Execution platform
TraderOps
Your backtest and your live trades finally agree.
Write a strategy once and it behaves the same whether you are backtesting it, paper-trading it, or trading it live with real money. One engine, eight Indian brokers. Going live doesn’t mean porting anything or rewriting anything — you switch the mode and the same code keeps running.
At a glance
- Market
- India
- Broker integrations
- 8
- Execution modes
- Backtest / Paper / Live
- Backtest ↔ live parity
- Paisa-exact
- Control surfaces
- Telegram + AI clients
The shift
What it replaced
Before this existed
A broker terminal for orders, a mismatched backtester for research, and spreadsheets for risk — backtests that fell apart the moment they went live, and stop-losses managed by hand across every account.
After
One engine where a no-code strategy is written once, checked against paisa-exact parity, and run live across eight brokers — driven from Telegram or an AI client, with stop-losses and a kill switch that don’t wait on you being at the screen.
What you can check
- 8 brokers
- Indian markets
- One engine, three modes
- 50+ active traders
Fit
Where it fits, and where it does not
Traders who already have a strategy
This automates the execution of logic you have already decided on. It won’t invent strategies for you, and it has no view on which ones are any good.
Anyone whose backtest stopped matching live
The parity is the reason it exists. If your research and your execution run on different code today, that gap is the exact thing it closes.
Multi-broker and multi-account operators
One strategy across several brokers, no per-broker branching, and one place where limits and kill switches apply to all of them at once.
Not built for
- Anyone after signals, tips or a strategy marketplace
- Discretionary traders who place every order by hand
- Traders outside India. Korea, Japan, the US, Canada and Brazil are planned, and none of them is live yet
- Brokers we have not integrated — there are eight, and they are all named further down this page
In the wild
What people do with it
Intraday index traders
One opening-range setup, several accounts
Configure the breakout once, allocate different capital per account, and let the scheduled square-off close everything at the end of the session rather than you doing it three times.
Options desks
Multi-leg positions that stop themselves
Legs go on together and the stop-loss sits on the position the strategy intended, so a bad fill on one leg does not quietly move where the whole thing exits.
Researchers moving to live
The month between backtest and real money
Run the strategy in paper mode against the same engine that will trade it. What you are checking is execution behaviour, not the logic — the logic is already identical.
Anyone away from the screen
Trading from a phone or an AI client
Check positions, pause a strategy or square off from Telegram, or connect Claude or ChatGPT and ask it in plain language. Every order-placing action still comes back for a confirm.
Architecture
How it holds
The trigger prices — stop-loss, target, trailing — are pinned to what the strategy intended, not to the price you actually got filled at. That is the core idea: the trigger ladder comes out identical in backtest, paper and live, and slippage only ever moves the fill. Underneath it, a single reconcile loop is the only thing allowed to write trade state, so order polling and events can’t race and end up disagreeing about a position. Everything that acts on a trade — a Telegram message, a call from an AI client, a webhook, the scheduler — runs through that same audited path. There is one place where a trade’s truth gets decided, and one place to look when you need to know why something happened.
Author once
You build a strategy by configuring it, not by writing code — opening-range breakout, wait-and-trigger, re-entry, multi-leg options. There is no separate step later to make it live-ready, and strategies import and export, so you can move a setup between accounts or hand it to someone else.
Anchor the triggers
Stop-loss, target and trailing prices are anchored to the entry the strategy intended, not the fill it happened to get. That is what keeps the trigger ladder identical across all three modes.
Dispatch through one broker layer
A fixed capabilities model per broker handles the differences — tag formats, product and exchange maps, order semantics — so the engine never has to branch on which broker it is talking to.
Reconcile on one path
One reconcile loop is the only thing that writes trade state. Nothing races, and there is a single place where what happened to a trade gets settled.
What it does
The decisions it is made of
Not a feature list. Each of these is a choice the build turned on, and each one was made against a cheaper alternative that would have worked until it did not.
Paisa-exact parity
Trigger prices are anchored to the intended entry, so backtest, paper and live all produce the same trigger ladder. Slippage moves the fill and nothing else, which is why what you test holds up once it is real.
Broker-agnostic layer
Eight brokers sit behind one capability model, so the engine never branches on a broker’s name. Adding the next one is additive — it doesn’t touch the strategies already running.
Run it from an AI client
Connect Claude, ChatGPT, Codex or any MCP client and manage your trading by talking to it — check positions, start or stop a strategy, place an order. A Telegram bot does the same from your phone. Anything that places or moves an order asks you to confirm first.
