Walk Forward Analysis: Validating Your System
Test your system's robustness before deploying it to live trading
- Understand what walk forward analysis is and why it matters
- Learn the difference between in-sample and out-of-sample testing
- Run a basic walk forward test in GSB Pro
- Interpret walk forward results to validate system robustness
- Identify red flags that indicate an over-fitted system
Video coming soon
Andrew will record this lesson video
Why Walk Forward Analysis Matters
When you build a trading system in GSB Pro, the system is optimized against historical data. This is called "in-sample" testing. While it shows that your system works on past data, it doesn't guarantee it will work on future data.
Walk forward analysis is the solution. It tests your system on data it has never seen before, simulating real-world trading conditions. This reveals whether your system is truly robust or whether it was just lucky on the data you optimized against.
Over-Fitting Risk
Over-fitting happens when a system is optimized so closely to historical data that it becomes brittle and unreliable on new data. Walk forward testing catches this before you deploy capital to live trading.
In-Sample vs. Out-of-Sample Testing
In-Sample (What GSB Does During Optimization)
- Data Used: Historical data you're optimizing against
- What It Shows: How well parameters work on known data
- Limitation: Cannot guarantee future performance
- Example: You optimize a 50/200 moving average system on 2 years of S&P 500 data
Out-of-Sample (What Walk Forward Does)
- Data Used: Historical data your system has never seen
- What It Shows: How well your system performs on "future" data
- Benefit: More realistic prediction of live trading performance
- Example: After optimization on 2 years, test on the next 6 months the system never saw
How Walk Forward Testing Works (Basic Overview)
Walk forward analysis divides your historical data into rolling windows:
- Optimization Window: GSB optimizes parameters on a period (e.g., years 1-2)
- Test Window: The optimized system is tested on the following period (e.g., 6 months immediately after)
- Roll Forward: The window shifts forward and repeats multiple times
- Compare Results: In-sample results are compared to out-of-sample results
Running a Walk Forward Test in GSB Pro
During your trial, you can run walk forward analysis on your sample project:
1. Load a completed build
Go to your previous build results and look for the walk forward option in the menu or toolbar
2. Select walk forward settings
Choose how to divide your data (e.g., optimize on 80%, test on next 20%)
3. Run the walk forward test
GSB will optimize multiple times and test each optimization on fresh data
4. Review results
Compare in-sample and out-of-sample equity curves and statistics
Interpreting Walk Forward Results
Good Walk Forward Results (Robust System)
Signs of Robustness
- • Out-of-sample results are similar to in-sample results (within 10-20%)
- • Both curves show consistent upward trends
- • Drawdowns are similar in both in-sample and out-of-sample
- • The system performs consistently across multiple walk forward periods
- • No dramatic difference in profit factor or win rate
Poor Walk Forward Results (Red Flags)
Over-Fitting Warning Signs
- • In-sample profit is 50%+ higher than out-of-sample
- • In-sample looks great but out-of-sample is flat or losing
- • Drawdowns are much larger out-of-sample
- • Results vary dramatically between walk forward periods
- • Win rate is high in-sample but low out-of-sample
- • System performed great on one specific time period only
Trial Scope: Walk Forward During Your 14 Days
During your trial, you'll have access to GSB Pro's core walk forward analysis features. This is enough to validate the systems you build and see whether they're truly robust.
Advanced walk forward optimization (like the Enhanced Walk Forward Optimizer) is a Members-only feature that automates the process further. As a trial user, you can still run basic walk forward tests to validate your discoveries.
Key Takeaway
Before you deploy any system to live trading, validate it with walk forward analysis.If your system's out-of-sample results are significantly worse than in-sample results, it's likely over-fitted and won't perform well on live data. A truly robust system shows consistent performance on both in-sample and out-of-sample data.