Two proprietary indicators. Seven asset classes. Six brokers. One operator cockpit — with a kill switch that flattens every position, in every account, in under three seconds.
The intelligence of the platform is two proprietary indicators, ported to Python and validated against live TradingView charts. One reads structure. The other reads order flow. A trade requires both to agree.
A multi-component confluence engine that combines many independent technical signals into a single graded decision. Emits tiered signals with three staged take-profits and an adaptive market-condition gate.
A cumulative-delta-volume engine that reads the balance of buying versus selling pressure bar by bar. Produces a weighted 0–20 order-flow score with graded tiers and an exhaustion measure that flags when a move is running out of participation.
Same candles, same configuration, same result — no wall-clock dependence, no randomness. Every signal walks eight gates before it becomes an order, and the risk engine has veto at every step.
Market data arrives and is normalised into a canonical bar format.
MASTER and CDV PRO compute on closed candles only.
Signals combined into a graded decision with tier and metadata.
Volatility regime and macro-bias suppress signals fighting context.
Immutable object: entry, three take-profits, stop-loss, metadata.
Position size from capital & stop distance. Limits verified.
Signal distributed to every eligible account, quantity scaled.
Bracket order placed, staged exits and trailing stop managed to close.
Nine independent services, each owning one responsibility, coordinating over Kafka. Any piece can scale, fail, or be replaced without taking down the rest.
Risk is a first-class subsystem, not an afterthought. Every uncertain state — a broker disconnect, an unclear fill, a stalled feed — resolves to not trading rather than trading blind.
Enforced in every mode, on every order, without exception.
Size is a function of each account's capital and the entry-to-stop distance — same stop everywhere, only quantity varies.
Aggregate limits verified before every fan-out. Over-limit trades are rejected, not clipped.
Any account hitting its daily loss limit halts. Others continue independently.
Broker disconnected, DB write fails, feed stalls — the system holds fire until the state is provably good.
Triggered from dashboard (two-step, 2FA), Telegram, or automatically on a daily-loss breach. Flattens all positions, halts all new signals, stays halted until a human restarts. Tested at < 3 s across multiple accounts.
A React cockpit with six views — live positions, strategy, risk, accounts, analytics, and system health. Config edits take effect immediately. Execution-affecting actions gated by 2FA.
A single approved signal fans out in parallel to every eligible account, quantity scaled to each account's capital. Adding a broker account is a single database row — no code change, no redeploy.
The recommended path to going live is to start in semi-automatic on a single account and asset class, then graduate to automatic only after the system has proven itself with real money under supervision.
Full pipeline runs against real market data but places simulated orders. No capital at risk.
Signals generated and surfaced. A human places every order.
System prepares each trade and waits for a human to approve or reject via dashboard or Telegram before executing.
Approved signals execute hands-free, end to end. Reached only after supervised proof.
Every planned milestone of the build is complete and paper-verified. Going live with real capital is intentionally a NO-GO until an operator explicitly approves it — a deliberately human decision against a go/no-go readiness report.
Tools chosen for reliability, not novelty. Nothing in the critical path is speculative.
Datareco builds production data and execution infrastructure for systematic trading. VoltaEdge Algo is our flagship product — a full trade lifecycle in a single deployable, wrapping a discretionary trader's real intuition in institutional-grade guardrails.
We do not build signals we do not understand. We do not deploy code we cannot roll back. And we do not go live with real money without a human pulling the go/no-go trigger.
Neither a signal-alert tool nor a backtesting notebook — it's a complete trade lifecycle. Market data in, signals generated, risk checked, orders executed and managed, P&L reconciled, all observable and controllable from an operator dashboard, with an emergency stop that flattens every position in every account in seconds.
Not in v1, by design. The intelligence is two deterministic technical indicators. What is built is the framework for AI — a plugin system with sandboxing and shadow-mode evaluation, proven with a placeholder plugin — but the ML classifier, news-sentiment model, and order-flow-imbalance model are explicitly future scope. Hooks ready, models not built.
Signals are computed only on closed candles. Once fired, a signal never changes retroactively. That's stricter than the source TradingView indicators, which update intrabar. We trade slightly more conservative timing for the guarantee that the system never acts on something that later vanishes.
One approved signal fans out in parallel to every eligible account. The stop distance is the same everywhere; the position size is scaled to each account's own capital. Accounts too small to take a valid position are skipped automatically.
All open positions are flattened via market orders across every account. All new signal processing halts. The system stays halted until a human manually restarts it. Tested at under three seconds across multiple accounts. Triggered via dashboard (two-step, 2FA), Telegram, or automatically on a daily-loss breach.
Zerodha as the primary adapter, plus Upstox, Angel One, Interactive Brokers, Binance, and MetaTrader. Adding another broker is a single adapter behind a stable gateway interface — the strategy code doesn't change.
No. Every planned milestone is complete and paper-verified — a 105-trade multi-asset paper run with rupee-level reconciliation, a 189-trade six-month backtest matching manual P&L with zero deviation, 268 tests passing. Live trading with real capital is intentionally a NO-GO until an operator explicitly approves it against a readiness report. That final decision is deliberately a human one.
We're onboarding a small cohort of pilot partners ahead of live capital rollout. Tell us a little about your setup and we'll come back within 48 hours.