Backtest Strategy

Type: backtest_strategy • Category: flow • Tags: backtest, trading, strategy, finance, risk, walk-forward, optimization

Description

High-performance strategy backtesting with comprehensive statistics

Parameters

Name	Type	Description	Required	Default
data	string	OHLC data from upstream worker (e.g., {{workers[1].[result]}})	yes
signals	string	Trading signals with time and action (long/short/close)	yes
initial_capital	number	Starting capital	no	10000
commission	number	Commission per trade (0.001 = 0.1%)	no	0.001
slippage	number	Slippage (0.0005 = 0.05%)	no	0.0005
time_field	string	Timestamp field name. Supports Unix timestamps (seconds/ms), ISO strings, or datetime objects.	no	"time"
open_field	string	Open price field name. Used for entry price when signal triggers at bar open.	no	"open"
high_field	string	High price field name. Used for take profit hit detection - the highest price in each bar determines if TP was triggered.	no	"high"
low_field	string	Low price field name. Used for stop loss hit detection - the lowest price in each bar determines if SL was triggered.	no	"low"
close_field	string	Close price field name. Used for position valuation, trailing stop updates, and exits at bar close.	no	"close"
ai_optimize	boolean	🤖 AI Find Best Strategy - Automatically analyze data volatility and test all parameter combinations to find optimal settings. No configuration needed!	no	false
ai_optimize_target	string	Metric to optimize for	no	"sharpe_ratio"
ai_sl_tp_type	string	SL/TP calculation type: percent (% of price), fixed (price units/pips), atr (ATR multiples)	no	"percent"
ai_show_trades_for	number	Show detailed trades for which result (1-N). Change this to see trades for different top results.	no	1
ai_top_results_count	number	How many top results to include in output (1-50). More results = more data to compare.	no	10
position_size_type	string	How to calculate position size	no	"percent"
position_size	number	Position size value (0.1 = 10% for percent type)	no	0.1
stop_loss_type	string	Stop loss calculation method	no	"percent"
stop_loss_value	number	Stop loss value (2.0 = 2% for percent type)	no	2
take_profit_type	string	Take profit calculation method	no	"percent"
take_profit_value	number	Take profit value (4.0 = 4% for percent type, or 3.0 = 1:3 RR)	no	4
trailing_stop	boolean	Enable trailing stop	no	false
trailing_stop_value	number	Trailing stop distance in %	no	1
analysis_blocks	number	Split data into N blocks for consistency analysis (0=disabled, recommended: 4, 6, or 12)	no	0
max_hold_duration	string	Max position hold time (e.g., '30m', '2h', '1d'). Leave empty to disable.	no	""
close_at_time	string	Close all positions at specific time daily (e.g., '16:00', '8pm'). Leave empty to disable.	no	""
close_at_time_offset	number	UTC offset in hours for time settings (e.g., -5 for EST, 1 for CET)	no	0
trading_window_start	string	Only open positions after this time (e.g., '9:30', '9am'). Leave empty for no restriction.	no	""
trading_window_end	string	Only open positions before this time (e.g., '15:30', '3pm'). Leave empty for no restriction.	no	""
optimize_sl	string	Stop loss range to optimize (e.g., '1-5:0.5' or '1,2,3,4'). Leave empty for no optimization.	no	""
optimize_tp	string	Take profit range to optimize (e.g., '2-10:1' or '2,4,6,8'). Leave empty for no optimization.	no	""
optimize_size	string	Position size range to optimize (e.g., '0.05-0.2:0.05'). Leave empty for no optimization.	no	""
optimize_target	string	Metric to optimize for when running parameter optimization	no	"total_return"

Help

Backtest Strategy

A high-performance, license-free backtesting engine for evaluating trading strategies. Built with pure numpy for maximum speed - no GPL dependencies (unlike vectorbt/backtrader).

