If you have ever built or tested a trading strategy, you have probably come across two approaches. Vectorized backtesting and event driven backtesting.
At first, they may seem similar. Both aim to test strategies on historical data. But under the hood, they work very differently.
This article simplifies event driven backtesting and explains why it matters for traders.
What is an Event Loop in Trading
An event loop is a system that processes things one at a time in sequence.
In trading, the system reacts to events such as:
New market data arrives
A signal is generated
An order is placed
An order gets filled
Each of these is an event.
The loop continuously waits for the next event and processes it step by step.
This mirrors how real markets operate. Everything happens in a sequence, not all at once.
Types of Events in Event Driven Backtesting
Event driven systems are built around different categories of events.
Market Data Events
New tick or candle data arrives
Signal Events
Your strategy generates buy or sell decisions
Order Events
Orders are created and sent to the broker
Execution Events
Orders get filled, partially filled, or rejected
Order Update Events
Orders are modified, cancelled, or updated after partial fills
Portfolio Events
Positions, cash, and profit and loss are updated
How Event Driven Backtesting Works
Event driven backtesting simulates real world flow step by step.
Simple loop:
- Market data event
- Signal event
- Order event
- Execution event
- Portfolio update
Then repeat.
Does Event Driven Mean Tick Data Only
No.
Event driven backtesting works with:
Tick data
One minute data
Hourly data
Daily data
Each data point becomes an event.
Tick data is not mandatory. It only becomes important for very high precision strategies.
Why This Matters
Real trading is not instant.
Orders can be:
Partially filled
Delayed
Modified
Cancelled
Event driven systems capture all of this.
Coming from Vectorized Tools like Amibroker and TradingView
If you are used to tools like AmiBroker or TradingView, this shift can feel very different.
Here are the key things to understand:
1. Bar by Bar Execution
You no longer compute on full arrays. Each bar is processed one at a time. The system only knows the past and present.
2. No Look Ahead Bias by Design
Future data does not exist yet. This makes your backtest naturally safer from accidental mistakes.
3. The Core Loop is an Event Queue
Everything flows through a sequence:
MarketEvent → SignalEvent → OrderEvent → FillEvent
Each event triggers the next.
4. Realistic Order Simulation
You can model:
Slippage
Partial fills
Different order types
Latency
Vectorized tools often assume perfect fills.
5. Portfolio State is Always Live
At every step, your system tracks:
Equity
Open positions
Available capital
This allows dynamic position sizing.
6. It is Much Slower
Vectorized backtests run extremely fast.
Event driven systems process each step, so they take more time, especially in Python.
7. Costs are Easier to Model
You can inject costs at execution level:
Brokerage
Taxes
Exchange fees
Impact cost
This gives more control than global settings.
8. Multi Asset Logic is Natural
Since state is updated continuously, you can easily implement:
Pairs trading
Hedging
Cross asset signals
9. Easier Transition to Live Trading
The same system can be connected to live data.
You replace historical data with live feed, and the logic stays the same.
10. Libraries to Explore
Good starting points:
Backtrader
Mature and widely used
bt
Simple and useful for learning
Nautilus Trader
High performance and modern
You can also explore QuantConnect LEAN for a production grade engine.
Why Event Driven Backtesting is Slower
Event driven systems simulate each step in time.
For every time step:
State is updated
Events are processed
Orders are simulated
This creates many small operations.
Vectorized systems process everything at once using optimized math libraries.
So:
Vectorized backtesting is fast because it simplifies reality
Event driven backtesting is slower because it simulates reality
When Should You Use Event Driven Backtesting
Use event driven when:
Execution matters
Strategy depends on timing
You want realistic results
Use vectorized when:
You need speed
You are exploring ideas
Most traders use both.
Final Thoughts
Event driven backtesting is about simulating the full trading lifecycle.
From data to signal to order to execution.
It is not about tick data. It is about sequence and realism.
If you want to move closer to real trading conditions, this approach is essential.
