Trend (0-20)
EMA alignment, ADX trend strength, SMA direction. Tells you if the market is moving in a clear direction.

Momentum (0-20)
RSI, MACD, Stochastic. Measures the speed and force of price movement.

Volume (0-20)
Relative volume, VWAP, accumulation/distribution. Shows if smart money is behind the move.

Structure (0-20)
Higher highs/lows, support/resistance, Bollinger position. Market structure shows the pattern.

Volatility (0-10)
Bollinger squeeze, ATR, Keltner compression. Low vol before big moves = opportunity.

Risk Quality (0-10)
Clear stop levels, data quality, trend clarity. Better risk = safer trade.

Trending
ADX ≥ 30 (or 25–30 with strong trend). Best for: Trend Following, Momentum, Swing, Position.

Ranging
Low ADX (0–20), no direction. Best for: Mean Reversion.

Compression
Bollinger squeeze active. Best for: Breakout, Scalping, Swing.

Volatile
High ATR, wild swings. Best for: Scalping. Reduce leverage, wider stops.

Mixed
No clear regime. Signals are weaker. Trade with caution or wait for clarity.

Trend Following 4H / 1D
Rides established trends. Enters on pullbacks in the trend direction. Uses EMA alignment and ADX confirmation. Works best in trending regimes.

Breakout 1H
Catches volatility expansion from compression zones. Looks for Bollinger squeeze + volume expansion + structure breakout. Best in compression.

Mean Reversion 1H
Fades extremes back to average. Uses RSI oversold/overbought + VWAP + Bollinger band mean. Best in ranging markets.

Momentum 1H
Follows strong directional moves. Requires volume confirmation + MACD alignment. Best in trending.

Scalping 15m / 1H
Short-term: tight structure, quick momentum, volatility spikes. Uses 15-minute candles for fast entries and 1H for confirmation. Best in volatile and compression. Targets are small but fast.

Swing 4H / 1D
Multi-day holds. Focuses on higher-timeframe trend and structure. Enters on 4H pullbacks aligned with 1D direction. Best in trending and compression. Larger targets, wider stops.

Position 1D / 1W
Long-term (weeks). Uses daily and weekly macro trends. Only trades with the big picture direction. Best in trending. Lowest leverage, widest stops, biggest targets.

15m — Scalping setups, fast entries. 96 candles (24 hours of data).

1H — Core analysis timeframe (25% weight). 168 candles (7 days).

4H — Intermediate trend and structure (35% weight). 100 candles (~17 days).

1D — Daily trend, dominant weight (40%). 30 candles (1 month).

1W — Weekly macro trend for Position strategy. 26 candles (~6 months).

Confluence — When 2/3 or 3/3 of the core timeframes (1H/4H/1D) agree on direction, the signal gets a confidence bonus. 3/3 = strong, 1/3 = weak.

RSI (Relative Strength Index)
Measures momentum. Below 30 = oversold (potential buy). Above 70 = overbought (potential sell). 14-period standard.

RSI Divergence
When price makes a lower low but RSI makes a higher low = bullish divergence (potential bottom). When price makes a higher high but RSI makes a lower high = bearish divergence (potential top). Weighted in scoring to reduce shorts at bottoms and longs at tops.

MACD
Shows trend momentum. When MACD crosses above signal line = bullish. Histogram shows strength of the cross.

MACD Divergence
Price lower low + MACD histogram higher low = bullish. Price higher high + MACD histogram lower high = bearish. Used with RSI divergence to identify reversal zones.

OBV (On-Balance Volume) Divergence
Price lower low + OBV higher low = bullish (volume accumulating despite price drop). Price higher high + OBV lower high = bearish (volume distributing despite price rise). Volume often leads price.

Stochastic Divergence
Same pattern as RSI: price lower low + Stoch higher low = bullish. Price higher high + Stoch lower high = bearish. 2+ divergence types agreeing = stronger reversal signal.

ADX
Measures trend strength (not direction). Above 25 = trending. Above 40 = strong trend. Below 20 = no trend.

VWAP
Volume Weighted Average Price. Institutions use this. Price above VWAP = bullish bias. Below = bearish.

Bollinger Bands
Shows volatility. Price near lower band in uptrend = buy zone. Squeeze (narrow bands) = big move coming.

Stochastic
Like RSI but compares price to range. K/D crossover in oversold = buy signal. In overbought = sell signal.

ATR (Average True Range)
Measures volatility in price terms. Used to set stop losses and position sizes. Higher ATR = wider stops needed.

