Blazing-Fast Vectorized Backtesting Engine

Quick Start · Why Tradix? · Features · Installation · API Reference · 한국어

Tradix is a high-performance backtesting engine built for quantitative traders and researchers. Powered by pure NumPy vectorization, it runs 1,000 parameter optimizations in under a second. No boilerplate, no configuration files, no data pipeline setup — just install and backtest.

The library ships with 33 battle-tested strategies, 60+ technical indicators, and a built-in data feed that auto-downloads from global exchanges including KRX (Korea Exchange). Whether you're prototyping a Golden Cross on Samsung Electronics or stress-testing a multi-factor portfolio against Black Swan scenarios, Tradix handles the entire pipeline: data → signals → execution → analytics → visualization.

Two lines.backtest · show — that's the entire workflow.

fromtradiximportbacktest, goldenCrossresult=backtest("AAPL", goldenCross())
result.show()

◈ Quick Start

pip install tradix

fromtradiximportbacktest, goldenCrossresult=backtest("005930", goldenCross())
result.show()
result.show(style="bloomberg")

backtest() auto-downloads price data via FinanceDataReader, applies the strategy with realistic commission and slippage modeling, and returns a comprehensive result object. Pass any ticker symbol — US stocks, Korean stocks, ETFs, indices, and crypto are all supported out of the box.

=== Backtest Result ===
Strategy: GoldenCross
Period: 2020-01-02 ~ 2024-12-30
Initial: 10,000,000 KRW
Final: 14,230,000 KRW
Return: +42.30%
Sharpe: 1.23
Max DD: -12.45%
Trades: 18
Win Rate: 61.1%

◈ Why Tradix?

Most backtesting frameworks require hundreds of lines of boilerplate — data loaders, strategy classes, broker configs, result parsers. Tradix eliminates all of that. Here's how it compares:

Dimension Tradix Backtrader Zipline bt Zero-config backtest ✅ ❌ ❌ ❌ 2-line API ✅ ❌ ❌ ❌ Vectorized engine ✅ ❌ ❌ ✅ 1K optimizations < 1s ✅ ❌ ❌ ❌ Built-in data feed ✅ ❌ ❌ ❌ 33 preset strategies ✅ ❌ ❌ ❌ Terminal dashboard ✅ ❌ ❌ ❌ Korean market native ✅ ❌ ❌ ❌ Walk-forward analysis ✅ ❌ ❌ ❌ Strategy DNA analysis ✅ ❌ ❌ ❌

Three pillars.numpy · pandas · scipy — that's the entire foundation. No C extensions, no binary dependencies, no compilation step.

◈ Features

Core Engine

The heart of Tradix is a dual-engine architecture: an event-driven engine for complex strategies with bar-by-bar logic, and a vectorized engine for lightning-fast batch operations. Both engines share the same indicator library and produce identical results.

Feature	Description
Vectorized Engine	NumPy-powered core, 100x faster than event-driven loops
60+ Indicators	SMA, EMA, RSI, MACD, Bollinger, ATR, Ichimoku, Supertrend, StochRSI, KDJ, and more
33 Preset Strategies	Trend, momentum, oscillator, volatility, multi-indicator, buy & hold, DCA
Strategy Builder	Declarative method chaining — no subclassing needed
Walk-Forward	Built-in overfitting prevention with time-series cross-validation
Optimization	Grid search and random search with any metric
Multi-Asset	Backtest across multiple symbols with rebalancing
Realistic Simulation	Commission, slippage, fill logic, position sizing
OCO / Bracket / Trailing Stop	Advanced order types for automated stop-loss and take-profit management

Multi-Source Data Feeds

Connect to global markets without writing a single data loader. Tradix ships with three data feed backends — FinanceDataReader for Korean/US/JP equities, Yahoo Finance for global ETFs and indices, and CCXT for 100+ cryptocurrency exchanges. All feeds share the same interface and include Parquet caching for offline replay.

Feature	Description
FinanceDataReader	KRX, US, JP stocks with auto-caching
Yahoo Finance	Global equities, ETFs, indices, crypto pairs via yfinance
CCXT (100+ Exchanges)	Binance, Bybit, Upbit, Coinbase — 1m to 1w candles with auto-pagination
Parquet Cache	Offline-first with configurable expiration
Unified Interface	All feeds implement the same DataFeed API

Advanced Analytics

Go beyond simple returns. Tradix provides institutional-grade analytics that dissect every aspect of your strategy — from 12-dimensional DNA fingerprinting to Black Swan resilience scoring. Understand not just how much your strategy made, but why it works and when it might fail.