Safe by construction
Kill switches you can scope to a single strategy or to everything at once, per-strategy capital limits, and a scheduled square-off for intraday positions. The safety comes from how the engine is built, so it holds without anyone watching the screen.
In practice
The shape of what it produces
trigger ladder · sample
| Mode | Entry trigger | SL trigger | Target trigger | Fill | Drift |
|---|---|---|---|---|---|
| Backtest | 24,780.00 | 24,712.50 | 24,880.00 | 24,780.00 | — |
| Paper | 24,780.00 | 24,712.50 | 24,880.00 | 24,781.15 | +1.15 |
| Live | 24,780.00 | 24,712.50 | 24,880.00 | 24,782.40 | +2.40 |
The same strategy in all three modes. The trigger column is identical by construction; only the fill moves.
Connects to
What it plugs into
Brokers
- Zerodha
- AliceBlue
- Tradejini
- Zebu
- Dhan
- Delta
- Kotak Neo
- Motilal Oswal
Control
- Telegram bot
- MCP server
- Webhooks
Strategy types
- ORB
- Wait-and-trigger
- Re-entry
- Multi-leg options
A new broker goes in through the capabilities model, not through engine changes, so support is additive and the strategies already running are left alone.
Operations
What happens when something goes wrong
Mode-scoped kill switches
Stop one strategy, a whole mode, or everything — without unwinding positions by hand across accounts.
Per-strategy capital limits
A strategy can’t commit more than you have allocated it, no matter what its logic decides to do.
Scheduled auto-square-off
Intraday positions close on a schedule instead of depending on someone being there to close them.
One audited pipeline
Telegram commands, AI-client calls, webhooks and the scheduler all land on the same audited path, so there is one honest record of what happened and why.
Security
How your data is handled
Broker credentials encrypted at rest
The API credentials you connect are stored encrypted rather than in plain text. TraderOps needs them to place orders on your behalf.
EvidenceEncryption scheme and key handling described on request.
Two-step confirm on every order
Anything that places or moves an order comes back for a confirmation first, whether it came from the dashboard, Telegram, an AI client or a webhook.
EvidenceObservable on any order-placing action in the app or the Telegram bot.
One writer of trade state
A single reconcile loop is the only thing allowed to record what happened to a trade, so there is one account of it rather than several that can disagree.
EvidenceTelegram, AI-client, webhook and scheduler actions all resolve through that same audited path.
No audited certification
THWorks holds no SOC 2, ISO 27001, HIPAA or PCI DSS attestation, and TraderOps is not covered by one. If your procurement requires it, we do not clear that bar today.
EvidenceStated here rather than left for you to find in a security questionnaire.
These are mechanisms, not certifications. Ask us anything here and you get the implementation, not a badge.
Displaces
What it takes off the stack
A broker terminal used for order entry
Orders originate from the strategy, not from a human watching a screen.
A separate backtesting tool
The thing that ran your research is the thing that runs your trades.
Spreadsheets tracking risk and positions
Capital limits and stop-losses are enforced by the engine rather than by attention.
Manual stop-loss management across accounts
Mode-scoped kill switches and scheduled square-off replace the bad-day scramble.
Gallery
Frames from the build
FAQ
TraderOps, answered
The questions that come up about TraderOps — what it does, how it is built, and what it takes to run it.
Stop-loss, target and trailing triggers are anchored to the entry your strategy intended, not to the fill it actually got. So the trigger ladder is worked out the same way in backtest, paper and live, and slippage only moves the fill price. That is what keeps results paisa-exact from one mode to the next.
India, today. All eight supported brokers are Indian, and the platform trades what they trade. Korea, Japan, the US, Canada and Brazil are planned, and none of them is live yet — so if you trade outside India, TraderOps is not usable for you at the moment.
Eight right now — Zerodha, AliceBlue, Tradejini, Zebu, Dhan, Delta, Kotak Neo and Motilal Oswal. A broker-agnostic layer normalises how each one behaves, so the same strategy runs on any of them, and adding a new broker doesn’t disturb the ones already live.
No. You compose strategies by configuring them — ORB (opening-range breakout), wait-and-trigger, re-entry and multi-leg options — and you can import and export them to move a setup between accounts. If you would rather, you can drive the whole thing from Telegram or an AI client instead of the dashboard.
Yes. TraderOps ships an MCP server, so you can connect Claude, ChatGPT, Codex or any MCP-capable client and manage your trading in plain language — check positions and P&L, start or stop a strategy, place an order. Anything that places or moves an order asks you to confirm before it goes through.
Keep reading