Market Structure
HH/HL = bullish (higher highs + higher lows). LH/LL = bearish. Break of structure = potential reversal.

Top & Bottom Detection
Potential bottom: bullish divergence + oversold RSI or near support. Potential top: bearish divergence + overbought RSI or near resistance. The engine penalizes shorts at potential bottoms and longs at potential tops to avoid getting trapped.

EMA (9/21)
Fast/slow exponential moving averages. When 9 EMA crosses above 21 = bullish crossover. Price above both = uptrend.

SMA (20/50)
Simple moving averages. 20 above 50 = bullish alignment. Price below both = downtrend.

Bollinger Bands
SMA(20) ± 2 standard deviations. Trade: Price touching lower band in uptrend = buy zone. Squeeze (narrow bands) = volatility expansion coming — prepare for breakout. Price above upper band = overextended, consider taking profit.

Moving Averages (SMA 20, EMA 9, EMA 21)
Trend direction and dynamic support/resistance. Trade: Price above all MAs = uptrend; buy pullbacks to EMA 9 or 21. Golden cross (EMA 9 above 21) = bullish. Death cross = bearish. Use MAs as trailing stop reference.

Keltner Channel
EMA(20) ± ATR×2. Similar to Bollinger but uses ATR for volatility. Trade: Price at lower channel in uptrend = buy. Break above upper channel with volume = continuation. Tighter than BB — good for mean reversion in ranges.

Donchian Channel
20-period high/low. Trade: Breakout above upper = long signal. Breakout below lower = short. Classic trend-following: enter on channel break, stop at opposite side.

Fibonacci
23.6%, 38.2%, 50%, 61.8% retracement levels from 4H swing. Trade: In uptrend, buy at 0.382 or 0.618 retracement. 0.618 (golden ratio) = strongest support. Place stops below 0.786.

Support / Resistance
Swing highs and lows with role reversal (break above R becomes S). Trade: Buy at support, sell at resistance. Break and hold = new level. Use for stop placement (below S for longs, above R for shorts).

POC (Point of Control)
Price level with highest volume — magnet for price. Trade: Price tends to return to POC. Buy when price dips to POC in uptrend. Sell when price rallies to POC in downtrend. Strong S/R when combined with structure.

Pivot Points
Classic P, R1, R2, S1, S2 from prior candle's H/L/C. Trade: Day traders use for intraday levels. Price above P = bullish bias, target R1/R2. Below P = bearish, target S1/S2. Good for scalping and session trading.

Chart Patterns
Flags, wedges, triangles, head & shoulders. Trade: Bullish pattern = buy on breakout above resistance. Bearish = short on breakdown. Target often = pattern height. Wait for volume confirmation.

Action Badges
Your trade actions (BE, TS, LOCK, EXIT, etc.) on the chart. Use: Visual history of where you moved stops or took profit. Helps review trade management.

Candlestick Patterns
Engulfing, hammer, doji, etc. Trade: Bullish patterns at support = buy. Bearish at resistance = sell. Best when aligned with higher timeframe trend.

Order Blocks
Last opposing candle before a strong move. Bullish OB = support zone. Bearish OB = resistance. Trade: In uptrend, buy when price retraces to bullish OB. In downtrend, short at bearish OB. Place stop below/above the block. Used by institutional traders.

Fair Value Gaps (FVG)
3-candle imbalance — price "gapped" without trading. Trade: Bullish FVG = buy zone (price often fills the gap from below). Bearish FVG = sell zone. Enter on retest of the gap. Often get filled quickly.

Liquidity Clusters
Grouped swing highs (above) and swing lows (below). Trade: Price often sweeps liquidity (stops) before reversing. Near cluster above = potential resistance/short zone. Near cluster below = potential support/long zone. Don't chase — wait for sweep.

VWAP
Volume Weighted Average Price. Trade: Price above VWAP = bullish bias. Below = bearish. Institutions use VWAP as benchmark. Buy dips to VWAP in uptrend. Sell rallies to VWAP in downtrend. Strong S/R for intraday.

Premium/Discount
Equilibrium (50%) level between swing high and low. Trade: Price in premium (above 50%) = expensive, consider shorts. In discount (below 50%) = cheap, consider longs. Mean reversion to equilibrium.

Swing Points
Marked swing highs (H) and lows (L). Trade: Identifies structure. HH + HL = bullish. LH + LL = bearish. Use for trend confirmation and stop placement (below last swing low for longs).