Feature	Description
Strategy DNA	12-dimensional strategy fingerprinting
Black Swan Defense	Extreme event resilience scoring (0-100)
Health Score	Overfitting risk, parameter stability diagnostics
What-If Simulator	Commission, slippage, capital sensitivity analysis
Drawdown Simulator	Historical worst-case scenario generation
Seasonality Analyzer	Monthly, weekday, quarterly pattern discovery
Correlation Matrix	Multi-strategy correlation and clustering
Trading Journal	Auto trade diary with MFE/MAE analytics
Strategy Leaderboard	Multi-strategy ranking with badge system

Innovative Analytics

Tradix pushes the boundary of what a backtesting library can do. Monte Carlo simulations with 10,000 paths estimate your ruin probability. Fractal analysis reveals whether your market is trending or mean-reverting. Regime detection identifies hidden market states so you can adapt your strategy dynamically.

Feature	Description
Monte Carlo	10K-path bootstrap with ruin probability and confidence bands
Fractal Analysis	Hurst exponent for market character classification
Regime Detector	GMM-based probabilistic regime detection with transition matrix
Information Theory	Shannon entropy, mutual information for signal quality
Portfolio Stress	6 crisis scenarios (crash, volatility spike, rate shock, etc.)

Terminal UI — TradingView-Inspired

No browser needed. Tradix renders professional trading dashboards directly in your terminal using Rich and Plotext. Choose from three display styles inspired by TradingView, Bloomberg Terminal, and hedge fund reports. The interactive Textual dashboard provides a full 5-view experience with keyboard navigation.

Feature	Description
3 Display Styles	Modern (TradingView), Bloomberg (dense 4-quadrant), Minimal (hedge fund)
14 Chart Types	Equity, drawdown, candlestick, returns, seasonality, heatmap, DNA, rolling metrics, and more
Plotly Candlestick	Interactive OHLCV chart with volume, SMA/EMA overlays, and trade markers
Rolling Metrics	Time-varying Sharpe, Sortino, and volatility visualization
Interactive Dashboard	5-view Textual app with keyboard navigation
CLI	tradix backtest, tradix chart, tradix compare, tradix optimize, tradix list

Korean Market

Tradix is the first backtesting engine built with native Korean market support. Transaction tax (0.18%), brokerage fees, and KRX stock name mapping are all built-in. You can even write your entire strategy in Korean — function names, variable names, and output are all available in 한국어.

Feature	Description
Native Support	Built-in transaction tax (0.18%), brokerage fees, KRX stock mapping
Korean API	Full Korean function names: 백테스트("삼성전자", 골든크로스())

◈ Installation

Using uv (Recommended)

uv is the fastest Python package manager.

uv init my-backtest &&cd my-backtest
uv add tradix
uv add "tradix[full]"# + Plotly, statsmodels, scikit-learn
uv add "tradix[yahoo]"# + Yahoo Finance global data
uv add "tradix[crypto]"# + CCXT 100+ crypto exchanges
uv add "tradix[tui]"# + Textual interactive dashboard

Using pip

pip install tradix # Core (NumPy + Pandas + Rich)
pip install "tradix[full]"# + Plotly, statsmodels, scikit-learn
pip install "tradix[yahoo]"# + Yahoo Finance global data
pip install "tradix[crypto]"# + CCXT 100+ crypto exchanges
pip install "tradix[tui]"# + Textual interactive dashboard

From source

git clone https://github.com/eddmpython/tradix.git
cd tradix
uv sync --dev
uv run pytest

Requirements: Python 3.9+ · NumPy · Pandas · Rich · Plotext · Typer

◈ Usage

Tradix offers multiple levels of abstraction. Start with the 2-line easy API, graduate to the declarative strategy builder for custom logic, or drop down to the full event-driven engine when you need bar-by-bar control.