---

⚡ Key Advantages

Performance

• Vectorized numpy operations - processes 100k+ bars in milliseconds
• O(1) signal lookup using hashmap for instant bar matching
• Memory optimized with __slots__ and contiguous arrays
• No external dependencies - only numpy (BSD) and pandas (BSD)

License-Safe

• 100% BSD/MIT licensed - safe for commercial use
• No GPL contamination - unlike vectorbt, backtrader, zipline
• No attribution required - use freely in paid products

Unique Features

• Block Analysis - Walk-forward validation with consistency scoring
• Trading Window - Restrict entry times (market hours only)
• Time-Based Exits - Auto-close at specific times or after duration
• Flexible Field Mapping - Works with any OHLC data format

Complete Risk Management

• Multiple position sizing methods (percent, fixed, risk-based)
• Stop loss & take profit (percent, ATR, fixed, R:R ratio)
• Trailing stops with customizable distance
• Commission and slippage modeling

Professional Statistics

• Risk-adjusted returns (Sharpe, Sortino, Calmar)
• Drawdown analysis (max DD, duration, recovery)
• Trade breakdown by direction and exit reason
• Equity & drawdown curves for visualization

---

📥 Inputs

OHLC Data

Connect data with time, open, high, low, close fields. Sources: postgres_sql, fetch_ohlc, fetch_csv, or any worker returning OHLC data.

Use the field dropdowns to select the correct columns from your data:

• time_field: Timestamp column (Unix seconds/ms, ISO string, or datetime)
• open_field: Entry price when signal triggers at bar open
• high_field: Used for take profit hit detection (highest price in bar)
• low_field: Used for stop loss hit detection (lowest price in bar)
• close_field: Position valuation, trailing stops, and bar close exits

Signals

List of trading signals with:

• time: Timestamp (Unix, ISO string, or datetime)
• action: "long", "short", or "close"
• size (optional): Override position size for this signal

---

💰 Position Sizing

Type	Description	Example
percent	Use X% of capital per trade	0.1 = 10% of capital
fixed	Fixed dollar amount	1000 = $1000 per trade
risk_based	Size based on stop loss distance	0.02 = risk 2% per trade

Risk-based sizing automatically calculates position size so that if stopped out, you lose exactly X% of capital. Professional money management!

---

🛡️ Risk Management

Stop Loss Types

Type	Description	Example
percent	Distance as % of entry price	2.0 = 2% below entry
atr	Multiple of ATR(14)	1.5 = 1.5x ATR
fixed	Fixed price distance	5.0 = $5 from entry

Take Profit Types

Type	Description	Example
percent	Distance as % of entry price	4.0 = 4% above entry
atr	Multiple of ATR(14)	3.0 = 3x ATR
rr_ratio	Risk:Reward ratio	3.0 = 1:3 (3x stop distance)
fixed	Fixed price distance	10.0 = $10 from entry

Trailing Stop

When enabled, stop loss follows price movement:

• Long: Stop moves up as price rises (never down)
• Short: Stop moves down as price falls (never up)
• Distance: trailing_stop_value in percent

---

⏰ Time-Based Rules

Max Hold Duration

Close positions after a specific time period:

max_hold_duration: "30m"  → 30 minutes
max_hold_duration: "2h"   → 2 hours
max_hold_duration: "1d"   → 1 day

Use cases: Scalping, momentum strategies, avoiding overnight gaps

Close At Time

Close all positions at a specific time daily:

close_at_time: "16:00"           → 4:00 PM
close_at_time: "8pm"             → 8:00 PM
close_at_time_offset: -5         → EST timezone

Use cases: Day trading, avoiding after-hours, market close exits

Trading Window

Only allow new positions during specific hours:

trading_window_start: "9:30"     → Market open
trading_window_end: "15:30"      → 30 min before close
close_at_time_offset: -5         → EST timezone

Examples:

Market	Start	End	Offset
US Stocks	9:30	16:00	-5
US Futures	18:00	17:00	-5
London Forex	8:00	16:00	0
Asian Session	0:00	9:00	9

Note: Overnight windows are supported (end < start).

---

🤖 AI Find Best Strategy

One-click optimization! Let AI automatically find the best parameters for your strategy.

How to Use

1. Configure only: data, signals, OHLC field mappings, initial_capital, commission, slippage
2. Check the "🤖 AI Find Best Strategy" checkbox
3. Select your optimization target (sharpe_ratio recommended)
4. Run the backtest

That's it! All other parameters will be hidden - AI determines everything automatically!

