Workflow
Research, Backtest, and Deploy in Nexus Algo
Learn the separate research, strategy-lab, backtest, deployment, and monitoring workflows in Nexus Algo—from an idea to a reconciled broker position.
Nexus Algo Editorial Team · 22 July 2026 · 10 min read
Research, backtesting, deployment, and monitoring answer different questions. Nexus keeps them connected, but a careful user should not collapse them into one decision.
This walkthrough explains each workflow separately and shows the hand-off between them.
Workflow 1: Research the market context
Research starts with a symbol or market universe and asks whether current evidence supports further investigation. Nexus can combine price and volume behavior, momentum, VWAP, completed candles, market regime, sentiment, and derivatives context.
The output should be read as a thesis with supporting and opposing evidence. A directional label or confidence score is not an instruction to trade.
- Choose the universe: Select the relevant equity, index, or F&O context rather than scanning instruments you cannot execute safely.
- Read the evidence: Review momentum, liquidity, volatility, regime, and signal freshness together.
- Write the invalidation: Define what would make the thesis wrong before choosing a strategy or position size.
Workflow 2: Match a strategy to the thesis
The Strategy Lab maps a market view to an instrument structure. A bullish view may fit a directional long option, a debit spread, or a credit spread depending on volatility, risk tolerance, liquidity, and the intended payoff. Neutral, volatility, calendar, hedging, and advanced structures answer different theses.
Evaluate maximum loss, maximum profit where bounded, break-even behavior, leg count, expiry exposure, and sensitivity to price, volatility, and time.
Workflow 3: Backtest without over-reading the result
- Use enough observations to cover more than one market regime.
- Separate training or tuning periods from validation periods.
- Include realistic fees, spread, slippage, and fill assumptions.
- Inspect drawdown, loss clusters, exposure, and tail outcomes—not only total return or win rate.
- Treat a backtest as evidence about a defined historical simulation, not proof of future performance.
Workflow 4: Configure risk and execution safety
Set account boundaries before enabling deployment: maximum loss, exposure, symbol concentration, open-position count, duplicate-order window, and any strategy-specific controls. These values define what the system is permitted to do when conditions move quickly.
Review the kill switch and the procedure for confirming that it has blocked new orders. Risk configuration is part of strategy design, not a final checkbox.
Workflow 5: Deploy a validated plan
- Refresh state: Confirm the broker session, quote timestamp, funds, positions, and existing orders.
- Review the plan: Verify symbol, side, expiry, strikes, legs, product, quantity, entry style, and protection.
- Pass prechecks: Let Nexus evaluate risk, exposure, margin, duplicates, liquidity, strategy quality, and any option-specific gates.
- Authorize: Use the manual, semi-automatic, or configured autopilot path appropriate to the account.
- Confirm broker truth: Treat broker acknowledgments, order state, and positions as authoritative after placement.
Workflow 6: Monitor, exit, and reconcile
Monitor protection state, live P&L, fills, rejections, partial fills, and broker connectivity. Multi-leg positions require leg-level visibility because partial execution changes the payoff and risk profile.
At exit, Nexus records the lifecycle and reconciles the cockpit with broker positions and orders. Review realized P&L and execution quality after the position is closed; do not infer the final outcome from a transient screen value.
Workflow 7: Use autopilot as a bounded loop
Autopilot repeats the scan-to-gate-to-deploy process on the server. Per-user in-flight protection prevents overlapping cycles, and stop actions invalidate future rescheduling of a current loop.
Before enabling it, confirm the active strategies, allocation, cadence, broker health, risk limits, and kill-switch behavior. Autopilot increases operational speed; it does not remove market risk or the need for supervision.
Frequently asked questions
Should I deploy immediately after a good backtest?
No. Check validation scope, costs, current regime, liquidity, account risk, and live broker conditions before deployment.
What is the difference between a signal and a strategy?
A signal describes evidence or direction; a strategy defines the tradable structure, sizing, risk, execution, and exit behavior used to express that view.
Can autopilot skip Nexus risk checks?
No. Autopilot operates through the same account and deployment boundaries and can be blocked by risk, broker, data-quality, or execution gates.