StrategyCheck runs your trading strategy in paper and tells you the truth about whether it works.

It does not predict markets, does not recommend what to trade, does not provide signals and does not hold your money.

You write your rule in your own words or paste your Python script. StrategyCheck runs it every market day with real prices, enforces risk limits that no text can override, logs every decision and sends you a short daily message. With fewer trades than the sample for its horizon it says "it can't be known yet". Updated November 2, 2026.

Join the waiting list Read the methodology

The beta opens on February 28, 2027. Leave your email and we notify you once. Paper is free forever with one strategy, no card.

StrategyCheck does not recommend what to trade, does not provide signals and does not hold your money. Paper and backtest are simulated results: they guarantee nothing in real. Trading involves risk of loss.

StrategyCheck · 2026-12-10 · version v3 · PAPER

Status

PAPER account: practice, no real money. Data as of 17:00. Open positions at close: 1.

Today

Today 2 trades closed. Net result for the day: +US$ 48.20 PAPER.

Cumulative

It can't be known yet whether it works: 61 trades so far for this version (PAPER), 100 needed. 39 to go.

This version adds up to +US$ 412.60 PAPER over 61 trades, after costs.

From its highest point, the PAPER account fell 3.1%; your limit is 10%.

Gate

To trade for real: 23 of 40 market days without changes and 61 of 100 trades (PAPER).

Incidents

There was a market data outage from 12:23 to 12:24 (60 s). No bars were lost today.

Your positions and orders match the broker's.

Next step

Let the strategy run without changes; every change restarts the count.

Example. Daily message composed by code from the product's closed vocabulary (36 sentences, no language model). The figures are illustrative, not from any user. Intraday strategy in discipline mode: that is why the sample is 100.

How it works: three gates

Backtest, paper and real, in that order. The "trade for real" button stays disabled until the system verifies the gate. Nobody skips it, including us.

Backtest Works today

You write or paste your strategy. It runs on past data with the same code that later runs live. Every figure carries the BACKTEST label: nothing was traded. Today, only with the test account's templates.

Paper Works today

The strategy runs every market day with real prices and modeled costs, without your PC. Today only one runs: the product's own test account, in production since October 29, 2026.

Real Under construction

Only if the version meets the gate for its horizon, without changes, can you choose to trade it with your broker. StrategyCheck does not recommend it. Trading for real on Alpaca awaits an Alpaca Compliance approval.

What the gate requires, by horizon

Horizon	Closed trades (minimum)	Market days (minimum)
Intraday (discipline mode)	100	40
Swing (days to weeks)	40	60
Position (weeks to months)	20	120
Rebalanced portfolio	4 rebalances	180

Also, in every horizon: net gain per trade greater than 0; profit factor greater than 1.3 (for portfolios: net return greater than the benchmark); maximum drawdown of the version within your limit; no version changes in the period, because every new version restarts the count. Below the sample, the verdict is "it can't be known yet" and no percentages are shown. Definitions and formulas in the methodology.

Three ways to connect, not one connector per broker

Every path starts in paper, with simulated money and no broker. The broker comes last, and only if your strategy passes the gate.

Lane	What it is	Brokers	Who keeps the credentials	In the beta (Feb 28, 2027)
1. Native (OAuth)	You authorize from the broker's site. StrategyCheck hosts and executes; it never sees your keys.	Alpaca	Nobody: a revocable token with trading scope, in an encrypted vault.	Yes, in paper Trading for real: awaits Alpaca Compliance approval.
2. Your own bridge	An open-source connector (`strategycheck-bridge`) you run wherever you want. It receives intents already passed through the risk guard, executes and returns positions and fills.	Interactive Brokers at launch; any broker with an API by writing an adapter.	You, on your machine. StrategyCheck stores nothing from the broker.	Yes Code in the coming months. "Without your PC" applies to lane 1 only.
3. Measurement only	You import the trades CSV from any broker and get the edge report. No execution connected.	All	Nobody.	Yes The free public audit ("upload your CSV") has a planned date: November 20, 2026. How it will be computed.

New brokers: we count requests. A bridge adapter is built when three paying users ask for the same one; a native lane only when the broker accepts the partnership. Tradovate, Tradier and IOL are on the list, with no date.

The honesty rules are code, not copy

What the product can say is written in a closed vocabulary with one test per sentence. If a sentence is not on the list, it does not come out.

Every figure carries its n. No amount, percentage or average appears without the number of trades that produced it.
Every figure carries its label. PAPER (simulated with real prices, not real money), BACKTEST (computed on the past, nothing traded) or REAL (according to your broker's statement). Next to the figure, not in a footer.
Below the sample, no percentages. The verdict is "it can't be known yet", with how many trades so far and how many to go.
No language model in the execution loop. No LLM decides an order, a size or a number. The daily message is composed by code from 36 fixed sentences.
Stale data is called stale. Older than 60 seconds, it is "data as of 17:00", never "live".
No outage is omitted. Every data or broker outage appears with start, end and duration; it is also stated when there was none.
Concentration is declared. "75% of the gain came from a single day" is said even when the gate has been passed.
The broker is the source of truth. Reconciliation differences are shown, not hidden.
No future tense, no promises. Passing the gate does not guarantee future gains, and the product says so every time.
Every number opens down to the query and the rows that produce it. Sentence, figure, evidence: three reading levels.

Read the full methodology

Is it for me?

It depends on how you work. Pick your case.