Market Structure
Label (BULLISH, BEARISH, RANGING, etc.) in bottom-left. Trade: Only long in BULLISH, only short in BEARISH. RANGING = wait for breakout or use mean reversion. BREAK_UP/BREAK_DOWN = structure change.

Tip: Combine 2–3 indicators for confluence. E.g. long at bullish OB + support + VWAP = strong setup. Don't overload — use Overlays + Levels + one Advanced group for clarity.

1. Never risk more than 2% per trade
If your account is $10,000, max risk per trade = $200.

2. Always use a stop loss
No exceptions. The system calculates stops using ATR + structure.

3. Risk/Reward minimum 1.5:1
Your potential profit should be at least 1.5x your risk.

4. Leverage matches confidence
Only use higher leverage when the score is 70+. High vol = low leverage.

5. 1 trade per pair
Don't stack positions. Wait for your trade to close before opening another on the same pair.

6. Follow the system
Journal shows you win more when you follow rules. Trust the process.

Score 80+: Up to 10x (very strong setup)

Score 70-79: Up to 7x (strong setup)

Score 60-69: Up to 5x (good setup)

Score 50-59: Up to 3x (moderate)

Score 40-49: Up to 2x (cautious)

Below 40: 1x only (weak setup, consider skipping)

Volatile market? Cut leverage by 50%. Extreme vol? Use 1x only.

How it works
Every closed trade is recorded with its strategy, regime, and result (win/loss + R:R). The learning engine adjusts strategy weights based on real performance. Those weights are now used live in the trading engine to score signals.

Weight adjustment
Uses both expectancy AND profit factor. Strategies with negative expectancy AND PF < 0.5 decay faster (10%). Positive expectancy AND PF > 2.0 boost faster (5%).

Regime-specific learning
Wins/losses tracked by regime. Strategies with >55% win rate in the current regime get a bonus. <40% win rate gets penalized. Min 5 trades per regime.

View it live: AI Engine | Strategy Comparison — side-by-side scores, regime stats, and score history.

Signals & scoring
1. Learned strategy weights used in scoring. 2. Min score (52) + dynamic confluence gate (score ≥58 needs 1 TF; 52–57 needs 2 TF) – weak setups show HOLD. 3. BTC regime filter – no alt longs when BTC STRONG_SELL, no alt shorts when BTC STRONG_BUY. 4. MTF divergence penalty when 1H and 4H disagree. 5. RSI, MACD, OBV & Stochastic divergence – weighted toward bottoms/tops; bullish div boosts longs, bearish boosts shorts; 2+ types agreeing = stronger modifier; penalizes counter-trend at extremes. 6. Top/bottom detection – penalizes shorts at potential bottoms and longs at potential tops. 7. Session filter – small score reduction outside 12–22 UTC. 8. Regime–strategy gating – e.g. no Mean Reversion in trending. 9. Min 5 trades per strategy before it can be selected.

Levels & risk
8. Dynamic ATR stops by regime (tighter in trending, wider in volatile). 9. Multi-timeframe S/R (1D/4H used for stops when available). 10. VWAP bands – bonus when price is in value zone (within 0.5 ATR of VWAP). 11. Fibonacci retracement – stops refined using 0.618 and key fib levels. 12. Volume Profile/POC – point of control adds S/R confluence.

Execution & position
13. Main trade buttons require score ≥55 and dynamic confluence gate (score ≥58 needs 1 TF; otherwise 2 TF). 12. Entry at live price – entry uses current market price when you open. 13. Multi-strategy trades – on each coin’s detail page you can open a trade using the main signal or any top strategy (Scalping, Momentum, Breakout, etc.) that scores ≥55; each strategy has its own stop and take-profit levels (scalping = tighter, swing/position = wider). 14. Trailing stop – stop moves to breakeven after 1R profit. 15. Slippage simulation (0.05%) on entry and exit. 16. Confidence-weighted position size (higher score = slightly larger size, cap 1.2×). 17. Slippage simulation (0.05%). 18. Streak-aware + Kelly sizing – position size adjusts for win/loss streaks and uses fractional Kelly criterion. 19. Cooldown – 4h same-direction re-entry block.

Data & learning
20. Strategy weights from AI Engine feed into scoring. 21. Expectancy drives strategy selection. 18. Quality gate and session messages appear in “Why This Trade” when relevant. 23. Faster weight learning for strategies. 24. Funding rate contrarian signals. 25. BTC correlation scoring.

