Testing Trading Strategies with Synthetic Market Data
Summary
The article explains how synthetic OHLCV data can supplement historical backtests when a market has limited data or when a researcher wants to probe strategy behavior in unfamiliar conditions. It proposes generating bars with simple randomized price changes for patterns such as upward or downward moves, narrow or wide ranges, and neutral variation, then saving them to CSV and serving them through a custom data source for backtesting. The generated series can also be plotted for inspection.
The article suggests using these scenarios to examine robustness, expose weaknesses, and explore parameter sensitivity. Its example checks basic bar relationships and continuity, but it does not offer a sophisticated market model or empirical validation of the simulated distributions. The author cautions that realism depends on the generator design and that synthetic tests supplement rather than replace testing on real market data. The setup is demonstrated in a platform-specific workflow, with a hosted service supplying saved records to the backtest system.
Key ideas
- Synthetic price series can broaden strategy testing beyond the regimes present in available historical data.
- A simple generator varies random price changes to create trend, range, and neutral bar patterns.
- Generated OHLCV records can be saved and exposed through a custom backtest data source.
- Bar validity and continuity should be checked before using generated records.
- Synthetic results depend on how realistically the generator represents market behavior and cannot replace real-data validation.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.