Declarative Strategy Builder

fromtradiximportQuickStrategy, backtest, sma, rsi, crossover, crossunderstrategy= (
QuickStrategy("MomentumRSI")
.buyWhen(crossover(sma(10), sma(30)))
.buyWhen(rsi(14) <30)
.sellWhen(crossunder(sma(10), sma(30)))
.sellWhen(rsi(14) >70)
.stopLoss(5)
.takeProfit(15)
)
result=backtest("AAPL", strategy)
result.show()

Parameter Optimization

Find optimal strategy parameters by sweeping across a parameter grid. Tradix runs all combinations using the vectorized engine — 1,000 combinations finish in under a second.

fromtradiximportvoptimizebest=voptimize(
"005930",
"goldenCross",
fast=(5, 20, 5),
slow=(20, 60, 10),
metric="sharpeRatio"
)
print(f"Best: fast={best['best']['params']['fast']}, slow={best['best']['params']['slow']}")
print(f"Sharpe: {best['best']['metric']:.2f}")

Vectorized Mode (100x Faster)

Skip the event loop entirely. The vectorized engine processes the entire price history as NumPy arrays, delivering results in microseconds.

fromtradiximportvbacktestresult=vbacktest("005930", "goldenCross", fast=10, slow=30)
print(f"Return: {result.totalReturn:+.2f}%")
print(f"Sharpe: {result.sharpeRatio:.2f}")
print(f"Max DD: {result.maxDrawdown:.2f}%")

Korean API (한국어 API)

Every function, every strategy, every output — available in native Korean. No wrappers, no translation layers. First-class Korean market citizen.

fromtradiximport백테스트, 골든크로스결과=백테스트("삼성전자", 골든크로스())
결과.보기()
결과.차트()

More Examples — Class-based strategy, walk-forward, multi-asset, risk

Class-Based Strategy

fromtradiximportStrategy, Bar, BacktestEnginefromtradix.datafeedimportFinanceDataReaderFeedclassDualMomentum(Strategy):
definitialize(self):
self.fastPeriod=10self.slowPeriod=30defonBar(self, bar: Bar):
fast=self.sma(self.fastPeriod)
slow=self.sma(self.slowPeriod)
iffastisNoneorslowisNone:
returniffast>slowandnotself.hasPosition(bar.symbol):
self.buy(bar.symbol)
eliffast<slowandself.hasPosition(bar.symbol):
self.closePosition(bar.symbol)
data=FinanceDataReaderFeed("005930", "2020-01-01", "2024-12-31")
engine=BacktestEngine(data, DualMomentum(), initialCash=10_000_000)
result=engine.run()
result.show()

Walk-Forward Analysis

fromtradiximportStrategy, Bar, WalkForwardAnalyzer, ParameterSpacefromtradix.datafeedimportFinanceDataReaderFeeddefcreateStrategy(params):
classMySma(Strategy):
definitialize(self):
self.fast=params["fast"]
self.slow=params["slow"]
defonBar(self, bar: Bar):
f, s=self.sma(self.fast), self.sma(self.slow)
iffands:
iff>sandnotself.hasPosition(bar.symbol):
self.buy(bar.symbol)
eliff<sandself.hasPosition(bar.symbol):
self.closePosition(bar.symbol)
returnMySma()
data=FinanceDataReaderFeed("005930", "2020-01-01", "2024-12-31")
space=ParameterSpace()
space.addInt("fast", 5, 20, step=5)
space.addInt("slow", 20, 60, step=10)
wfa=WalkForwardAnalyzer(
data=data,
strategyFactory=createStrategy,
parameterSpace=space,
inSampleMonths=12,
outOfSampleMonths=3,
)
result=wfa.run()
print(f"Robustness: {result.robustnessRatio:.1%}")

Multi-Asset Portfolio

fromtradiximportMultiAssetEngine, MultiAssetStrategyclassEqualWeight(MultiAssetStrategy):
defonBars(self, bars):
self.rebalance({
"005930": 0.4,
"000660": 0.3,
"035420": 0.3,
})
engine=MultiAssetEngine(strategy=EqualWeight())
result=engine.run()
result.show()

Risk Simulation

importpandasaspdfromtradiximportbacktest, goldenCrossfromtradix.riskimportRiskSimulator, VaRMethodresult=backtest("005930", goldenCross())
returns=pd.DataFrame({"returns": pd.Series(result.equityCurve).pct_change().dropna()})
simulator=RiskSimulator()
simulator.fit(returns)
var95, cvar95=simulator.calcVaR(confidence=0.95, method=VaRMethod.HISTORICAL)
print(f"95% VaR: {var95:.2%}")
print(f"CVaR: {cvar95:.2%}")
mc=simulator.monteCarloSimulation(horizon=252, nSim=10000)

◈ Strategies

33 production-ready strategies organized by trading style. Each strategy is a callable that returns a configured strategy object — no subclassing, no boilerplate. Pass parameters to customize, or use the defaults.

