- Home
- Skills
- Finance & Accounting
- Trading Strategy Backtest: Don't Fool Yourself
More screenshots
Works with the AI Agents you already use
Trading Strategy Backtest: Don't Fool Yourself
Test trading ideas against history without fooling yourself — look-ahead discipline, signal and execution separation, honest accounting, and bias disclosure.
$5
Trading Strategy Backtest: Don't Fool Yourself
Example session with this skill installed
We have a momentum strategy that backtests at 28 percent a year. Before we put money behind it, check whether the result is real.
- Read your context and instructions
- Compiled the trading strategy backtest
Rebuilt the backtest from the raw bars with a strict rule: a signal computed on a bar may only trade on the next bar, never the same one. The original code had used the closing price of the signal bar as its fill price — look-ahead bias, worth about nine points of the claimed return. Separated signal from execution next: the signal fired on the close, but the fill was assumed at the close price with no slippage and no commission. Re-priced fills at the next open with a realistic spread and a per-trade commission, which removed another six points. Checked the universe: the original test had run on the current index constituents, which silently drops every company that failed and left the index — survivorship bias. Rebuilt the universe as it actually stood at each date, which removed a further five points. What survived was an 8 percent strategy, not a 28 percent one — and it was concentrated in two short periods, with a drawdown of 31 percent that the original summary had smoothed over by reporting only annual averages. The verdict was written plainly: the idea has a real edge, but a small and lumpy one, and the sizing must assume the 31 percent drawdown will happen again. The deliverable is the corrected backtest with every assumption listed, the bias disclosure attached, and the exact conditions under which the team would stop trading it — written before any money moves, not after the first loss.
Connects securely to your tools. The creator never sees your data.
About this skill
A backtest that flatters its author is worse than no backtest: it spends real money on a fantasy. This skill tests trading ideas against history with look-ahead discipline, a clean separation between signal and execution, warmup handling, and honest accounting of costs and slippage. It covers the choice between an event study and a full strategy backtest, the biases that inflate results — survivorship, selection, overfitting — and how to disclose them instead of hiding them. It also covers adversarial self-review: the checks you run on your own result before anyone else's money touches it.
It tests ideas against history; past performance does not predict future results, however clean the test.
It needs honest data — point-in-time universes and realistic costs. A backtest built on today's constituents or zero-cost fills proves nothing.
It is not investment advice. The output is a method for testing an idea, not a recommendation to trade it.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 3 days ago
Permissions
No special permissions declared or detected