Skip to content
Solvex Space
Algorithmic SystemsFrom $2,499 · quoted to scopeAdvisory engagement

Strategy Backtesting & Research Validation

An independent examination of whether a backtest demonstrates anything — realistic costs, walk-forward and out-of-sample testing, and a deliberate hunt for the leakage, survivorship and overfitting that make most impressive equity curves meaningless.

Request a quote

Verified Solvex Specialist

Verified by Solvex

Direct specialist contact for Solvex engagements

What this means. An authorized Solvex administrator registered and approved this exact public identity. What it does not. Solvex has not inspected the account on the platform, and this is not the platform's own verification.

Solvex specialists never ask for your passwords, recovery phrases, one-time codes, or payments to a personal account. Work, scope and invoices are agreed in writing through the official channels on this site.

What's covered

  • Reconstruction of the backtest under controlled, reproducible conditions
  • Realistic transaction costs: commission, spread, slippage and financing
  • Walk-forward and out-of-sample validation
  • Monte Carlo and trade-order randomisation to test path dependence
  • Overfitting analysis, including parameter sensitivity
  • Survivorship and selection bias review of the data set
  • Look-ahead and data-leakage detection
  • Performance attribution — what actually produced the return

What you receive

  • Independent validation report with reproducible methodology
  • Cost-adjusted performance under realistic assumptions
  • Out-of-sample and walk-forward results
  • Robustness analysis and parameter sensitivity
  • Explicit list of biases and leakage found, with their impact
  • A plain statement of what the results do and do not support

Evidence and reporting

How the work is kept honest
  • Evidence, frozen at issuanceFindings tie to something observed. When the report is issued, the evidence behind it is frozen in the same transaction and cannot be edited afterwards.
  • A signed reportAn Ed25519 signature covers both the report content and the delivered file. Alter a byte of either and verification fails.
  • Written scope firstAdvisory work runs to a written scope agreed before it starts, so what you receive is what was agreed.

Anyone holding a Solvex report can verify it publicly without seeing its contents.

Our boundaries

What this engagement does not do, stated before it starts.

  • We do not endorse, recommend or rate any strategy for investment
  • A validated backtest is evidence about the past, never a forecast
  • Live execution differs from simulation in ways no backtest fully captures
  • Data quality is bounded by the data you provide or license

How this engagement runs

  1. 01

    Intake

    Tell us the system, the goal and the constraints. If the work is not a good fit, we say so before anyone is invoiced.

  2. 02

    Scope and authorization

    Written scope and signed authorization before anything is touched. Security testing runs only against systems you own or are contractually entitled to have tested.

  3. 03

    Investigation or build

    Specialists matched to the work. Findings are proven by hand — scanner output is a lead, never a finding.

  4. 04

    Evidence

    Every finding ties to something observed. When a report is issued, its evidence is frozen in the same transaction, so what backed the report cannot change afterwards.

  5. 05

    Delivery

    A signed report: an Ed25519 signature over both the content and the file, with a short verification reference you can read down a phone.

  6. 06

    Verification and retest

    Anyone holding the report can verify it publicly without seeing its contents. Fixes are retested as part of the engagement — “fixed” means we confirmed it.

Questions we are asked

Our backtest shows a Sharpe of 4. Is that good?
It is a reason for suspicion rather than celebration. A Sharpe that high in a retail-accessible strategy usually indicates leakage, unrealistic fills, survivorship in the universe, or parameters fitted to the very data being reported. Finding out which is exactly what this engagement is for.
What is data leakage in practice?
Using information the strategy could not have had at decision time. Adjusted closes that embed future corporate actions, a universe filtered by who survived, a stop filled at a price never actually traded, a feature normalised over the whole sample. Each looks harmless and each can invent the entire edge.
Will you tell us the strategy does not work?
If that is what the evidence shows, yes — plainly, and with the analysis that led there. That is the value: a strategy that fails validation costs a fee, and one that fails with live capital costs considerably more.