See what your MT5 server is doing. Without logging in.
Stedwatch is a read-only monitor that runs on the Windows VPS itself. It watches the machine and every MetaTrader 5 terminal, keeps the history, and sends you a Telegram message if something stops running. One glance replaces the RDP check you do several times a day.
Stedwatch is a Windows app that runs on your VPS. Open this page there to download it.
The real app, sample data. Light and dark, like this page.
Every terminal's real status, at a glance.
Running, connected, orders blocked: one row per terminal, straight from the process, the OS and the journals. The window can look fine while the process is gone, the broker link is dead or every order bounces - Stedwatch reads the signals that cannot pretend.
- Per-terminal status, named honestly: connected, needs attention, not running, unknown
- Uptime, CPU and RAM per terminal, average and max
- Broker ping per terminal, with the age of the reading
Is this box strong enough for peak hours?
Forum rules of thumb cannot answer that; your own peaks can. Stedwatch reads the same counters Task Manager uses and keeps the peaks instead of averaging them away. Add a terminal, watch the next busy session, decide with numbers from your own server.
Average AND max, over the window you pick. The peak is the number that decides.
Windows rebooted overnight. Did everything come back?
The most expensive reboot is the one you learn about from your P&L. Stedwatch checks, per terminal, whether an auto-start is actually configured, and names the one that is not - before the next reboot, not after.
Honest tradeoffs stay your call: the tool reports, you decide. Nothing is changed automatically.
Stop logging in just to check.
If everything runs, Stedwatch stays quiet. When a terminal dies, AutoTrading flips off or the broker connection drops, a message with the terminal's name lands in your Telegram. Per-alert toggles and quiet hours keep it calm; a mobile app with native push is coming soon.
- Terminal gone or crashed, an EA blocked by AutoTrading, broker connection lost
- Unexpected reboot, disk running low, drive health warning
- Every event also lands in the dashboard's event list, timestamped
Sample Telegram alerts, exactly as they arrive:
The same events, timestamped in the dashboard's event list.
Every alert states when it fires and why it matters. You choose; urgent ones always come through.
See who tried to sign in, and whether it matters.
Bot noise on an internet-facing RDP is normal, and Stedwatch says so instead of crying wolf. What matters is whether any of it can get in, and you see exactly that.
- Failed sign-in attempts by hour, with top usernames and source IPs
- The checks that actually protect you: RDP port, NLA, firewall, disabled accounts
- Windows records attempted usernames, never passwords, and the page says so
Find the terminal that eats your RAM.
RAM that grows week over week, GDI objects creeping toward the 10,000 quota, a disk aging toward failure: slow problems that only show up if someone keeps the history.
- Server RAM over time, average and peak - growth is visible, not a feeling
- RAM average and max per terminal in the terminals table: the leak has a name
- GDI and USER objects against their real quotas, and SMART disk health, read as SYSTEM
Everything it watches
The complete list, honestly scoped. If a value cannot be proven from outside, it is shown as "unknown", never guessed.
Every MT5 terminal
- Process alive, with crash and clean close told apart via the Windows event log
- CPU and RAM per terminal, average and max, over time
- Broker connection and ping, with the reading's age
- AutoTrading state, and the check it flips off on
- EA errors read from the journals (retcodes, init failures)
- Uptime and auto-start state after reboots
The server itself
- CPU, RAM, disk and network, with history
- Disk SMART health, wear and predicted failure
- Failed sign-ins and the security checks that matter
- GDI and USER objects against their Windows quotas
- Reboots, internet outages, pending updates
- Stedwatch's own footprint, shown honestly
Checks that explain themselves
Not just red and green dots: every check tells you what was measured, why it matters and what you can do. Setup, security and performance, in plain language.
Will everything come back after a reboot? Now you know before it happens.
More than one server? All of them in one view, over your own network or VPN.
And a built-in guide in plain language: why read-only, when to add RAM, what the pills mean.
Running in about five minutes
One exe, a short wizard, no account needed on the server. The 90-day beta key unlocks everything; when it expires, the app simply drops to the free tier. Monitoring never stops.
The real installer. It says what it does, does exactly that, and nothing else.
Where the beta stands
- July 20, 2026 · Open betav0.8 released: free 90-day keys, everything included.
- Now · This is the startWe ship improvements continuously. Your feedback lands directly in the next update, delivered over the app's signed auto-update.
- Coming up · More on the wayThe app keeps growing during the beta - the mobile app with native push is on the list. Every release lands here.
- Honest timing wording throughout the app. The dashboard, settings, setup wizard and the generated report now state the real check interval instead of a faster one, drive-health (SMART) checks are named as the weekly sweep they actually are, and the security re-check no longer calls itself instant.
- The frame-rate probe now runs once per session instead of every 30 seconds: its own two-second measurement window was itself forcing extra rendering while you scrolled.
