Getting started
Install Wallbot, start it with the launcher, create your dashboard password and learn the strategy on the built-in simulator. Then move to Binance paper trading on real market data.
Install
Wallbot is a Python program with a web dashboard. On Windows, open a terminal in the Wallbot folder (for example C:\wallbot) and create its private Python environment once:
cd C:\wallbot
python -m venv .venv
.venv\Scripts\pip install -r requirements.txt
That is all the installation there is. Nothing is installed system-wide, and no exchange account is needed yet.
Note. On a Linux VPS the same steps use python3 -m venv .venv and .venv/bin/pip. Keep the dashboard on 127.0.0.1 and reach it through an SSH tunnel, for example ssh -L 8787:127.0.0.1:8787 you@your-server (see Viewing the dashboards remotely).
On a Linux VPS: wallbot.sh
On a VPS, wallbot.sh runs each instance in its own tmux session, under the launcher described below (start_wallbot.py):
./wallbot.sh start live # Trading instance, port 8788, tmux session "wallbot-live"
./wallbot.sh start research # Research instance, port 8787, tmux session "wallbot-research"
./wallbot.sh status # sessions, CPU and memory per instance, disk use, server memory
./wallbot.sh attach live # watch the console; detach with Ctrl+B then D
./wallbot.sh stop live # clean stop; waits until the bot has exited
Everything lives under one folder, for example /wallbot: the code, the Trading data in /wallbot/data_live, the Research data in /wallbot/data and the log files in /wallbot/logs (wallbot_live.log and wallbot.log). start refuses to run a second copy if the session or the port is already in use.
Add wallbot.sh boot as an @reboot crontab entry. After a server restart it waits a minute, then starts every instance whose data folder is on that server. The Trading instance starts with new entries paused (reboot_start_paused), and you get a high-priority "VPS rebooted — Wallbot started paused, check positions" notification. Only "Resume entries" lifts that pause.
Start it with the launcher
Double-click start_wallbot.bat, or run it from a terminal. It starts the Research instance on port 8787 and writes everything the bot prints to logs\wallbot.log. When the log file reaches 50 MB (log_rotate_mb) the launcher renames it to .1 and starts a new one; 5 old files are kept (log_keep).
The launcher watches the bot for you:
- If the bot exits unexpectedly, the launcher waits 10 seconds and starts it again (
launcher_restart_wait_s). - It restarts at most 5 times in any hour (
launcher_max_restarts_per_hour). After that it gives up and tells you, because something is wrong that a restart will not fix. - Each unexpected exit sends a "BOT EXITED" alert by email and push, if you have set them up. The email includes the last 30 log lines with secrets removed.
- Ctrl+C in the launcher window stops everything cleanly, without an alert. So does Ctrl+Break on Windows (
launcher_ctrl_break), and SIGTERM or SIGHUP on Linux, for example when the terminal or tmux session closes (launcher_posix_signals).
To keep it running after you close the terminal, start it in its own window:
start "Wallbot" start_wallbot.bat
The Trading instance has its own launcher, start_wallbot_live.bat, on port 8788. You do not need it to get started. See Instances and modes.
Keep the PC awake
If Windows sleeps, the bot stops and recording has a gap. On mains power you can turn sleep off:
powercfg /change standby-timeout-ac 0
powercfg /change hibernate-timeout-ac 0
powercfg /change monitor-timeout-ac 10
On a laptop, also set "When I close the lid: Do nothing" and keep it plugged in. Set Windows Update active hours so it does not restart during the day.
First login
Open http://127.0.0.1:8787 in your browser. The first time, the page says "First run. Choose a dashboard password (at least 10 characters). It protects your exchange keys." Pick a strong password and press "Create password".
After that the page asks you to "Sign in". The password is stored only as a secure hash, and login attempts are rate-limited. Each instance has its own password: the Trading instance on 8788 asks you to create a separate one the first time you open it.
Warning. The dashboard password protects your encrypted API keys. Do not reuse a password from anywhere else, and do not expose the dashboard to the internet.
Learn on the simulator
A new database starts on the simulator: Settings → Account & keys → Exchange shows "Simulator (offline demo)". It needs no account and no internet. It generates fake order books with both real walls and spoofed walls, so you can watch every part of the strategy happen quickly:
- walls appearing on the depth chart, with a ring that fills while each wall proves it is not a spoof;
- entries placed just in front of confirmed walls;
- pending entries cancelled when a wall is pulled;
- stops hit when a wall truly breaks, and trailing stops locking in profit;
- the daily loss limit pausing entries.
Spend some time here with the Dashboard tour open. The simulator says nothing about whether the strategy makes money on a real market. It only shows you how the bot behaves.
Switch to Binance paper trading
Paper trading on Binance uses Binance's real order books and trades, with simulated fills. No money is at risk and no key is required.
- Open Settings → Account & keys.
- Under Mode, make sure "Paper" is selected.
- Set Exchange to "Binance USDⓈ-M" and press "Save changes".
- Watch the top bar: "Exchange" should read Binance and "Feed" should change to "streaming".
- After the first scan, the Coins & scanner table fills and coin tabs appear above the depth chart.
Optionally, save a Binance key in the Live slot. In paper mode it is used read-only to read your real leverage brackets and fee rates; any request that could change your account is refused by the bot. Without a key, paper mode assumes one leverage bracket and default fees, and the dashboard says so. If you do save one, create it with withdrawals disabled and restricted to your IP. See Credential slots.
Paper mode starts with a simulated balance of paper_balance (100 USDT by default).
Optional: record market data
With record_market_data on, the Research instance saves the order book and trades of watched coins to backtest_data\live\ for later backtests. Plan for 1 to 2 GB per day with the default number of watched coins. The top bar shows the disk used.
On a VPS started with wallbot.sh, the disk guard (disk_guard) keeps everything under the Wallbot folder within a budget of 20 GB (disk_budget_gb). When raw recordings pass 15 GB (disk_recordings_budget_gb), the oldest recording days are deleted first, never today's. Near full, recording pauses; trading never does. The guard only runs where the disk-budget folder (WALLBOT_DISK_ROOT) is set, which wallbot.sh does. On a Windows PC nothing is measured or deleted.
Getting-started checklist
- Created
.venvand installed the requirements. - Started
start_wallbot.batand openedhttp://127.0.0.1:8787. - Created a dashboard password of at least 10 characters.
- Watched the simulator long enough to see an entry, a spoof cancel, a stop and a trailing exit.
- Switched Exchange to Binance in Paper mode and saw "Feed: streaming".
- Read the trade cards and the Activity log while paper trades happen.
- Set up notifications (Notifications) and sent a test.
- Read Safety rules and Daily operation before going further.
- Only after weeks of paper results: set up the Trading instance and testnet (How you switch).