# Autrady

> Autrady audits trading strategies. Upload the signals behind a bot, a paid newsletter or your own backtest, and Autrady tests whether they beat the market after trading costs and allowing for luck. 14 days free with no card, then €19 a month including VAT.

Last updated: 2026-09-25

## What is Autrady

Autrady is research software from Autrady EOOD that audits the signals behind a trading strategy. A signal is a US stock or ETF ticker plus the moment a buy idea became public. Autrady enters at the next market open after each signal was knowable, charges spread and currency costs on both sides of every trade, and judges the result against the market's own move (and against size, value and momentum when factor data is available). The verdict is one of: passes, fails, or not enough evidence.

Autrady does not sell signals, does not trade for you and does not give investment advice.

## Who it is for

- People who pay for a signal service, newsletter or bot and want to know if its picks beat the market after costs
- Traders who want an honest second opinion on their own backtest before trading it
- Signal sellers who want a pre-registered, fingerprinted study to point to

## How an audit works

1. Upload a CSV with one row per buy signal: the ticker and the moment it became public. A date without a time counts as public after that day's US close; a date and time needs its time zone, such as 2024-02-02T15:30:00-05:00. An optional value column holds a number to filter on, such as a rating. Upload every call from one source (ideally two years or more), not a single tip. Example and template: https://autrady.com/files/autrady-signals-template.csv

   ```csv
   ticker,as_of,value
   AAPL,2024-01-08,
   NVDA,2024-01-22,5
   JPM,2024-02-02,
   ```

2. Write down the hypothesis and choose the holding period (1 to 252 trading days). The hypothesis, rule and holding period are fingerprinted before any result exists.
3. See the power check first: how large an effect the study can detect with this many signals.
4. Read the development verdict: alpha after the market (Newey-West t-statistic), both halves of the period, and a bar that rises with every study run on the same file.
5. Open the sealed holdout once, only after a development pass on prices that include delisted companies.

## What an audit checks

| Question | What the audit does | What it caught in Autrady's research (September 2026) |
|---|---|---|
| Does it beat the market, or ride it? | Judges returns after the market's move, and after size, value and momentum | 34 of 324 technical rules looked like winners over 30 years; none beat the market after risk |
| Would it survive the companies that died? | Flags results that rely on today's survivors and keeps the holdout sealed until retested with delisted companies | Insider buying scored t 2.95 on survivors and t 1.34 with delisted companies |
| Could anyone have traded it? | Enters at the next open after each signal was public; charges spreads and currency costs both ways | A viral grid bot's own backtest showed +24.5% a month; on 1-minute prices it lost 1.6% a month |
| Is the data clean? | Screens out impossible price moves | One corrupt price made an activist-stake strategy look like +2.35% a month |
| Is it luck? | Power check first, a bar that rises with every idea tested, one look at the holdout | 5 of 5 pre-registered strategies from academic research failed |

Results computed on prices that exclude delisted companies are labelled provisional in the app.

## Free luck check

The homepage has a calculator for claimed track records. Inputs: average monthly return, typical monthly swing, months of results and the number of versions tried. Outputs: the record's t-statistic, the bar after that many tries (one-sided, Bonferroni), the chance that the best of that many no-edge strategies looks this good, and how many months a real edge of that size needs to clear the bar with 80% probability.

## Use Autrady from an AI assistant (MCP)

Autrady runs a public MCP server at https://autrady.com/mcp (streamable HTTP, no sign-in, rate limited). Add it as a connector in Claude, ChatGPT, Cursor or any MCP client. Tools:

- `luck_check`: average monthly return, typical monthly swing, months of results and versions tried in; the chance that luck explains the record, the t-statistic against the bar for that many tries, and the months a real edge of that size needs.
- `explain_audit`: what a full audit of the actual signals checks, the price, the free trial and where to start.

Starter prompts:

- "Use Autrady's luck check: a strategy averaged +1.2% a month with a 4% monthly swing over 24 months, and it was the best of 20 versions I tried. Could that be luck, and how many months would prove it?"
- "Ask Autrady what a full audit would check for my newsletter's stock picks, and what it costs."

Audits of your own signal files run in the web app at https://autrady.com.

## Pricing

- Plan: Autrady, one plan
- Price: €19 per month, VAT included
- Free trial: 14 days, no card needed
- Cancel: anytime
- Limits: up to 20 signal files of 500,000 signals each; unlimited studies
- Sign-in: Google
- Machine-readable pricing: https://autrady.com/pricing.md

## Important disclaimer

Autrady is research software. It tests historical evidence, cannot predict future returns and does not provide investment advice. Past results, real or simulated, do not guarantee future performance.

## Contact

Email office@autrady.com. Autrady EOOD, Varna, Bulgaria (UIC 206828616, VAT BG206828616).

## Links

- [Home](https://autrady.com/): what an audit checks, the free luck check, pricing and FAQ
- [Pricing](https://autrady.com/pricing.md): structured plan details for agents
- [Privacy](https://autrady.com/Home/Privacy): privacy policy
- [Terms](https://autrady.com/Home/Terms): terms of service, trial, cancelling and refunds