WORKSPACE / 05 / STRATEGY TESTER

Interrogate the idea.

A deterministic replay should make the assumptions as visible as the outcome.

Testing methodology

REPLAY REQUEST / 90 DAYS

Test the rules.
Keep the assumptions.

Saved thesis

LookbackModelled slippage · bps

Choose a thesis and set the replay assumptions. Results become available after a completed test.

Create a thesis first

REPLAY WINDOW / OUTCOME SERIESAwaiting data

+ RETURN

Your outcome series will appear here.
Prepare a request to define the test.

DAY −90−60−30TODAY

—Firing events

—Net P&L

—Max drawdown

What a test has to account for.

The data and assumptions that make an outcome reproducible.

01 / THE DATA

Versioned inputs.

Hyperliquid candles, funding and open interest. Robinhood Chain pool history. Fear & Greed snapshots. X-post history is a later, cost-dependent extension.

02 / THE FILL MODEL

Execution has a cost.

Simulated fills use the mark plus or minus modelled slippage. Funding, venue fees, liquidity, and partial fills need explicit assumptions before interpreting the result.

03 / THE IDENTITY

A test you can repeat.

The specified test ID is SHA-256 over the specification, snapshot IDs, and tester version. Identical inputs should reproduce the same replay.

Evidence is not a forecast.

Historical results do not predict future returns. A good test reveals sensitivity and failure cases; it does not remove market or execution risk.