DCA (Dollar Cost Average)
Optional feature (off by default). When enabled, adds to losing positions when the signal re-confirms the trade direction. How it works: 1. Price dips X% from avg entry (default 2% per level, stacking: 2%, 4%, 6%...). 2. Engine re-analyzes the coin -- signal must still agree with trade direction and score must be above min threshold (default 52). 3. Adds a configurable % of original position size (default 100%). 4. Max 3 adds (configurable 1-10). 5. Average entry price recalculates (volume-weighted). 6. Balance must have sufficient margin for each add. Configure in Performance > Feature Toggles > DCA section.

Trailing Take-Profit Mode
Alternative to fixed TP1/TP2/TP3. When set to Trailing, no fixed TPs are used -- the stop trails behind price from entry, riding the trend until stopped out. Two distance modes: 1. ATR-based (dynamic): trail distance = N x ATR at entry (default 1.5x, configurable 0.5-5.0). Adapts to each coins volatility. 2. Fixed % (static): trail distance = X% of entry price (default 2%, configurable 0.5-10%). As price makes new highs (longs) or new lows (shorts), the stop follows immediately. Exit happens when price reverses enough to hit the trailing stop. Configure in Performance > Feature Toggles > Take-Profit Mode. Also available as a backtest toggle.

Order blocks, FVG and liquidity clusters
19. Order blocks – last opposing candle before a strong move (bullish OB = support, bearish OB = resistance); structure bonus when price is near a valid OB. 20. Fair value gaps (FVG) – 3-candle imbalances; price in or near a FVG zone adds confluence. 21. Liquidity clusters – swing highs/lows grouped by proximity; near a cluster adds structure context and appears in reasoning.

1. Funding Rate Contrarian Signal
Fetched from Bitget Futures every 60 seconds for all tracked coins. Extreme positive funding (>0.1%) means the market is overleveraged long — bearish contrarian signal. Extreme negative (<-0.1%) means overleveraged short — bullish contrarian. The score penalizes trades going with the crowd and rewards contrarian positions. During trade monitoring, funding rate warnings appear in the score check.

2. BTC Correlation Score
Calculates Pearson correlation between each alt's hourly returns and BTC's returns. When correlation is high (>0.7) and BTC's direction disagrees with the alt's signal, the score is penalized. This prevents entering alt trades that are likely to be overridden by BTC's movement. During monitoring, a "High BTC correlation — watch BTC" message appears for correlated pairs.

3. Volume Profile / Point of Control (POC)
Calculates the price level with the highest traded volume across the last 100+ candles. The POC acts as a strong support/resistance level — price tends to gravitate back to it. When price is within 0.5% of POC, the score gets a bonus (+3). POC is also used as additional S/R when calculating stop loss and take profit levels.

4. Fibonacci Retracement Levels
Calculates key Fibonacci levels (23.6%, 38.2%, 50%, 61.8%, 78.6%) from the 4H swing high/low. These levels refine stop loss placement — if a strong fib level (especially 0.618, the golden ratio) sits between your entry and calculated stop, the stop is tightened to just below/above that fib level. Fibonacci levels also provide additional S/R for trade level calculations.

5. Win/Loss Streak Awareness
The system tracks your consecutive wins and losses. After a losing streak:
• 2 consecutive losses: position size reduced by 25%
• 3+ consecutive losses: position size reduced by 40%
After a winning streak (3+ wins): position size boosted up to 15%. This protects capital during drawdowns and lets you lean in when you're on a roll.

6. Adaptive Position Sizing (Kelly Criterion)
Uses the fractional Kelly criterion (25% of full Kelly) based on each strategy's win rate and average risk/reward ratio from the learning engine:
Kelly% = WinRate - (1 - WinRate) / AvgRR
If Kelly is positive, position size is capped to 25% of Kelly × balance. If Kelly is negative (strategy is losing money), position size is reduced by 50%. This ensures profitable strategies get more capital while unprofitable ones are de-risked automatically. Requires at least 15 trades of data.

7. Stricter Regime-Strategy Gating
Strategies are now more aggressively blocked in wrong market conditions:
• Mean Reversion: blocked in trending AND compression (only allowed in ranging/mixed/volatile)
• Trend Following: blocked in ranging AND volatile
• Breakout: blocked in trending (needs consolidation first)
• Swing: blocked in volatile (gets stopped out)
• Position: blocked in volatile AND compression
This reduces false signals from strategies being used in the wrong market regime.

