# How to verify Stochastly

Each statement below links to the place where it can be checked: a public register, a published paper, a dated measurement, or a run on your own data.

## The publisher

- business register: SIREN 109 270 363, active since 28 August 2026, activity code 58.29C (application software publishing), Annecy, France: [public record in the French government company directory](https://annuaire-entreprises.data.gouv.fr/entreprise/109270363)
- identity and host: publisher, publication director, host and network provider, as article 1-1 of the French LCEN requires: [legal notice](https://stochastly.com/legal)
- people: the founder and the co-founder, and the part of the software each of them built: [credits](https://stochastly.com/credits)
- contact: contact@stochastly.com

## Your rights as a customer

- consumer mediator: CM2C, free of charge for a consumer whose written complaint stays unsettled, within one year of that complaint: [cm2c.net](https://www.cm2c.net) · [appointment in the legal notice](https://stochastly.com/legal)
- withdrawal: 14 days from the contract date for a consumer buying at a distance: [terms of use and sale](https://stochastly.com/terms)
- access: early access, free for life on one computer, no payment is taken: [terms of use and sale](https://stochastly.com/terms)
- regulatory status: a research software publisher that provides no investment service and issues no personal recommendation, so no AMF authorisation or ORIAS registration applies; the reasoning is in the [legal notice](https://stochastly.com/legal)
- live trading: direct broker execution and managed accounts are planned once the applicable financial regulation is in place: [roadmap](https://stochastly.com/roadmap)

## The statistics come from papers

The statistics a run reports come from published research, cited in the guides with a DOI where one exists. The deflated Sharpe ratio can be recomputed in the browser, starting from the worked example of the 2014 paper.

- deflated Sharpe ratio: Bailey and López de Prado (2014), [The Journal of Portfolio Management](https://doi.org/10.3905/jpm.2014.40.5.094) · [recompute it](https://stochastly.com/tools/deflated-sharpe-ratio-calculator)
- probabilistic Sharpe ratio: Bailey and López de Prado (2012), [Journal of Risk](https://doi.org/10.21314/JOR.2012.255), with the minimum track record length
- the other methods: PBO, walk-forward analysis and the tests behind them, with sources: [guides](https://stochastly.com/learn) · [statistical methods in the engine](https://stochastly.com/research)

## The product's own figures

A figure published about the product carries its definition, the command that remeasures it and the day it was measured.

- measurement protocol: how a figure is defined, remeasured and dated: [protocol](https://stochastly.com/research/method)
- dated product facts: node census, chart census, export targets, and engine speed with its hardware: [product facts](https://stochastly.com/evidence.html)
- exported files: what a Pine Script or MQL5 file reproduces of the run, target by target: [fidelity](https://stochastly.com/research/fidelity)
- look-ahead leakage: how the causality gate truncates the bars and replays the graph before a run: [causality](https://stochastly.com/research/causality)
- self-checks: the validations the desktop app runs on itself: [guards](https://stochastly.com/product/guards) · [validity domain](https://stochastly.com/research/gaps)
- changes: app, site and account changes listed by date, newest first: [version history](https://stochastly.com/version)

## What leaves your computer

The application sends one daily usage summary with no identifier, which you can turn off in Settings; it does not send your market data, strategies or results. It sends run data only if you opt in on the sign-in screen to verdict sharing for the public research counter or to backtest descriptions for model training, each sent under a random ID with no name or email; both start unticked and either can be turned off in Settings. The licence check runs on a timer, and the optional assistant is hosted, so its conversations leave the computer when it is switched on. The outbound hosts are measured from the source code and listed by name, and a network monitor running beside the app can confirm them. [Outbound hosts](https://stochastly.com/product/local) · [privacy policy](https://stochastly.com/privacy).

## Check it on your own data

Early access users install the desktop app on their own computer. A strategy you already know, run on your own price files and compared with your own records, gives a check that depends on no claim made by the publisher.

[Join the early access](https://stochastly.com/join) [Install guide](https://stochastly.com/docs/install)

---
Source: https://stochastly.com/trust