- Remote-session speed, round three: chart tooltips are now drawn inside the chart itself instead of as a separate floating window - on RDP, moving that window used to cost a full repaint on every mouse step. Text also switches to grayscale smoothing in remote sessions (what fast remote apps do), which rasterizes and compresses much better.
- The app now measures its own achieved frame rate while you scroll and writes it to the performance log, so slowness reports come with hard numbers.
- Fixed: with a beta (or Pro) key, expanding a terminal row now shows the full MT5 health detail it promises - AutoTrading with who turned it off, broker connection and ping, account and build, the EAs seen in the logs, journal errors with a benign-rejection count, and the data folder. It used to show the Pro teaser even on licensed installs.
- Big performance fix for high-resolution remote sessions (for example Retina RDP from a Mac): the app's cached surfaces are now kept at the session's real pixel resolution, so every frame is a plain 1:1 pixel copy instead of a software upscale of the whole window. Text and charts also get crisper.
- Remote-session performance: each chart is now rasterized once into a real image and repainted as a plain pixel copy, so hovering and scrolling the dashboard no longer re-draws chart geometry in software.
- Honest update dialog: it no longer promises an admin prompt, because Windows skips that prompt on many servers depending on UAC settings.
- Chart hover is back everywhere, including remote sessions: charts now draw from a cached image with the crosshair layered on top, so hovering costs almost nothing.
- Honesty fix: the dashboard's "max" figures now always use the same peak the charts show (the true spike, never the flatter average-series maximum).
- The dashboard is fast over RDP: chart hover is a desktop-session feature now (each crosshair step re-rendered a whole chart in software), and the dashboard refreshes its charts every few minutes on remote sessions instead of rebuilding every minute - status changes and new events still appear immediately.
- A proper update dialog replaces the plain Windows popup.
- Smooth over RDP: every card now renders from a cached bitmap on remote sessions, dragging the scrollbar scrolls live again, and text uses the faster rendering mode.
- Charts decimate to their point budget (spikes always survive) and caches render at 1x on remote sessions.
- One-click updates: the update banner installs directly after a single confirmation, with live progress right in the banner.
- Read-only monitoring for Windows trading VPSes: CPU, RAM, disk and SMART health, security posture, and every MT5 terminal (AutoTrading state, broker connection, EA errors from the journals).
- Telegram alerts with per-alert toggles and quiet hours.
- Native dashboard with history and a multi-server view, in one self-installing app.
- Signed, health-gated auto-updates that roll back automatically if a build misbehaves.
- Honesty rule throughout: "unknown" rather than a guess, and strictly read-only.
Every release from here on lands in this list and reaches installed apps through the signed update feed: open Settings, Update in the app to see your version.
Questions you should ask
Is it safe to run next to live trading?
Stedwatch is read-only by design: it watches and reports, and there is no command or configuration-write endpoint against the monitored system, MT5 or your trading. It cannot start, stop or trade anything. The dashboard even shows Stedwatch's own CPU and RAM footprint on your server, so you can judge it yourself.
Which brokers and terminals does it work with?
Any MetaTrader 5 terminal on Windows, broker-independent: it watches the process and reads the terminal's own journals. Non-English terminal languages are handled with language-independent anchors (retcodes, symbols); where a localized phrase cannot be proven, the value shows "unknown" instead of a guess. If your terminals run in another language, telling us which is genuinely useful beta feedback.
How many servers and terminals can I watch?
All terminals the server runs, and more than one server in the same view over your own network or VPN. The beta key covers everything, fair use. MT4 is not supported: MT5 only for now.
Does my data leave the server?
Your monitoring data stays on your VPS. There is no Stedwatch cloud and no account required to run the app. Alert texts you enable travel through Telegram to reach your phone; that is your choice and your channel.
What happens when the 90 days end?
The app drops to the free tier and keeps monitoring: live dashboard and about 48 hours of history, one server. Pro depth (per-terminal detail, long history, alerts, SMART and security views) pauses until you decide. Nothing locks up, nothing nags.
Why does Windows warn when I run the installer?
The beta exe is not code-signed, so SmartScreen shows its standard warning for unknown publishers: "More info", then "Run anyway". Every update the app installs is hash-verified and signature-checked against the release feed before it is applied.
What does it honestly not do?
It does not restart terminals, does not manage or copy trades, and does not read your strategies or account balance. It also cannot see a terminal that hangs but keeps its process alive - no outside tool reliably can, and the ones that claim to false-alarm on every idle weekend (here is why). Values it cannot prove from outside (like encrypted terminal settings) are shown as "unknown" instead of guessed. The mobile app with native push is coming soon and is not available yet.
Found something broken? Perfect.
That is exactly what the beta is for. Tell us what is wrong, confusing or missing: reply speed is our thank-you.