Action Badges

How Auto-Actions Work
Two systems run in the background:

1. Stop Management (every 30 seconds)
• Take-profit split – When price hits TP1, 40% of the position closes. At TP2, 30% closes. At TP3, the remaining 30% closes.
• Breakeven (BE) – When price moves 1R in your favor (1R = distance from entry to your original stop loss), your stop automatically moves to your entry price. Zero risk from this point.
• Trailing Stop (TS) – At 1.5R profit, the stop begins trailing 1R behind the best price. As price makes new highs (longs) or new lows (shorts), the stop follows, locking in gains.
• Stepped Lock-in – If trailing is off, stops lock at 0.5R, 0.75R, and 1R as the trade progresses 50%, 75%, 90% toward TP2.
These are controlled by "Auto move to breakeven" and "Auto trailing stop" in your Settings.

2. Score Check Actions (every 5 minutes)
The engine re-analyzes the market and compares the current score to your entry score. Based on how the score has shifted:
• Hold – Score moved 20+ in your favor. Thesis is strong, hold.
• Tighten stop – Score dipped slightly (-5) or heat is yellow. Moves stop to breakeven.
• Lock in profit – 50%+ progress toward TP2. Steps stop to lock 0.5R/0.75R/1R.
• Take partial (PP) – Near TP1 and signals weakening. Closes 33% of position.
• Reduce position (RP) – Score dropped -15+ against the trade. Closes 50%.
• Consider exit (EXIT) – Red heat + danger + score -25+ against. Closes entire position.

Score check actions only auto-execute if "Auto-execute score check actions" is enabled in Settings. Otherwise they appear as suggestions only.

Leverage-Aware Thresholds
At 3x+ leverage, the system uses tighter thresholds because losses are amplified:
• Exit threshold: -20 (vs -25 at 1x)
• Reduce threshold: -10 (vs -15 at 1x)
• Yellow heat + score slipping at 3x+ = automatic "Reduce position" suggestion.
• Trades open 48+ hours with less than 1% price movement get a "Trade stale" warning.

Progress Indicator
Each trade card shows how close the trade is to triggering the next action:
• "BE at 1R: 45%" – Trade is 45% of the way to breakeven trigger.
• "Stop at breakeven" – BE already triggered, showing progress to trailing (1.5R).
• "Trailing active" – Stop is actively following the price.
• "(underwater)" – Trade is in the red, no auto-action will fire yet.

How Chart Patterns Work
Chart patterns are geometric formations formed by price over many candles — distinct from single/multi-candle patterns (Tier 1-3). These are the classic technical analysis formations like flags, triangles, wedges, head & shoulders, and double tops/bottoms.

The engine scans each timeframe for these patterns using swing point detection and linear regression trendlines. Each pattern is scored based on:
• Pattern type reliability — historical win rates from backtesting
• Completion % — how close the pattern is to triggering a breakout
• Volume confirmation — declining volume during consolidation = higher reliability
• Breakout volume spike — volume on the breakout candle(s) > 1.2× pattern average = confirms real breakout, not fake-out (1.25× weight)
• Close beyond boundary — last candle must close beyond the pattern boundary (not just wick through) = reduces false breakout signals (1.2× weight)
• Trend alignment — continuation patterns in the trend direction get a 1.3x boost

Chart patterns add up to +6 Structure points and feed bull/bear direction points into the scoring engine. They appear as trendlines drawn directly on your chart (toggled with the "Chart Pat." checkbox).

Continuation Patterns — trade with the trend

Reversal Patterns — trade against the prior trend

Neutral Patterns — breakout direction determines the trade

Chart Pattern Scoring & Integration
Here's how chart patterns influence trade decisions:

How Candlestick Patterns Integrate
The engine scans every timeframe (15m through 1W) for 24+ candlestick patterns. Each detected pattern receives a context-weighted score that feeds into the Structure (+5 max) and Momentum (+3 max) sub-scores.

Bullish patterns (green) add bullPoints → support LONG signals.
Bearish patterns (red) add bearPoints → support SHORT signals.

A pattern alone is a weak signal. A pattern at the right place with confirming indicators is powerful. The system multiplies pattern weight based on context:
• At support/resistance: 1.5x
• With RSI divergence: 1.3x
• High volume confirmation: 1.3x
• At order block: 1.3x
• In BB squeeze: 1.2x
• Higher TF trend aligned: 1.2x
• Against higher TF trend: 0.6x (reduced)
• Against S/R level: 0.6x (reduced)

Single-Candle Patterns (Tier 1)

Multi-Candle Patterns (Tier 2)

Complex Patterns (Tier 3)

Pattern + Context = Real Edge
Here's how patterns stack with the existing system to improve signal quality: