Turn your data into distributions
Synthetic Data Generation for
Trade Backtesting
infinitely scale your data
How does it work?
Xample learns distributions from your time-series data. Customers then use these distributions to generate infinite time-series data, with the same statistical properties as the original data.
App Screenshot
Financial Data is Difficult
Generating synthetic time-series is hard, especially financial data.
How do you strip out the effect of stock splits?
What about days where the market was limit down?
What about new instruments which don't have a lot of price history?
How do you handle heteroskedasticity?
How do you handle index additions and deletions?
We get it
Xample's founding team include former derivative traders at Wall Street banks, portfolio managers at systematic hedge funds, and machine learning engineers at F1 teams. We've dealt with all of these issues before, and we've built a system that takes care of it.
What will you do with infinite data?
Backtest your trading strategies the robust way
Backtests of trading strategies are incredibly path-dependent. If you only ever backtest against historical data, you are only testing one path, one that's already happened. With Xample, you can generate thousands of statistically significant paths, and test your strategy against all of them.
Fully understand your risk
Run your risk scenarios (max-drawdown, peak-to-valley, max-shortfall) on thousands of paths, not just one. Understand the distribution of your vulnerabilities, not just the point estimate.
Train better machine learning models
Machine learning models trained on financial data often overfit. By generating synthetic data, you can vastly scale the size of your dataset, while still maintaining the core statistical properties of the data. This enables you to train a more performant model, with far less risk of overfitting.
Get started today
Free
Generated datapoints per day:
10,000
Maximum training set size:
5,000
Saved distributions:
5
On-prem deployment:
No
Montly cost:
$0
Enterprise
Generated datapoints per day:
Maximum training set size:
1,000,000
Saved distributions:
500
On-prem deployment:
Yes
Monthly cost:
$1,000