Trend Following (9)

Strategy	Description
goldenCross()	SMA crossover (fast/slow)
emaCross()	EMA crossover
tripleEma()	Triple EMA crossover
trendFollowing()	ADX-filtered trend following with trailing stop
superTrend()	Supertrend indicator reversal
ichimokuCloud()	Ichimoku cloud breakout
parabolicSar()	Parabolic SAR reversal
donchianBreakout()	Donchian channel breakout
breakout()	Channel breakout (Turtle Trading)

Momentum & Oscillator (8)

Strategy	Description
rsiOversold()	RSI reversal (oversold/overbought)
macdCross()	MACD histogram crossover
stochasticCross()	Stochastic K/D crossover
williamsReversal()	Williams %R reversal
cciBreakout()	CCI overbought/oversold breakout
rsiDivergence()	RSI divergence detection
momentumCross()	Momentum zero-line crossover
rocBreakout()	Rate of Change breakout

Volatility (5)

Strategy	Description
bollingerBreakout()	Bollinger band breakout
bollingerSqueeze()	Bollinger squeeze expansion
keltnerChannel()	Keltner channel breakout
volatilityBreakout()	ATR-based volatility breakout
meanReversion()	Bollinger mean reversion

Multi-Indicator (5)

Strategy	Description
tripleScreen()	Elder's triple screen system
dualMomentum()	Absolute + relative momentum
macdRsiCombo()	MACD + RSI combined signal
trendMomentum()	Trend + momentum filter
bollingerRsi()	Bollinger + RSI combined

Special (6)

Strategy	Description
gapTrading()	Gap up/down trading
pyramiding()	Pyramiding position building
swingTrading()	Swing high/low trading
scalpingMomentum()	Short-term momentum scalping
buyAndHold()	Passive buy and hold
dollarCostAverage()	Dollar cost averaging

◈ Indicators

60+ technical indicators implemented in pure NumPy. Each indicator is available as both a standalone function and a strategy-builder condition. All calculations are vectorized — even complex indicators like Ichimoku Cloud run in microseconds.

Category	Indicators
Moving Averages	smaemawmahmatemademavwmaalma
Momentum	rsimacdstochasticrocmomentumcciwilliamsRcmostochasticRsikdjawesomeOscillatorultimateOscillator
Volatility	atrbollingerkeltnerdonchianbollingerPercentBbollingerWidth
Volume	obvvwapmfiadlchaikinemvforceIndexnvipvivrocpvtklingerOscillator
Trend	adxsupertrendpsarichimokutrixdpolinearRegression
Price	pivotPointsfibonacciRetracementzigzagelderRaytwap
Other	ulcerpercentChangehighestlowest

◈ Performance

Tradix is built for speed. The vectorized engine processes entire price histories as contiguous NumPy arrays, eliminating Python loop overhead. Benchmarked on 10 years of daily data (2,458 bars):

Operation	Time
SMA calculation	0.006ms
RSI calculation	0.009ms
MACD calculation	0.040ms
Full backtest (single)	0.132ms
1,000 param optimization	0.02s

◈ Terminal UI

Professional trading dashboards rendered directly in your terminal. No browser, no GUI toolkit, no Jupyter dependency. Tradix uses Rich for styled tables and panels, Plotext for ASCII charts, and optionally Textual for a full interactive experience.

Display Styles

result.show() # Modern — TradingView metric cardsresult.show(style="bloomberg") # Bloomberg — dense 4-quadrant layoutresult.show(style="minimal") # Minimal — clean hedge fund report

CLI

tradix backtest AAPL -s goldenCross --dashboard
tradix backtest AAPL -s bollingerSqueeze --style bloomberg
tradix chart AAPL -n 60
tradix compare AAPL -s goldenCross,rsiOversold
tradix optimize AAPL -s goldenCross
tradix list

Charts (12 Types)

fromtradix.tui.chartsimport (
plotEquityCurve, plotDrawdown, plotCandlestick, plotReturns,
plotSeasonality, plotMonthlyHeatmap, plotRollingMetrics,
plotTradeScatter, plotTradeMarkers, plotCorrelationBars,
plotStrategyDna, plotDashboard,
)
plotEquityCurve(result, smaPeriods=[20, 60])
plotDrawdown(result)
plotCandlestick(df, smaPeriods=[5, 20])
plotDashboard(result, lang="ko")

