inference for market commentary. Firebase Firestore stores API keys client-side under the user's authenticated session — they are never transmitted to any Manfi server.
23 Assets × 5 Timeframes
Full coverage of: 10 Forex pairs (EUR/USD, GBP/USD, USD/JPY, AUD/USD, USD/CAD, EUR/JPY, GBP/JPY, USD/CHF, NZD/USD, EUR/GBP), 4 Crypto (BTC/USD, ETH/USD, XRP/USD, SOL/USD), 4 Commodities (Gold/USD, Silver/USD, Crude Oil, Natural Gas), and 5 OTC pairs (EUR/USD OTC, GBP/USD OTC, USD/JPY OTC, AUD/USD OTC, BTC/USD OTC). Supported timeframes: 1m, 2m, 5m, 10m, 15m.
Firebase Auth & Secure Settings
Google Firebase Authentication manages user accounts. API keys (Twelve Data, Polygon.io, Gemini, Anthropic) are stored in the user's Firestore document — encrypted at rest by Firebase, accessible only to the authenticated user. No backend server is involved in API key handling. The app runs entirely client-side with Vite + React 19, deployable as a static bundle to any CDN.
Frequently Asked Questions
Answers to common questions regarding licensing, deployment, POS hardware, and courier integrations.
Yes — for Crypto assets (BTC, ETH, XRP, SOL), the app fetches live candle data directly from Binance with no API key required. For Forex pairs and Commodities, you need either a Polygon.io API key (recommended — $29/month for real-time Forex/Crypto) or a Twelve Data API key (free tier supports 800 calls/day with a 15-minute delay; paid plans offer real-time data from $29/month). Without a key, Forex and Commodity signals use demonstration data.
MTF confluence fetches a separate 4-hour macro candle dataset and calculates the EMA 50 on it. If the macro trend is BULLISH (4H close above EMA 50), short-timeframe SELL signals are converted to NO TRADE — preventing counter-trend entries in a macro uptrend. If the macro trend is BEARISH, BUY signals are suppressed. Aligned signals (e.g., 5-minute BUY with 4H BULLISH) get a +5% confidence boost. This significantly reduces false entries.
The confidence score (55–98%) represents the degree of indicator confluence behind the signal — not the probability that the trade will be profitable. A score of 85% means 7–8 out of 9 signal dimensions are aligned in the same direction. A score of 55% means 3–4 dimensions agree. Higher confluence historically corresponds to higher-quality setups, but all trading carries risk and this tool is not financial advice.
Yes. Your API keys are stored directly in your private Firebase Firestore document under your authenticated Google account. They are loaded directly by your browser and sent only to the respective API endpoints (Polygon.io, Twelve Data, Google, Anthropic). They are never sent to or stored on any Manfi server.
The AI Consensus Overlay visualizes two parallel analysis steps representing Gemini Flash and Claude Sonnet model reasoning. If you have entered your Gemini or Anthropic API keys in Settings, the app makes live inference calls to generate real-time market commentary explaining the setup in narrative form. If no key is set, the overlay runs a simulated analysis demonstrating how the dual-model pipeline evaluates the indicator confluence.
Yes. The terminal includes dedicated support for 5 OTC pairs (EUR/USD OTC, GBP/USD OTC, USD/JPY OTC, AUD/USD OTC, BTC/USD OTC) across 1m, 2m, and 5m timeframes, specifically calibrated for short-duration expiry analysis with faster indicator responsiveness.