Skip to main content

Overview

Deltalytix includes an advanced AI trading coach that analyzes your trading data to provide personalized insights, identify patterns, and offer recommendations. Built on large language models with specialized trading knowledge, the AI understands your performance and helps you improve.

AI Chat Widget

Natural Conversation

Ask questions in plain English about your trading. The AI understands context and provides detailed answers.

Data-Driven Analysis

AI analyzes your actual trades, not generic advice. All insights are based on your real performance data.

Visual Tools

AI can generate custom equity curves, charts, and visualizations to illustrate insights.

Daily Conversations

Conversations are saved per day and persist. Return anytime to continue where you left off.

Getting Started with AI

From dashboard/components/chat/chat.tsx:
1

Add Chat Widget

Add the AI Chat Widget to your dashboard. It’s available in the “Other” category of the widget menu.
2

Start Conversation

Click Start Chat on the welcome overlay. The AI will greet you and offer to help analyze your trading.
3

Ask Your First Question

Try asking:
  • “What’s my overall performance?”
  • “Show me my equity curve”
  • “What are my best and worst trading days?”
  • “How do I compare to last month?”
4

Follow-Up Questions

The AI remembers context. Ask follow-up questions without repeating information:
  • “What about last week?”
  • “Show me just my short trades”
  • “Why is that happening?”

AI Capabilities

Performance Analysis

The AI can analyze your trading statistics and provide insights: Questions to Ask:
  • “What’s my win rate and profit factor?”
  • “Am I improving over time?”
  • “What’s my average trade duration?”
  • “Show me my biggest winners and losers”
  • “How consistent am I?”
AI Response Example:
“Your overall win rate is 62% with a profit factor of 1.8, which indicates strong profitability. Your average winning trade is 250whileyouraverageloseris250 while your average loser is 140, giving you a good risk/reward ratio of 1.78:1. However, I notice your win rate drops to 45% after 2pm - we should explore why your afternoon performance differs.”

Pattern Recognition

The AI identifies patterns in your trading behavior: Pattern Questions:
  • “Do I trade better at certain times of day?”
  • “Which days of the week are my best?”
  • “Am I better at long or short positions?”
  • “Do my tagged ‘A+’ setups actually perform better?”
  • “What patterns do you see in my losing trades?”
AI Pattern Detection:
  • Time-of-day performance variations
  • Weekday trends
  • Instrument-specific edges
  • Tag/strategy performance differences
  • Hold time vs profitability correlations
  • Emotional state impact (from mindset journal)
The AI cross-references your mindset journal entries with trade performance to identify psychological patterns affecting your results.

Equity Curve Generation

From dashboard/components/chat/chat.tsx (lines 560-621), the AI has a built-in tool to generate custom equity curves: Request Examples:
  • “Show me my equity curve for the last 30 days”
  • “Generate an equity curve for my ES trades only”
  • “Create an equity curve comparing my two accounts”
  • “Show my equity with and without my worst trades”
Generated Equity Chart Includes:
interface EquityChartData {
  chartData: Array<{
    date: string
    equity: number
    dailyPnL: number
  }>
  accountNumbers: string[]
  showIndividual: boolean
  dateRange: { start: Date, end: Date }
  totalTrades: number
}
The chart appears inline in the chat conversation with:
  • Interactive tooltips
  • Zoom and pan capabilities
  • Daily P&L markers
  • Account comparison overlays
  • Drawdown highlighting

Strategic Recommendations

The AI provides actionable recommendations based on your data: Recommendation Types:
  1. Time Management
    • “Consider avoiding trading between 11am-1pm when your win rate drops to 40%”
    • “Your best performance is 9:30am-11am EST. Focus your energy there.”
  2. Position Sizing
    • “Your largest losses occur when trading >5 contracts. Consider reducing size on high-conviction trades.”
    • “You’re leaving money on the table - winning trades average 1.2x your target. Consider wider profit targets.”
  3. Strategy Refinement
    • “Your ‘breakout’ tagged trades have a 70% win rate vs 58% overall. Trade more breakouts.”
    • “Short trades during trending markets show better performance than long trades. Adapt to market conditions.”
  4. Risk Management
    • “You’re holding losers 2x longer than winners. Consider tighter stop losses or time stops.”
    • “Your worst losing streak was 7 trades. Implement a circuit breaker after 3 consecutive losses.”

Journal Analysis

The AI reads your trade comments and mindset journal: Journal Questions:
  • “What do my trade notes say about my losing trades?”
  • “Analyze my journal entries for emotional patterns”
  • “What lessons have I documented multiple times?”
  • “Show me trades where I noted ‘entered too early’”
  • “Compare my performance on days I rated emotions below 50 vs above 70”
AI Journal Insights:
  • Recurring themes in comments (FOMO, impatience, overconfidence)
  • Correlation between documented emotions and results
  • Rule breaks mentioned in notes
  • Setup quality vs actual performance
  • Words/phrases associated with winning vs losing trades

Advanced AI Features

Multi-Turn Reasoning

From chat.tsx (lines 461-470), the AI uses step-by-step reasoning:
case "step-start":
  return <ThinkingMessage key={message.id} />
When processing complex queries, you’ll see:
  • “Thinking…” indicator
  • Step-by-step breakdown of analysis
  • Tool execution messages (“Generating equity chart…”)
  • Final synthesized answer
Complex Question Example:
User: “Compare my performance on trending days vs ranging days, but only for my ES trades in the morning session, and show me if my win rate differs.”
The AI will:
  1. Filter trades by instrument (ES)
  2. Filter by time (morning session)
  3. Analyze market conditions from tags or price action
  4. Calculate win rates for each condition
  5. Generate comparison and explanation

Contextual Memory

The AI remembers your conversation throughout the day: Session Context:
  • Previous questions and answers
  • Charts already generated
  • Filters you’ve discussed
  • Recommendations already given
Follow-Up Examples:
  • User: “Show me my equity curve”
  • AI: [Generates chart]
  • User: “What caused that big drawdown in the middle?”
  • AI: [Analyzes the specific drawdown period from the chart]
Conversations are saved daily in your Mindset journal. You can return to previous day’s conversations to review past insights.