I have a rule and I don't code

Write it in your own words. StrategyCheck gives it back in plain language, line by line, so you can fix whatever is not what you meant. Then it runs it in paper and sends you one message a day. US stocks and ETFs, daily or slower decisions.

I have a Python script

Paste it or install the SDK. The validator flags what to change (lookahead, dependencies, time per cycle) and your code runs in an isolated container with no network. You can open any decision and see the data it was made with. Open-source SDK, MIT license, code in the coming months.

I trade with a prop firm

Not yet. Practicing your prop firm's rules in paper is on the plan, with no date. If you need it, ask for it: we count requests and write to you only when there is news about this.

Request it

Who it is not for

If you want to get rich quick. StrategyCheck measures whether a strategy works; it does not make it work.
If you want to scalp or trade at high frequency. Intraday exists only in discipline mode, with mandatory brackets and a cap on trades per day.
If you expect signals or to be told what to buy. It does not do that, and it never ranks strategies by expected return.
If you want to hand over your money. It has no custody and cannot withdraw funds; it executes rules you set.
If you trade futures, crypto, forex or options. They are not available today.
If you need financial advice. StrategyCheck is not an investment adviser.

Pricing

No plan can be bought today. Payments start when "trade for real" exists. The only thing you can do now is sign up to be notified.

Paper

US$ 0 forever

Test and measure one strategy. No card.

1 strategy in paper, with rules or with Python code (SDK)
Daily message
Edge report with n and labels
Replay of every decision
History audit from CSV
Email and Telegram alerts

Join the waiting list

Operator

US$ 19 per month

or US$ 190 per year (two months free). Trade one strategy for real.

Everything in Paper
Trade 1 strategy for real with 1 broker (native Alpaca or bridge)
The gate and the risk guard in real
Reconciliation with your broker

Notify me when it's ready

Available when "trade for real" is enabled for you.

Pro

US$ 49 per month

or US$ 490 per year (two months free). Trade and measure several strategies.

Everything in Operator
Up to 5 strategies
Prop-firm mode (no date)
API (no date)

Notify me when it's ready

Same as Operator.

Prices in US dollars. Taxes are computed for your country at checkout. Once billing exists: you cancel from Account and plan in two clicks and keep access until the end of the paid period. No calls, no mandatory questions.

FAQ

Does StrategyCheck trade for me?

It does not decide for you. It executes the rule or the script you wrote, in paper, and logs every decision with the data that produced it. It does not pick what to trade, does not change your strategy and does not rank strategies by expected return. Trading for real, when it exists, is your decision, and the system only enables it after the gate.

Is it financial advice?

No. StrategyCheck is software. It is not an investment adviser or broker-dealer, does not provide signals, does not recommend what to buy or sell and does not hold money. It measures whether a strategy you wrote covers its costs, and says so with the number of trades in view.

How long does it take to say whether a strategy works?

It depends on the horizon. Intraday in discipline mode needs 100 trades and 40 market days; swing, 40 trades and 60 days; position, 20 trades and 120 days; rebalanced portfolio, 4 rebalances and 180 days. Before that sample the verdict is "it can't be known yet" and no percentages are shown.

Which brokers does it work with?

In the beta: Alpaca through OAuth authorization (paper only; trading for real depends on an Alpaca Compliance approval that does not exist yet), Interactive Brokers through the open-source strategycheck-bridge connector, which runs on your machine, and any broker for measurement by importing a trades CSV. Tradovate and prop-firm rules are not available yet.

What happens to my code?

Your Python script inherits from an SDK class and returns intents, not orders. It runs in an isolated container with no network and with CPU, memory and time per cycle limited. A validator checks it first: syntax, allowed dependencies, lookahead and time per cycle. The SDK will be open source under the MIT license; the strategycheck and strategycheck-bridge packages are reserved on PyPI and the code is published in the coming months.

How much does it cost?

Paper: US$ 0, forever, with one strategy. Operator: US$ 19 per month. Pro: US$ 49 per month. The annual plan includes two months free (US$ 190 and US$ 490). No plan can be bought today: payments start when "trade for real" exists. The only thing you can do now is join the waiting list.

What is the difference between PAPER, BACKTEST and REAL?

BACKTEST: computed on past data; nothing was traded. PAPER: simulated with real market prices and modeled costs; not real money. REAL: trades in your account at a broker, according to its statement. Every figure carries its label, and paper and backtest results guarantee nothing in real.

Why does it say "it can't be known yet"?

Because with few trades a result looks more like luck than like a method. Until the sample for its horizon is complete, StrategyCheck shows how many trades there are and how many are missing, and shows no win rate, no average and no verdict. It is the honest answer, not a failure.

What happens to my money and my keys?

StrategyCheck does not hold your money, cannot withdraw it and never asks for your broker password. With Alpaca you authorize from Alpaca's site and can revoke the permission at any time. With the bridge, the credentials stay on your machine. Every order passes through risk limits that no text can override.

When can I use it?

The private beta is planned for February 28, 2027. Today the engine runs in production with a single strategy in paper, the product's own test account. There is no public sign-up: you can leave your email on the waiting list and we will notify you once when it opens.

Waiting list

We will let you know when the beta opens. Nothing else.

Changelog

Date	Version	What changed
2026-11-02	site 0.1	Landing, methodology v1 and author page published. Unchanged: still no public sign-up and no CSV audit.
2026-10-30	engine	First automatic daily close of the product, on the test account, in paper. Verdict: insufficient sample (0 closed trades).