Backtest Doctor

Your backtest looks profitable.

But are the numbers actually correct?

A backtest can look great while small record errors quietly distort the result.

Missing trading costs.
Incorrect net P&L.
Price or timing inconsistencies.

Backtest Doctor helps you find those problems before you trust the result.

It checks supported backtest records, finds inconsistencies, and shows the evidence behind each finding.

Not just “something looks wrong.”
You see exactly what does not match.

See what changed — and why

Backtest Doctor can inspect supported:

• P&L records
• trading costs
• prices
• holdings
• timing fields

When something does not reconcile, you can see:

Recorded value → Expected value

plus the record-level evidence behind the finding.

Real demo example

The included demo reported:

Return: +2.50%

Backtest Doctor found:

3 confirmed net P&L mismatches

The supplied fees and slippage did not reconcile with the recorded net P&L.

A supported repair candidate produced:

+2.50% → +2.21%

Net P&L: $25.00 → $22.10

The important part is knowing exactly why the recorded result changed.

Demo data · Example output · Results depend on the supplied records.

Evidence, not guesswork

Each supported finding can show:

• what was recorded
• what was expected
• where the mismatch occurred
• the evidence behind it

Where supported, Backtest Doctor can also generate a separate repair candidate for review.

Your original records are not silently overwritten.

When should you use it?

My backtest looks too good to be true.

Check whether the reported P&L matches the underlying supported records.

My results changed after a code change.

Find inconsistencies in the exported records instead of guessing.

AI wrote my backtest. Can I trust the output?

Check the exported records before relying on the result.

What you get

Depending on the supplied input and findings:

• Clear issue summaries
• Recorded vs. expected values
• Record-level evidence
• Supported repair candidates
• Verification results
• Markdown and JSON output

The goal is simple: know whether the numbers behind your backtest actually agree.

Built for the moment when something feels off

Maybe the return suddenly jumped.

Maybe a small code change produced a completely different result.

Maybe an AI-generated backtest looks suspiciously good.

Maybe you simply do not want to risk trusting the wrong numbers.

A convincing equity curve is not enough.

Before you spend more time improving a strategy — or start trusting its results — check whether the records behind it actually add up.

Find the mismatch before you trust the backtest.

How it works

1. Provide supported backtest records

Use the supported input format.

2. Run the inspection

Backtest Doctor checks supported relationships across prices, holdings, timing, costs, and P&L.

3. Review the evidence

See the affected record, recorded value, expected value, and reason for the mismatch.

4. Review the repair candidate

Where supported, inspect a separate corrected record candidate.

5. Verify the result

Use the evidence to decide whether the reported numbers should be relied on.

Important scope

Backtest Doctor is a record-audit and diagnostic tool.

It does not predict future profitability, guarantee returns, recommend strategies, execute trades, or automatically run and rewrite customer Python code.

It helps you inspect supported backtest records and identify evidence-backed inconsistencies before you rely on them.