The story behind EdgeAudit.
EdgeAudit is built by a UK software engineer with several years of experience building algorithmic trading systems and statistical research tools across sports and financial markets. EdgeAudit is the latest of those projects, and the first one shipped as a product.
How I got here
My background is in building automated trading and analysis systems for retail-scale markets. Over the last few years I've shipped algorithmic bots for sports betting markets, sports analytics tooling, and a research-grade backtesting framework with bootstrap confidence intervals, walk-forward validation, and Bonferroni-corrected significance applied to every strategy I tested.
Working in those markets at scale taught me a specific lesson: rigorous statistical discipline is the single biggest gap between research-grade work and the typical retail toolkit. The methodology that's standard in academic finance — applied across every paper you'd see in the Journal of Financial Economics — is almost never the default in the tools available to retail algorithmic traders.
That gap is what EdgeAudit closes.
Why a product
I'd been applying these statistical checks privately on my own strategies for some time. The pipeline — parse strategy, fetch data, run backtest, apply Bonferroni for multiple comparisons, bootstrap returns for confidence intervals, split chronologically for walk-forward, return a structured verdict — was already battle-tested.
Packaging it as a product was the obvious next step. Retail algorithmic traders shouldn't have to rebuild a research pipeline from scratch every time they want to test a strategy properly. EdgeAudit gives them the same statistical rigour I use, accessible through a Discord slash command, with a structured verdict back in under a minute.
What's behind the product
EdgeAudit isn't a thin wrapper. The backtest engine is purpose-built for OHLC-bar iteration with realistic commission and frictions modelling. The strategy parser uses DeepSeek to convert plain-English descriptions into a structured DSL of indicators and conditions. The statistics layer applies Bonferroni correction across the strategies you submit, computes 10,000-resample bootstrap confidence intervals on returns, Sharpe, and drawdown, and reports walk-forward holdout performance separately from the training window.
The whole pipeline is open to inspection in the methodology blog post, and the codebase is built with the kind of testing discipline you'd expect from production research code — over 140 unit tests at the time of writing, covering every layer of the engine.
How EdgeAudit makes money
Simple subscription model:
- Free tier with daily-volume cap, so anyone can evaluate the methodology before paying.
- Standard tier at £30/month for unlimited backtests.
- Pro tier at £75/month for batch comparison and exportable reports.
- Founders deal at £15/month for life for the first 50 customers — direct support for the early supporters who help make the product better.
No upsells, no signal-selling, no "lifetime" gimmicks. Cancel any time from Stripe's customer portal.
What I'm not selling
EdgeAudit doesn't sell trade signals. It doesn't predict markets. It doesn't connect to brokers or move money. It's a research and validation tool — its job is to tell you whether the strategy you came up with would have held up under proper statistical scrutiny. What you do with that answer is up to you.
I'm not in the business of telling people what to trade. I'm in the business of helping people honestly evaluate the ideas they already have.
Contact
Solo founder, UK-based. Email hello@edgeaudit.app — I read every one.
If you have a strategy you want stress-tested, install the bot and run it. If something doesn't work as expected, tell me. If you have a feature you'd find useful, tell me. The product is shaped by the people who actually use it.