Image Understanding

From chat.tsx (lines 437-450), you can attach images to your questions: Upload Screenshots:
  • Chart setups you’re considering
  • Trade screenshots for post-analysis
  • Strategy diagrams
  • Questions about specific patterns
The AI can analyze images and provide feedback:
  • “Is this a good setup to trade?”
  • “What do you see in this chart?”
  • “Did I exit too early here?” [attach exit chart]

Chat Interface Features

Message Types

From chat.tsx (lines 423-651): User Messages - Your questions and uploaded content Assistant Messages - AI responses with rich formatting:
  • Markdown text
  • Code blocks
  • Lists and bullet points
  • Emphasis (bold, italic)
  • Links
Tool Execution Messages - Real-time feedback:
  • “Generating equity chart…”
  • “Analyzing 347 trades…”
  • “Calculating statistics…”
Equity Charts - Interactive visualizations generated inline Error Messages - Clear error states with retry options

Chat Controls

Reset Conversation (from chat.tsx:348-349)
  • Click Reset button to clear the current day’s conversation
  • Starts fresh conversation with AI greeting
  • Previous days’ conversations remain in history
Stop Generation (from chat.tsx:678)
  • If AI is generating a long response, click Stop to interrupt
  • Useful if the AI misunderstood your question
  • You can then rephrase and ask again
Resume Scroll Button (from chat.tsx:81-112)
  • Auto-scrolls to latest message
  • Appears when you scroll up to read history
  • Click to jump back to bottom

Mobile Experience

The chat widget is fully responsive:
  • Touch-friendly message bubbles
  • Optimized keyboard behavior
  • Swipe gestures for navigation
  • Compact chart rendering
  • Voice input support (device-dependent)

Best Practices

Be specific with questions. Instead of “How am I doing?”, ask “What’s my win rate over the last 30 days compared to my 6-month average?”
Use natural language. The AI understands conversational English. No need for rigid query syntax.
Ask ‘why’ questions. Don’t just ask for stats. Ask “Why is my afternoon performance worse?” to get insights.
Leverage context. After getting a response, ask follow-up questions without repeating information.
The AI analyzes your data but can’t see your live trading or screen. It won’t know about trades you haven’t imported into Deltalytix.

Example Conversation Flow

User: What's my overall performance this month?

AI: This month you've taken 47 trades with a 64% win rate and +$2,340 net profit. 
Your profit factor is 2.1, indicating strong profitability. Would you like me to 
break this down by day or instrument?

User: By day please

AI: [Generates daily P&L breakdown]
Your best days are Wednesdays ($680 avg) and Thursdays ($520 avg). Mondays show 
the weakest performance at only $120 average. 

User: Why are Mondays weak?

AI: Analyzing your Monday trades, I see:
1. Lower win rate (52% vs 64% overall)
2. Slightly larger average losses ($180 vs $140)
3. Your journal entries mention "choppy open" 4 times on Mondays

Consider waiting 15-30 minutes after the open on Mondays to let the market establish
a clearer direction before entering your first trade.

User: Good idea. Show me my equity curve excluding Mondays

AI: [Generates filtered equity curve]
Excluding Mondays, your equity curve is smoother with fewer drawdowns. Your profit
would be $2,100 (vs $2,340 with Mondays), but your win rate jumps to 68%. This 
suggests skipping Mondays might improve consistency at minimal profit cost.

Privacy & Data Security

Your data privacy:
  • AI queries are processed securely via encrypted API calls
  • Trade data is never used to train public AI models
  • Conversations are private to your account
  • You can delete chat history anytime by resetting conversations
  • No trade data is shared with third parties

Limitations

AI Limitations to Keep in Mind:
  1. No Live Market Access - AI can’t see current market conditions or provide real-time trade signals
  2. Historical Analysis Only - Insights are based on your past trades, not predictive of future results
  3. Data Quality Dependent - Garbage in, garbage out. Incomplete or inaccurate trade data leads to poor insights
  4. No Account Management - AI can’t execute trades, modify positions, or access your broker accounts
  5. English Language - Currently optimized for English queries (French support improving)

Example Questions to Ask

Getting Started:
  • “Give me an overview of my trading performance”
  • “What are my key statistics?”
  • “Show me my equity curve”
Pattern Discovery:
  • “When do I trade best?”
  • “Which setups have the highest win rate?”
  • “Do I perform differently on trending vs ranging days?”
  • “Compare my last 30 days to my last 90 days”
Problem Solving:
  • “Why am I in a losing streak?”
  • “What’s causing my drawdown?”
  • “How can I improve my win rate?”
  • “Should I trade more or less?”
Strategy Validation:
  • “Is my ‘breakout’ strategy working?”
  • “Are trades tagged ‘A+’ actually better?”
  • “Should I trade both longs and shorts, or focus on one direction?”
  • “What’s my optimal position size?”
Advanced Analysis:
  • “Show me correlation between my emotion ratings and P&L”
  • “Analyze my journal entries for recurring mistakes”
  • “Compare my performance across different accounts”
  • “Generate an equity curve for only my winning strategies”

Next Steps

Trade Journal

Document trades with rich notes for AI to analyze

Performance Analytics

Understand the metrics and patterns AI references

Build docs developers (and LLMs) love