Telegram trading signals, understood and executed on MetaTrader in real time.
MarketSync reads the Telegram channels a trader already follows, parses each signal with a fast-path parser backed by a language-model fallback, maps it to the right broker symbol, applies that account's own risk rules, and places the order in the cloud. No VPS and no forwarding bot.
Most signals arriving from a channel are already in a shape the system recognises: a symbol, a direction, an entry, a stop and one or more take-profit levels. Those go down a fast path, a deterministic parser that resolves them without calling a model at all. It is cheaper, it is faster, and it cannot hallucinate.
The system watches channels, groups and forum topics a trader is already subscribed to, so nothing has to be forwarded into a bot first. When a provider edits a message after posting it, the edit is reprocessed rather than ignored, because a corrected entry price is often the difference between a good fill and a bad one.
The awkward ones fall through to a language model, which is asked for strict JSON against a fixed schema rather than free text. Output that fails to parse is retried, and a concurrency cap stops a burst of thirty signals in one minute from turning into a cost or latency spike.
Anything that still cannot be resolved is logged as unparsed rather than guessed at. A missing stop loss is not something to improvise.
Sixteen locales are supported, so a signal posted in Turkish, Urdu, Arabic or Portuguese is handled the same way as one posted in English.
A signal is treated as a proposal, not an instruction. Each account carries its own configuration profile, and that profile decides what actually gets placed: lot sizing, which take-profit levels to act on, whether to split a multi-entry signal across several orders, how trailing stops behave, and how much spread compensation to allow.
Two accounts can follow the same provider and end up with completely different positions, which is the point. The trader's rules outrank the signal every time.
MetaTrader is the main target, alongside DXtrader and TradeLocker, each behind its own adapter so supporting a new platform is an adapter rather than a rewrite.
Broker rejections are treated as part of the normal flow rather than an error to swallow. A no-quotes response, for example, triggers a recovery path instead of failing silently, because a trader needs to know whether a position actually exists.
The usual way to automate MetaTrader is to rent a VPS, install a terminal on it and hope it stays up. That makes uptime the trader's problem and puts a Windows box between them and their money.
MarketSync runs execution as a service, so there is nothing to install and nothing to keep alive. That was a product decision as much as an architectural one: it removes the most common reason people abandon signal automation.
The trading path is roughly half of it. The backend also carries multi-gateway subscriptions and billing, a referral system, a signal marketplace, a paper-trading simulator for testing a provider before risking real money, white-label support, and an AI support assistant with its own caching layer.
Around 450 API routes sit behind a NestJS backend, with a Next.js front end and real-time updates in all sixteen supported languages.
A cloud-based copy-trading platform running on a backend of around 450 API routes, live at marketsync.ai and serving traders worldwide.
| Signal sources | Telegram channels, groups and forum topics |
|---|---|
| Parsing | Deterministic fast path, language-model fallback with strict JSON |
| Languages | 16 locales |
| Execution platforms | MetaTrader, DXtrader, TradeLocker |
| Order handling | Multi-entry, trailing stops, per-account lot sizing, spread compensation |
| Risk model | Per-account configuration profiles |
| Backend | NestJS, around 450 API routes |
| Front end | Next.js with real-time updates |
| Also includes | Subscriptions, referrals, signal marketplace, paper-trading simulator, white-label |
Not with MarketSync. Execution runs as a cloud service, so there is no terminal to install and no virtual machine to keep online. The usual VPS setup makes uptime the trader's own responsibility, which is the most common reason signal automation gets abandoned.
MarketSync supports sixteen locales, including Turkish, Urdu, Arabic, Hindi, Portuguese and Polish. Signals are parsed on structure rather than language, and a language-model fallback handles phrasing the deterministic parser does not recognise.
The edit is reprocessed rather than ignored. Providers often correct an entry price or a stop level seconds after posting, and treating the first version as final would put the trade on the wrong side of the correction.
Yes. Risk lives on the account, not on the signal. Each account has its own configuration profile controlling lot sizing, take-profit handling, multi-entry splitting, trailing stops and spread compensation, so one signal can produce different positions per account.
A trading community with an AI coach that actually knows your numbers.