Backtest Evidence Review

A second read of your backtest — before you trust the curve.

You send one strategy (trade-results spreadsheet or CSV export + rules). I send back a short report on whether the evidence holds up: costs, out-of-sample data not used to tune the strategy, comparison with randomized results, sample size. Not whether to trade it.

Independent · Spain · 3 founding slots left

€79 one backtest · report in 2 business days after complete submission and accepted files

Get the review — €79

Not sure it fits? Ask first

Ask first opens a fit-check email draft — not a purchase. Stripe-hosted checkout. After I verify payment, I email intake + a private Proton Drive folder for uploads.

A worked example — including a failed result

I run a personal research lab where I pre-register hypotheses before testing them. Six experiments so far (H001–H006). None passed my own gates for live capital — including H001, which looked fine on simulated out-of-sample profit until the randomized-results comparison and trade count failed.

H001 verdict: insufficient evidence (simulated PnL +113.92 on 1,000 start, randomized comparison failed, 13 trades < 15 required)

Out-of-sample trades:
13
Simulated PnL (1,000 start):
113.92
Randomized comparison:
FAIL (113.9 < 95th percentile threshold 455.5)
Sample gate:
FAIL (13 < 15 min trades)

Read the full public H001 summary →

What I need from you

One strategy. A trade-results spreadsheet or CSV export. Written rules and cost assumptions. Optional: a superficial review of up to 200 lines, without running the code.

No API keys. No “make this profitable” requests. No signals.

What you get back

Limits, file sizes, data handling →

How it works

  1. Buy — Stripe-hosted checkout (Service Terms and Privacy Policy accepted there).
  2. Intake — I email a form link and a temporary upload folder after payment is verified.
  3. Report — within 2 business days of complete submission and accepted files.

Who I am

Xavier Gisbert Blanquer — independent methodology reviewer, Spain. I pre-register hypotheses and walk-forward splits before running out-of-sample tests; I apply a comparison with randomized results (null test) and minimum trade-count gates before treating any curve as evidence. The H001 sample above is one of six lab runs — all failed those gates.

This is methodology feedback on materials you provide. Not investment advice. Past backtests do not predict future results. You remain responsible for trading and compliance decisions. Legal · Privacy · Terms