TL;DR
Open the exact library recipe, inspect its seven buy rules, set dates and capital, then run QuickTest once and wait for the result. The AI recipe's 2024 product example returned 70.97% from initial capital versus SPY's 25.30%, with larger drawdown of 23.49% versus 9.13%. Its sequential buying power percentages are not fixed portfolio weights, and it has no sell rule. Save fees, inspect purchases and holdings, and keep any changed candidate separate before testing dates excluded from selection.
- Authoring
- Plain-language instructions
- Evidence
- Historical simulation
Follow one worked path
Open the exact seven-rule recipe
Inspect the Artificial Intelligence Revolution library preview and all seven buy rules before running the test.
Open the exact example recipeThe original sequential cash percentages are visible. Opening the recipe does not submit a backtest.
Set the worked historical window
Keep the rules unchanged. Select January 1 through December 31, 2024 and $10,000 starting capital, then review the displayed fee settings.
The inputs are explicit. The recorded 2024 example is retrospective, not an unseen test.
Run QuickTest once and read the result
Submit once and wait for the original test. Inspect the statistics table, drawdown and filled events before making any deployment decision.
- Return from initial capital
- AI recipe: 70.97% · Hold SPY: 25.30%
- Maximum drawdown
- AI recipe: 23.49% · Hold SPY: 9.13%
- Sharpe
- AI recipe: 1.75 · Hold SPY: 1.42
- Sortino
- AI recipe: 2.38 · Hold SPY: 1.90
The actual example's table and chart use different return bases, explained in the evidence below. Higher return came with a larger drawdown than Hold SPY.
See the completed example
Run this strategy in three steps
Show run this strategy in three steps
Open the recipe
Use Open the example strategy to inspect all seven buy rules in the library preview.
Set the test
Keep the rules unchanged. Set January 1 through December 31, 2024 and $10,000 starting capital.
Run QuickTest once
Opening the preview does not submit a test. Select Run QuickTest, then read its status and completed statistics before submitting again.
Exact rules and recorded result
Show rules, result values and assumptions
Open the exact built-in recipe below. Its seven rules buy CRM 5%, AVGO 5%, META 15%, TSLA 15%, NVDA 20%, MSFT 20% and GOOGL 20% of remaining buying power when evaluated. Inspect every rule before testing: these are sequential cash percentages, not fixed portfolio weights. The template has no sell rule and can buy again while cash remains.
In the library preview, keep those seven rules unchanged, set January 1 through December 31, 2024 and $10,000 starting capital, then select Run QuickTest once. Opening the preview does not submit the test. Read its status and completed statistics instead of submitting again while it runs.
We completed that product QuickTest on October 10, 2026. The table uses the initial-$10,000 return basis. The chart header's 71.14% uses its first visible point; it differs from the statistics table's 70.97% basis. Higher return came with a larger maximum drawdown than Hold SPY.
The final book held seven positions, with $47.25 cash. Dollars sold was zero, win rate was N/A and profit factor was 0.00. Open the Filled events to follow each purchase and the capital remaining afterward. For your own test, save the selected fees with the result and compare the holdings around its drawdown dates.
Scroll sideways to compare all columns. Select a heading to sort.
| Return from initial capital | 70.97% | 25.30% |
| Maximum drawdown | 23.49% | 9.13% |
| Sharpe | 1.75 | 1.42 |
| Sortino | 2.38 | 1.90 |
No matching rows. Clear the filter to see all records.
Optional: build your own rule with Aurora
Show optional: build your own rule with aurora
Open Aurora using the link below and describe one rule. On the returned card, select View Portfolio to inspect the definition. Ask Aurora to propose the test dates, capital and benchmark, then approve the historical run after checking those values. You can also use the portfolio Backtest control and select Run Backtest after setting its dates.
| Part of your request | A concrete example to define |
|---|---|
| Asset | SPY |
| Entry | A stated indicator condition, with its lookback period |
| Sizing | A stated dollar amount or allocation with a maximum exposure |
| Exit | A stated sell condition and which position it closes |
| Timing | Daily evaluation with explicit test dates |
No matching rows. Clear the filter to see all records.
Try a separate SPY candidate in Aurora
Use this request to create a separate SPY draft. Review its rules and test settings in Aurora before running it.
Create a new $10,000 SPY strategy draft. Buy 10 shares only when SPY price is strictly above its 200-day simple moving average and the portfolio holds no SPY. Sell all SPY only when price is strictly below that average and a SPY position is held. Equality takes no action. Use cash dividends. Show the exact entry, sizing and exit rules for review. After I approve the draft, propose a daily backtest for January 1, 2024 through December 31, 2025 with buy-and-hold SPY and explicit fees. Wait for my approval before submitting that test.Inspect more than the return
Show inspect more than the return
After the job finishes, compare its chart with the benchmark and inspect the drawdown dates. Follow the signal and Filled events to see which rules acted, what quantities traded and how fees affected cash. Inspect the holdings during those periods to understand the portfolio's exposure.
Record the selected rule and the variants you tried, then test a period you did not use to choose the candidate.
Decide whether to keep researching
Show decide whether to keep researching
Save the accepted definition and operation identity. Change one hypothesis deliberately rather than asking for a prettier result. If the rule advances, begin a separately reviewed paper observation with simulated money.
Aurora stores explicit strategy rules and uses the same engine as the SDK. Review the proposed run in Aurora before submitting it; opening this guide does not start a backtest or deployment.
No comments yet.