What AI Does

1. Analyzes your data's volatility:

   • Average bar price change %
   • 95th percentile price change
   • Detects timeframe (tick/intraday/hourly/daily)
   • Adjusts parameters based on volatility

2. Auto-determines parameter types:

   • Stop Loss Type: percent (universal)
   • Take Profit Type: percent (universal)
   • Position Size Type: percent (universal)

3. Calculates smart ranges based on volatility:

   Parameter	How Range is Calculated
   Stop Loss	0.5x to 3x average bar change
   Take Profit	1x to 5x average bar change
   Position Size	5% - 25% of capital
   Trailing Stop	Disabled, tight, normal, wide (based on volatility)
   Max Hold Duration	Based on detected timeframe (5min-2weeks)

4. Tests all combinations (up to 600) and finds optimal settings

5. 🔄 Recursive Search - If best result is unprofitable, AI automatically:

   • Expands parameter ranges (tighter AND wider)
   • Increases granularity (more test values)
   • Tries up to 3 additional searches
   • Compares results and returns the best

6. Runs full backtest with optimal parameters to get detailed trade data for charting

🔄 Recursive Search (Auto-Retry)

If the best strategy found is losing money or has negative Sharpe ratio, AI doesn't give up! It automatically expands the search with wider parameter ranges:

Attempt	SL/TP Range Expansion	Test Values
Initial	1.0x (baseline)	5-7 values
Retry 1	1.5x wider + tighter	7-9 values
Retry 2	2.0x wider + tighter	9-11 values
Retry 3	2.5x wider + tighter	11-13 values

Trigger conditions:

• total_return < 0% (losing money)
• sharpe_ratio < 0 (negative risk-adjusted return)
• sharpe_ratio < 0.3 AND return < 2% (poor results)

Output includes:

"recursive_search": {
  "performed": true,
  "depth": 2,
  "max_depth": 3,
  "reason": "Searched 2 additional time(s) for better parameters",
  "improved": true
}

Recommendations will show:

• 🔄 AI performed 2 additional search(es) with expanded parameter ranges
• ⚠️ Strategy remains unprofitable even after exhaustive search. Consider revising signal logic.

🔍 Verify Top 10 Results

Want to verify that calculations are correct for any of the top 10 results? Use the ai_show_trades_for parameter!

How to use:

1. Run AI optimization normally
2. Look at top_10 array - you'll see 10 best results with different parameters
3. Change ai_show_trades_for to any number 1-10
4. Re-run the worker
5. Now trades, trades_summary, equity_curve, drawdown_curve will show data for that specific result

Example workflow:

Step 1: ai_show_trades_for = 1 (default)
        → See trades for best result (#1)
        
Step 2: ai_show_trades_for = 5
        → See trades for 5th best result
        → Compare equity curve, verify calculations
        
Step 3: ai_show_trades_for = 10
        → See trades for 10th result
        → Understand why it ranked lower

Output includes info about which result is shown:

"showing_trades_for": {
  "rank": 5,
  "parameters": {
    "stop_loss": 1.2,
    "take_profit": 3.6,
    "position_size": 0.15,
    "trailing_stop": true
  },
  "results": {
    "total_return_pct": 8.5,
    "sharpe_ratio": 0.85,
    "win_rate": 52.3,
    "total_trades": 127
  },
  "note": "Change 'ai_show_trades_for' parameter (1-10) to see trades for different top results"
}

This allows you to:

• ✅ Verify calculations are correct for each strategy
• ✅ Compare equity curves between different parameter sets
• ✅ Visualize trades on chart for any top 10 result
• ✅ Understand why certain parameters rank higher/lower

