Chinese Bot

01 — trained_on_tick_data

The models are trained on years of tick-level price data — not daily OHLCV, but every single price movement across 40+ forex pairs, commodities, and indices. This gives the AI an understanding of market microstructure that traditional indicator-based signals completely miss. A 5-second model sees patterns a human scalper would need years to internalize.

02 — per_timeframe_models

One model for one timeframe doesn't work — a 5-second candle is 90% noise, a 1-hour candle is 90% signal. Chinese Bot runs 16 separate models, each optimized for its specific timeframe. Ultra-short TF models prioritize momentum and order flow. Long TF models weight trend strength and volatility regimes. No fixed parameters — each model learns what matters for its window.

03 — cross_timeframe_validation

Every signal is cross-checked against the next higher timeframe's model output. A 5-second DOWN signal only fires if the 30-second model agrees on direction. A 5-minute signal requires 15-minute confirmation. This cascade eliminates ~40% of false positives compared to single-TF systems. The AI doesn't just look at one chart — it sees the whole market structure.

04 — confidence_scoring

The model doesn't just output UP or DOWN — it outputs a confidence score (0-100%) based on how strongly the input pattern matches historical winning setups. Each signal also includes a volatility rating (low/medium/high). Skip sub-60% confidence signals. Size up on 85%+ signals. Most platforms give you a binary direction — Chinese Bot tells you how certain the AI is.

05 — real_time_inference

Free signal pages refresh every 15 minutes. Chinese Bot runs model inference on every incoming tick — the instant price moves, the model evaluates and fires a signal if conditions are met. Delivery is via Telegram bot, WebSocket API, and web dashboard. In binary options, a 5-second delay is the difference between ITM and OTM.