Platform

Four parts of one research workflow

Each part does one job. Here is what it does, which data it uses, what you get — and where it stops.

Strategy Notebook

Saved to your account

A research note for each idea: the idea in your words, a structured hypothesis, the test setup, every saved run and your conclusion.

What it does: Keeps the hypothesis, inputs and results together so you can see later exactly what you tested and why.
Data used: Only what you type, plus the runs you save. Nothing is imported from other services.
What you get: A dated history of runs and edits, export as JSON, CSV or Markdown, and deletion at any time.
Limits: Up to 200 notes per account and 50 runs per note. Notes are private to your account; there is no sharing yet.

ConditionFast MA crosses below slow MA

Expected outcomeSmaller drawdown than holding

Horizon252 bars

Fails ifWorse net result in 3 of 4 scenarios

Runs saved4

Example note structure. Values are illustrative.

Sample dataSynthetic

Four sample market regimes to test the same idea against, each generated from a documented model with a fixed seed.

What it does: Lets you see whether a result depends on the kind of market rather than on the rule.
Data used: Synthetic daily bars from a random-walk model with regime-specific drift and volatility. Never market prices.
What you get: Steady trend, range-bound, volatility shock and drawdown-and-recovery scenarios, 252 bars each.
Limits: Four paths are a teaching set, not a representative sample of markets. Real prices have features the model leaves out.

Steady trenddrift +0.07%/bar · vol 0.8%

Range-boundmean reversion to 100

Volatility shock−8% gap at bar 141

Drawdown & recovery−0.22%/bar for 90 bars

Scenario parameters as documented on the Methodology page.

Paper tradingSimulation

A deterministic paper simulation of three strategy templates with explicit costs. Nothing leaves the workspace.

What it does: Applies the rule bar by bar: signal on the close, fill at the next open, fees and slippage on every fill.
Data used: The selected sample scenario and your parameters: windows, fee, slippage, position size.
What you get: Equity curve against buy-and-hold, net and gross result, drawdown, trades, exposure and cost share.
Limits: Long-or-flat only, no leverage, no partial fills, no order book or latency. It is not a backtest on real prices.

TemplateMoving-average crossover

Fast / slow10 / 40 bars

Fee + slippage5 + 5 bps per fill

Position size100% of equity

Executionnext bar open

Default settings for a new note.

Rule-basedAI optional

Plain-language statements derived from each run by published rules, so you can see why the summary says what it says.

What it does: Flags small samples, heavy costs, deep drawdowns, low exposure and other patterns you should notice.
Data used: Only the metrics of the run. The thresholds are listed on the Methodology page.
What you get: A short, consistent summary for every run. If the operator enables an AI provider, you can also ask a model to explain the result; that text is labelled with the model name.
Limits: A summary describes what happened in the simulation. It is not a forecast and not a recommendation.

Trades: 6“Too few to tell skill from chance”

Costs / gross: 41%“Small edges are fragile”

Max drawdown: 18%“Would you have kept following the rules?”

Examples of rule-based statements and their triggers.

Confirm your email, write one idea down and run it against all four scenarios. It takes about five minutes.