AI Output

• data_analysis: Full volatility analysis with detected timeframe
• auto_determined_types: Explains why AI chose percent types
• best_parameters: Optimal settings ready to use (copy-paste!)
• best_results: Summary statistics for optimal configuration
  • initial_capital: Starting capital
  • final_capital: Ending capital after all trades
  • total_return_pct: Percentage return
  • total_return_value: Absolute return in currency
  • max_drawdown_pct: Maximum drawdown as percentage
  • max_drawdown_value: Maximum drawdown in currency
• recommendations: Actionable insights about your strategy
• top_10: Top 10 performing parameter combinations (with all metrics)
• trades: Full list of all trades with entry/exit details for charting
• trades_summary: Breakdown of trades by type and exit reason
• equity_curve: Equity over time for charting
• drawdown_curve: Drawdown over time for charting
• ranges_tested: Shows what ranges AI tested based on volatility
• statistics: Distribution of results across all tests
• recursive_search: Info about recursive search attempts

Trade Details (for charting)

Each trade in the trades array includes:

Field	Description
entry_time	When position was opened (ISO timestamp)
exit_time	When position was closed (ISO timestamp)
direction	long or short
entry_price	Price at entry
exit_price	Price at exit
stop_loss	Stop loss price level
take_profit	Take profit price level
size	Position size / lots
profit	Profit/loss in currency
profit_pct	Profit/loss as percentage
exit_reason	stop_loss, take_profit, signal, trailing_stop, max_duration
bars_held	Number of bars position was held
commission	Commission paid for this trade

Example Output

{
  "ai_optimization": true,
  "message": "🤖 AI analyzed 280 combinations based on your data's volatility...",
  "best_parameters": {
    "stop_loss_value": 0.9,
    "take_profit_value": 2.7,
    "position_size": 0.15,
    "trailing_stop": true,
    "rr_ratio": 3.0
  },
  "trades": [
    {
      "entry_time": "2024-01-15T09:30:00+00:00",
      "exit_time": "2024-01-15T14:22:00+00:00",
      "direction": "long",
      "entry_price": 150.25,
      "exit_price": 154.31,
      "stop_loss": 148.90,
      "take_profit": 154.31,
      "size": 66,
      "profit": 267.96,
      "profit_pct": 2.7,
      "exit_reason": "take_profit",
      "bars_held": 58
    }
  ],
  "trades_summary": {
    "total_trades": 127,
    "winning_trades": 74,
    "losing_trades": 53,
    "exits_by_tp": 48,
    "exits_by_sl": 45,
    "exits_by_trailing": 26
  },
  "equity_curve": [...],
  "drawdown_curve": [...]
}

---

🔍 Manual Parameter Optimization

For advanced users who want full control over parameter ranges.

How to Use

Fill in the optimization range fields:

• optimize_sl: Stop loss values to test
• optimize_tp: Take profit values to test
• optimize_size: Position size values to test (optional)

Range Format

Format	Example	Result
Single value	2	[2.0]
List	1,2,3,4	[1.0, 2.0, 3.0, 4.0]
Range	1-5	[1.0, 2.0, 3.0, 4.0, 5.0]
Range with step	1-5:0.5	[1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0]

Examples

Test SL from 1% to 3%, TP from 2% to 6%:

optimize_sl: "1-3:0.5"     → [1.0, 1.5, 2.0, 2.5, 3.0]
optimize_tp: "2-6:1"       → [2.0, 3.0, 4.0, 5.0, 6.0]
optimize_target: "sharpe_ratio"

This tests 5 × 5 = 25 combinations.

Test specific values:

optimize_sl: "1,1.5,2,2.5,3"
optimize_tp: "3,4,5,6"
optimize_size: "0.05,0.1,0.15,0.2"

This tests 5 × 4 × 4 = 80 combinations.

Optimization Targets

Target	Description
total_return	(Default) Maximum profit
sharpe_ratio	Best risk-adjusted return
profit_factor	Highest gross profit / gross loss
win_rate	Highest winning percentage
expectancy	Best expected value per trade
sortino_ratio	Best downside risk-adjusted return

Output

The optimization returns:

• best_parameters: Optimal SL, TP, position size, and R:R ratio
• best_results: Performance metrics for best combination
• top_10: Top 10 performing combinations
• heatmap: Data for SL vs TP heatmap visualization
• statistics: Distribution of results across all tests

Limits

Maximum 1000 combinations per optimization to prevent excessive computation.

