01
Make the rules yours.
Describe your strategy or bring Python. Start with a simple allocation, then add signals, constraints, or other models.
Plain language → inspectable code
Portfolio backtesting, on your terms
Your portfolio idea.
Put to the test.
Build an investment model from your idea. Explore its historical results, understand the risks, and keep improving it in one workspace.
No card required. Credit expires after 7 days.
Builds and runs are metered. See usage pricing.
ENSEMBLE / ModelsInteractive preview
Sample workspace
COREBalanced Core
- Annualized return
- 4.5%
- Volatility
- 5.6%
- Max drawdown
- -11.3%
Growth of 100Hypothetical backtest
2011-08-012026-09-25
Look-through allocation2026-09-25
IEF51.4%
VTI17.1%
VXUS12.5%
CTA9.5%
GLD9.4%
The underlying assets, across both child models. Historical sample weights, not a recommendation.
Balanced CoreEqual risk contribution
Global EquityVTI + VXUSInverse volatility
DiversifiersIEF + GLD + CTAInverse volatility
- Lookback
- 60 days
- Rebalance
- Monthly
- Execution
- Next close
Combine reusable models and inspect the rules behind the allocation.
2 child models5 assetsDaily price data
From an idea to something you can use
A backtest is the beginning.
Give your strategy a home, with the tools to understand it and the flexibility to make it your own.
02
Keep the whole picture.
Inspect results alongside code, inputs, and data assumptions. Change a parameter and keep both runs for comparison.
Code + inputs + run history
03
Continue wherever you work.
Open the same saved model in the web app, with your AI agent, or through API and CLI. Enable scheduled refreshes when ready.
Web · MCP · API · CLI
Start with one questionWhat would you
What would you
like to test?
Your first model can be simple. Describe the assets, the weights, and when to rebalance.
See a worked 60/40 exampleAn example starting point
“Build a portfolio with 60% SPY and 40% IEF. Rebalance monthly. Show the backtest, drawdowns, and allocation.”
Saved modelInspectable codeHistorical results
Start with $100 credit No card required. Credit expires after 7 days.
Builds and runs are metered. See usage pricing.
Before you start
A few useful answers.
Research software. Your decisions.
ENSEMBLE does not place trades.
Do I need to write code?
No. Describe the assets, allocation rules, and rebalance schedule in plain language, or work with your connected AI agent. You can also supply Python. Review the generated code and assumptions before relying on a model.
What does it cost to try?
Start with $100 in credit, valid for 7 days, with no card required. Model builds and runs are metered, so the credit is not unlimited usage. Continued usage requires available credit or funding. Optional monthly plans and usage rates are on the pricing page.
What can I learn from the results?
Inspect historical returns, volatility, drawdowns, and allocations alongside the model's code and inputs. Available history varies by asset. Backtests are hypothetical, backfilled periods are synthetic, and past performance does not predict future returns.
Can ENSEMBLE keep my model up to date?
Yes. You can enable scheduled refreshes and inspect the resulting runs. Refreshes depend on data availability and your spending cap, and consume metered credit. ENSEMBLE produces research and allocations; it does not place trades.