Forecast and Analytics for Bitcoin (BTC)

Data Analysis and Algorithm

The forecast is generated by an ensemble of machine learning models (based on XGBoost) and is updated automatically at the close of each candle. The neural network collects a window of the last 100 bars and analyzes historical prices, tick volumes, volatility, as well as technical oscillators.

Monte Carlo Method

The system does not provide a single rigid scenario. After evaluating the current trend, the algorithm conducts 30 independent simulations of probable futures, incorporating historical market noise. The final signal (BUY or SELL) is determined by the direction taken by the majority of these 30 scenarios.

Confidence

Displays the percentage of simulations (out of 30) that align with the final signal:

• Below 60% — "Market Noise": The signal is weak, and the models' predictions are divided. Market entry is not recommended.
• 60% – 80% — "Moderate Confidence": A statistical advantage supports the signal. Standard trading environment.
• Above 80% — "High Confidence": Strong consensus among the models. Maximum probability of scenario execution.

Dynamic Levels (Stop Loss / Take Profit)

Stop Loss and Take Profit are calculated adaptively based on current market volatility (ATR indicator). During periods of high market turbulence, stops are widened to prevent the trade from being stopped out by random noise, and they are tightened during a calm market.

Classic divergence is a powerful reversal signal. However, using oscillators has one vulnerability: against a strong global trend, they generate numerous false signals. To increase entry accuracy, we implemented an MTF filtering system.

Core of the Algorithm

The platform doesn't just look for divergence on the current chart, but automatically checks the trend direction on the higher, controlling timeframe. The global trend is determined by the price position relative to the 200-period Exponential Moving Average (EMA 200)

Timeframe pairing

• Signals on M5 and M15 are checked against the trend on H1.
• Signals on M30 are checked against the trend on H4.
• Signals on H1 and H4 are checked against the daily trend D1.

How to read the statuses in the table?

• 🟩 Confirmed: Ideal setup. The direction of the divergence matches the global trend of the higher timeframe (trend trading after a pullback)
• 🟥 Counter-trend: Risky setup. The indicator shows a reversal, but the higher global trend is still pointing in the opposite direction. It is recommended to skip the signal or trade with reduced volume.

To search for chart patterns, our AI does not use rigid price frames, which often fail. Instead, it applies a dynamic algorithm for finding extremes (peaks and troughs), normalized via the ATR volatility indicator. This filters out market noise and finds mathematically precise patterns in both quiet and impulsive markets.

Reversal Patterns

• Double Top / Double Bottom: Classic patterns indicating trend exhaustion. The AI looks for two peaks (or troughs) at the same level. They signal the price's inability to break a strong resistance or support level, followed by a reversal.
• Head and Shoulders: One of the most reliable patterns. The algorithm identifies three consecutive extremes, where the central one (head) is higher than the others. It foreshadows a shift from an uptrend to a downtrend (or vice versa for an inverted pattern).

Consolidation Patterns

• Triangles (Ascending, Descending, Symmetrical): Occur when volatility drops and support and resistance lines converge. The AI measures the slope angles of the lines to pinpoint the "coiling" moment before a strong impulsive breakout.
• Wedges (Bullish and Bearish): Unlike triangles, both lines of a wedge point in the same direction (up or down). They indicate that the current trend is "exhausted" and often lead to a sharp breakout in the opposite direction.

Continuation Patterns

• Flags and Pennants: Short-term pauses after an aggressive price move (flagpole). The AI looks for a sharp impulse followed by tight consolidation. Usually, the move is expected to continue in the direction of the initial impulse.
• Rectangle: A sideways channel (flat), where the price consolidates between parallel horizontal levels. A breakout of one of the boundaries sets the asset's further direction.

Confidence: The algorithm evaluates the quality of each pattern (from 60% to 95%). The higher the percentage, the closer the pattern's proportions match academic technical analysis models.