Interactive Dashboard

pip install tradix[tui]

fromtradix.tui.dashboardimportlaunchDashboardlaunchDashboard(result)

◈ Advanced Analytics

Tradix includes a full suite of institutional-grade analytics. Each analyzer takes a backtest result and returns a structured report with scores, metrics, and actionable insights.

fromtradiximport (
StrategyDnaAnalyzer, BlackSwanAnalyzer, StrategyHealthAnalyzer,
WhatIfSimulator, DrawdownSimulator, SeasonalityAnalyzer,
CorrelationAnalyzer, TradingJournal, StrategyLeaderboard,
)
result=backtest("005930", goldenCross())
dna=StrategyDnaAnalyzer().analyze(result)
printStrategyDna(dna)
health=StrategyHealthAnalyzer().analyze(result)
printHealthScore(health)
blackSwan=BlackSwanAnalyzer().analyze(result)
printBlackSwanScore(blackSwan)

fromtradiximport (
MonteCarloStressAnalyzer, FractalAnalyzer, RegimeDetector,
InformationTheoryAnalyzer, PortfolioStressAnalyzer,
)
mc=MonteCarloStressAnalyzer().analyze(result, paths=10000)
print(f"Ruin probability: {mc.ruinProbability:.2%}")
fractal=FractalAnalyzer().analyze(result)
print(f"Hurst: {fractal.hurstExponent:.3f} → {fractal.marketCharacter}")
regime=RegimeDetector().analyze(result)
print(f"Current regime: {regime.currentRegime}")

◈ API Reference

Core Functions

Function	Description
backtest(symbol, strategy)	Run a backtest with a preset or custom strategy
vbacktest(symbol, strategy, **params)	Run a vectorized backtest
voptimize(symbol, strategy, **ranges)	Grid search parameter optimization
BacktestEngine(data, strategy)	Event-driven backtest engine
VectorizedEngine(initialCash)	Vectorized backtest engine
QuickStrategy(name)	Declarative strategy builder

Strategy Builder Methods

Method	Description
.buyWhen(condition)	Add buy condition
.sellWhen(condition)	Add sell condition
.stopLoss(pct)	Set stop loss percentage
.takeProfit(pct)	Set take profit percentage
.trailingStop(pct)	Set trailing stop percentage

Condition Builders

Function	Returns
sma(period)	SMA indicator
ema(period)	EMA indicator
rsi(period)	RSI indicator
macd(fast, slow, signal)	MACD indicator
bollinger(period, std)	Bollinger Bands
atr(period)	ATR indicator
price()	Current price
crossover(fast, slow)	Crossover condition
crossunder(fast, slow)	Crossunder condition

Indicators support comparison operators: sma(10) > sma(30), rsi(14) < 30

◈ Architecture

Tradix follows a modular architecture. Each subsystem is independently importable, so you can use just the indicators, just the risk engine, or the full pipeline.

tradix/
├── engine.py # Core backtest engine
├── multiAssetEngine.py # Multi-asset portfolio engine
├── strategy/ # Strategy base + 60+ indicators + ensemble
├── easy/ # 2-line API, presets, Korean API
├── vectorized/ # Vectorized engine + indicators (Pure NumPy)
├── datafeed/ # Data feeds (FinanceDataReader + Parquet cache)
├── broker/ # Commission, slippage, fill simulation
├── risk/ # Position sizing, VaR, Monte Carlo
├── optimize/ # Grid / random search optimizer
├── walkforward/ # Walk-forward analysis
├── analytics/ # Strategy DNA, Black Swan, Health Score, etc.
├── portfolio/ # Portfolio tracking + optimization
├── quant/ # Factor analysis, statistical arbitrage
├── signals/ # Signal prediction + adaptive signals
├── advisor/ # Market regime + strategy recommendation
├── entities/ # Bar, Order, Position, Trade
├── events/ # Event system
├── tui/ # Terminal UI (Rich + Plotext + Textual)
├── cli.py # Typer CLI
└── tests/ # 87 tests

◈ Contributing

Contributions are welcome. Fork the repo, create a feature branch, and submit a pull request.

git clone https://github.com/eddmpython/tradix.git
cd tradix
uv sync --dev
uv run pytest

◈ Support

If Tradix helps your trading research, consider supporting the project:

◈ License

MIT — Use freely in personal and commercial projects.

Trade smarter. Backtest faster.