---

📊 Block Analysis (Walk-Forward Validation)

Divide your backtest into N blocks to validate strategy consistency. This helps detect overfitting - a robust strategy should perform similarly across all periods.

analysis_blocks: 6  → Split data into 6 equal periods

Why Use Block Analysis?

• Detect overfitting: Strategy that works only in one period is likely curve-fitted
• Validate robustness: Consistent profits across all blocks = reliable strategy
• Walk-forward simulation: Each block is like a forward test

Consistency Score (0-100)

Score	Rating	Meaning
80-100	⭐ Excellent	Strategy works consistently across all periods
60-80	✅ Good	Minor variations, generally reliable
40-60	⚠️ Moderate	Notable differences, needs review
20-40	❌ Poor	Likely overfitted to specific periods
0-20	🚫 Very Poor	Strategy fails in multiple periods

Block Output

Each block includes:

• Start/end dates
• Total return & trades
• Win rate & profit factor
• Sharpe ratio

---

📈 Output Metrics

Summary Statistics

Metric	Description
total_return	Total profit/loss in currency
total_return_pct	Total return as percentage
annual_return	Annualized return
sharpe_ratio	Risk-adjusted return (volatility)
sortino_ratio	Risk-adjusted return (downside only)
calmar_ratio	Return / Max Drawdown

Risk Metrics

Metric	Description
max_drawdown	Largest peak-to-trough decline
max_drawdown_duration	Longest recovery period
volatility	Annualized standard deviation
win_rate	Percentage of winning trades
profit_factor	Gross profit / Gross loss
expectancy	Average expected profit per trade

Trade Analysis

Field	Description
total_trades	Number of completed trades
avg_win / avg_loss	Average winning/losing trade
best_trade / worst_trade	Extremes
avg_bars_held	Average holding period
statistics_by_direction	Separate stats for long/short
exit_analysis	Breakdown by exit reason

Visualizations

Output	Description
equity_curve	Capital over time (for charting)
drawdown_curve	Drawdown over time
trades	Full trade list with details

---

🔧 Complete Parameter Reference

Required

Parameter	Type	Description
data	string	OHLC data from upstream worker
signals	string	Trading signals array

Capital & Sizing

Parameter	Default	Description
initial_capital	10000	Starting capital
position_size_type	percent	percent, fixed, or risk_based
position_size	0.1	Size value (0.1 = 10% for percent)

Costs

Parameter	Default	Description
commission	0.001	Per-trade commission (0.1%)
slippage	0.0005	Slippage (0.05%)

Risk Management

Parameter	Default	Description
stop_loss_type	percent	percent, atr, or fixed
stop_loss_value	2.0	Stop loss distance
take_profit_type	percent	percent, atr, rr_ratio, or fixed
take_profit_value	4.0	Take profit distance
trailing_stop	false	Enable trailing stop
trailing_stop_value	1.0	Trailing distance (%)

Time Rules

Parameter	Default	Description
max_hold_duration	""	Max hold time (30m, 2h, 1d)
close_at_time	""	Daily close time (16:00, 8pm)
close_at_time_offset	0	UTC offset in hours
trading_window_start	""	Entry window start (9:30)
trading_window_end	""	Entry window end (15:30)

Advanced

Parameter	Default	Description
time_field	time	Timestamp field name
open_field	open	Open price field
high_field	high	High price field
low_field	low	Low price field
close_field	close	Close price field
analysis_blocks	0	Block analysis (4, 6, or 12)

AI Optimization

Parameter	Default	Description
ai_optimize	false	🤖 Enable AI auto-optimization
ai_optimize_target	sharpe_ratio	Metric for AI to optimize
ai_sl_tp_type	percent	SL/TP type: percent (%), fixed (pips/price), atr (ATR×)
ai_top_results_count	10	How many top results to include (1-50)
ai_show_trades_for	1	Show trades for which result (1-N). Change to verify different results!

Manual Optimization

Parameter	Default	Description
optimize_sl	""	SL range (e.g., "1-3:0.5" or "1,2,3")
optimize_tp	""	TP range (e.g., "2-6:1" or "2,4,6")
optimize_size	""	Position size range (e.g., "0.05-0.2:0.05")
optimize_target	total_return	Metric to maximize