How to install AlgoRover Research: cloud and local guide

Choose cloud or local execution

Cloud Research runs historical backtests on servers without a local backtesting runtime. The cloud plugin currently supports Codex, with published strategies, available presets and markets, and 90- or 365-day backtests. The cloud workflow is explained first below.

The local edition runs research on your computer using Codex or Claude Code. See the local section below. Installation requires no registration; backtesting requires an AlgoRover product login. Neither workflow requires exchange accounts or exchange API keys for backtests.

Step 1: Install and connect cloud Research

In a Codex CLI version that supports plugins, run these three terminal commands in order. The last command starts AlgoRover sign-in and authorization.

If plugin is not recognized, check that your Codex CLI supports plugins. Consult the official installation guide for current requirements.

codex plugin marketplace add algorover/algorover-research
codex plugin add algorover-research-cloud@algorover-research-github
codex mcp login algorover-research --oauth-client-registration dcr --scopes research:read,research:run

Step 2: Sign in and authorize

When the browser opens, register or sign in to your AlgoRover product account as directed. After Google sign-in, return to the Research authorization page and select Allow. Then return to Codex and open a new chat.

research:read retrieves your jobs and reports; research:run starts or cancels backtests. Sign-in and plugin authorization are separate steps. Complete authorization before continuing.

Step 3: Find a strategy and choose available options

Ask Codex to find published strategies and show available versions, presets and markets. Use options actually offered by the catalog. You can also choose a strategy on the website and tell the AI its name or identifier.

Use AlgoRover Research to find published BTC strategies. List their strategy identifiers, versions, available presets and markets. Let me select a strategy first; do not start a backtest.

Step 4: Preview the setup, then confirm the run

Ask for a 90-day preview, or specify 365 days. Check the strategy identifier, version, preset, market, dates and other displayed settings. Request another preview if changes are needed.

The request below only previews the setup. After checking it, say: I confirm the setup you just showed; start this cloud backtest. Confirm after reviewing the settings.

Preview a 90-day cloud backtest for the strategy I selected. Show its identifier, version, preset, market, date range and available configuration details. Do not run it until I confirm.

Step 5: Check progress or cancel a running job

Keep the job ID after starting. Ask the AI for current status, specifying the ID when you have multiple jobs. Use the same AlgoRover account to view research history on the website.

To cancel, explicitly ask: Cancel the running backtest with job ID … and check its resulting status. The job page shows the final status.

Check the AlgoRover Research backtest I just started. Show its job ID and current status. If it is unfinished, explain how to check its progress again.

Step 6: Review performance and download results

After a job succeeds, review performance and fills, then download the Chinese-language report and configuration from Codex or My backtests on the website. Reports are currently in Chinese even when you use an English or Korean interface.

Reports remain available until the date on the job page, for up to 30 days. Request a new download link while the report is retained if the previous link expires. Download needed files before report expiry. Historical results are for research and do not predict future performance.

Retrieve my successful backtest, show performance and fills, and provide downloads for its Chinese-language report and configuration. If a link expired, check whether the report is still retained and request a fresh link.

Local execution: install Research

Select Local execution on the Research product page. Copy the installation prompt into Codex or Claude Code, or expand Install with terminal commands and use the matching commands. Local installation differs from the cloud plugin commands.

Follow the AI and official guide to prepare your local environment. Confirm installation is complete before researching.

Add the plugin marketplace https://github.com/AlgoRover/algorover-hub and help me install the local edition of AlgoRover Research. Check the official instructions and current environment requirements first, then guide me through installation.

Local execution: sign in and connect

After installation, ask the AI to start using AlgoRover. Open the Dashboard workspace as directed, register or sign in, then return to the conversation to check your connection. Product login is separate from exchange access; backtesting needs no exchange keys.

Start using the local edition of AlgoRover Research. Check installation and account connection. If sign-in is needed, open the corresponding Dashboard and guide me through registration or sign-in in its workspace, then verify the connection.

Local execution: review and confirm a backtest

Query strategies and the presets, markets and data settings supported by your local version. Ask the AI to show the strategy version, dates, parameters and cost assumptions, then explicitly confirm before running. Cloud 90/365-day options are not fixed limits for the local edition.

Use local AlgoRover Research to find an available BTC strategy and list supported presets and markets. After selection, show the version, test dates, parameters and cost assumptions. Wait for my confirmation before running locally.

Local execution: find results and files

Ask for the completed results and actual local paths to reports, charts and inputs. You manage local files on your device; cloud report retention does not apply.

For failed runs, retain the error and follow guidance for your local version. Cloud job cancellation commands do not manage local processes.

Review my completed local backtest, explain the main results and cost assumptions, and show the actual local paths to reports, charts and input files.

Manage cloud data and get support

Sign out, revoke plugin access, delete an individual report, or select Delete my Research data under Backtest status. Confirmed deletion requests receive a receipt. Cloud data can still be managed on the website after uninstalling the plugin.

For installation, authorization or backtest issues, email agent_strategy@algorover.ai with the time, Codex version, job ID and error message. Do not send passwords, verification codes, tokens or exchange API keys. Send privacy requests to lily_wang@algorover.ai.