Help & FAQ

1. Login with your account credentials (default admin: admin / Admin@123 — change immediately in Settings).
2. Set minimum confidence in the top navbar slider (default 60%). Signals below this threshold are hidden across all pages.
3. Pick a market from the left sidebar: Crypto, Forex, Commodities, Indian Stocks, or Indices.
4. View live signals on the market page or the unified All Signals page.
5. Use Auto Generate to run the signal engine on-demand for any asset and timeframe.
6. Explore advanced tools: AI Insights, Backtesting, Risk Manager, TA Summary, MTF Analysis.

How signals are generated

The signal engine runs a 7-stage pipeline for every asset + timeframe combination:

1. Session Gate — rejects signals outside the market's active trading hours (e.g. forex during Asian session overlap, stocks outside exchange hours).
2. Volatility Gate — rejects if ATR% is below 0.1% (dead market) or above 6% (chaotic market). Labels regimes: dead / normal / elevated / chaotic.
3. Multi-Timeframe Gate — checks the next higher timeframe trend. Signals opposing the higher-TF direction are blocked.
4. Momentum Gate — validates RSI is not overbought/oversold at entry; MACD histogram must confirm direction.
5. Volume Gate — current bar volume must be ≥ 80% of the recent average (rejects low-liquidity setups).
6. Confidence Scoring — multiplicative model combining trend strength, MTF alignment, and volume confirmation → final score 0–100.
7. Signal Packaging — structure-aware stop-loss (swing high/low + ATR fallback), two profit targets (T1 = 1.5R, T2 = 2.5R).

Signal cards

• BUY / SELL badge: direction of the trade.
• Confidence %: 60–100. Higher = stronger multi-factor agreement.
• Entry / SL / T1 / T2: price levels computed at generation time.
• Volatility regime: normal / elevated / chaotic label.
• Lockout: once a signal fires, that asset+TF is locked out for a cooldown period (1m=5min, 1h=2h, 4h=8h) to prevent signal spam.

AI Insights runs a machine-learning ensemble (Random Forest + XGBoost + LightGBM) trained on 28 features derived from price action, momentum, volatility, and volume.

How to use

1. Select an asset from the dropdown.
2. Tick one or more timeframes (1m, 5m, 15m, 30m, 1h, 4h, 1d).
3. Click Run AI Analysis.
4. The right panel shows a Confluence Summary (agreement across selected TFs) and per-TF prediction cards.

Understanding results

• Bull Probability: likelihood of a price increase in the next few bars (0–100%).
• Bias: BULLISH / BEARISH / NEUTRAL (threshold ±60%).
• Confidence: model agreement score — higher means the three models broadly agree.
• Confluence: when ≥2 timeframes show the same bias, it is highlighted as a high-conviction setup.
• Models are retrained automatically when the cached model is older than 24 hours. A heuristic fallback (momentum indicators) is used if training data is insufficient.

Technical Analysis Summary provides a quick snapshot of indicator consensus across 7 timeframes (1m → 1d) for any asset.

• Oscillators: RSI, Stochastic, CCI, MACD — each rated BUY / NEUTRAL / SELL.
• Moving Averages: EMA-9/21/50/200, SMA-20/50 — trend direction vs current price.
• Overall score: weighted vote of all indicators → STRONG BUY / BUY / NEUTRAL / SELL / STRONG SELL.
• Results for each TF are computed independently; disagreement across TFs is normal and informative.

TA Summary and AI Insights use different logic — TA Summary is rule-based (fixed indicator thresholds); AI Insights is ML-based (learns from historical patterns). Getting different results is expected and useful.

Multi-Timeframe Analysis shows the alignment of trend and momentum indicators across all timeframes in a single grid.

• Each cell shows the trend direction for that asset × timeframe.
• Green row: all or most TFs agree bullish — high-confidence long setup.
• Red row: bearish alignment.
• Mixed: conflicting — avoid or wait for resolution.
• The signal engine's MTF Gate uses this same data to filter signals on execution.

The Scanner sweeps all tracked assets on a selected timeframe and ranks them by signal confidence.

• Filter by market type, minimum confidence, and direction (BUY / SELL / both).
• Results refresh automatically when new signals are generated.
• Click any row to navigate to the asset's market page for details.

Backtesting replays the signal engine on historical data and measures realistic performance.

Parameters

• Asset + Timeframe: any tracked asset.
• Start / End date: history window.
• Commission (default 0.1%): deducted on every entry and exit.
• Slippage (default 0.05%): fill-price penalty (BUY fills higher, SELL lower).
• Min Confidence: only trades from signals above this threshold are taken.

Trade mechanics

• Partial exit at T1: 50% position closed at first target; stop-loss moved to breakeven.
• Full exit at T2: remaining 50% closed at second target.
• Max hold limit: position forced-closed after N bars if neither target nor stop is hit (1m=30, 1h=10, 4h=6, 1d=5 bars).
• Volatility-scaled sizing: position size reduced in elevated/high ATR regimes (75% / 50% of normal).

Metrics

• Win Rate %, Profit Factor, Max Drawdown %
• Sharpe Ratio (annualised), Sortino Ratio (downside-only)
• Total Commission and Total Slippage paid — shows realistic drag
• Avg Bars Held, Exit Reasons breakdown (SL / T1 / T2 / timeout)

Calculates recommended position size for a trade based on your account and risk tolerance.

Standard calculator

• Inputs: Account Size, Risk %, Entry Price, Stop-Loss Price, Lot Size.
• Output: units to trade, monetary risk, R:R ratio.

Volatility-adjusted calculator

When ATR is provided, the calculator automatically applies a volatility scalar:

• Normal regime (ATR ≤ 60th percentile): 100% of calculated size.
• Elevated regime (60–80th percentile): 75% of size.
• High regime (>80th percentile): 50% of size.
• ATR percentile is calculated from the ATR history values you supply.

Log and review your actual trades with notes, emotions, and outcomes.

• Record entry/exit prices, P&L, trade setup notes, and emotional state.
• Filter by date range, asset, or outcome (win / loss / breakeven).
• Journal entries feed into My Performance for statistics.

A daily summary page generated each morning with:

• Top signals from the last 24 hours across all markets.
• Market sentiment overview (bullish / bearish / mixed).
• Key economic events for the day.
• Watchlist movers — assets on your watchlist with notable price changes.

Portfolio

Track your open and closed positions, total P&L, and allocation breakdown by market.

Watchlist

Pin assets for quick access. Watchlisted assets appear in the Morning Briefing movers section. Free accounts: up to 5 assets. Premium: up to 50.

• Min Confidence Filter: signals below this are hidden sitewide (also adjustable from the navbar slider).
• Account Size & Risk %: used as defaults in the Risk Manager calculator.
• Theme: Dark / Light — also toggleable from the navbar moon icon.
• API Keys: Binance and other data-source keys (for live data feed).
• Notification Preferences: email and push notification settings.

P0 — Signal Engine Overhaul (2026-07)

• Replaced single-indicator signal logic with a 7-stage gated pipeline (session → volatility → MTF → momentum → volume → confidence → packaging).
• Session gate blocks signal generation outside each market's active hours (UTC).
• Volatility gate rejects dead (<0.1% ATR) and chaotic (>6% ATR) markets.
• Structure-aware stop-loss using recent swing high/low; ATR fallback if swing not found.
• Per-TF signal lockout cooldowns to eliminate signal spam.
• DB indexes added on signals, signal_history, notifications, audit_logs, system_logs, journal_entries.
• Nightly cleanup job (02:00 UTC): prunes system logs >7 days, signal history >60 days, closed signals >30 days.

P1 — Market Data Infrastructure (2026-07)

• Central Market Data Collector: single background job (30s interval) fetches OHLCV for all assets into a shared in-memory cache. All signal generation reads from cache — no direct API calls during signal processing.
• Dropped DB tables: market_data and technical_indicators removed. Data was never used from DB; served from API cache directly. Reduces DB size and write load.
• Retry decorator: 3 attempts with 1.5× exponential backoff on all external API calls.
• Circuit breakers: Binance and Yahoo fetchers open after 5 consecutive failures; probe recovery after 120s. Prevents cascade failures.
• Default OHLCV candle limit reduced 300 → 220 (covers all indicators with margin).

P2 — AI & Risk Improvements (2026-07)

• AI model persistence: trained models saved to data/models/ via joblib. 24h TTL before retraining. Eliminates cold-start retraining on every restart.
• Expanded feature set: 28 features including lagged returns, RSI delta, MACD histogram cross, ATR ratio, Bollinger Band position, OBV change, body/wick ratios.
• Volatility-adjusted position sizing: ATR percentile determines a volatility scalar (1.0 / 0.75 / 0.50) applied to all position size calculations.
• Circuit breakers and retry logic extended to Yahoo Finance fetcher.

P3 — Realistic Backtesting (2026-07)

• Commission (default 0.1%) and slippage (default 0.05%) deducted on every entry and exit.
• Partial exit mechanic: 50% closed at T1, stop moved to breakeven, 50% runs to T2.
• Max-hold-bars timeout per timeframe to avoid indefinitely open positions.
• Sortino Ratio added (downside-only volatility penalty, more relevant to traders than Sharpe).
• Exit reason tracking: SL / T1 / T2 / reversal / timeout counts per backtest run.
• Commission and slippage totals reported so you can see realistic cost drag.