Setup
Training vs Live Mode
Simulate trades risk-free or go live with real capital.
Every AgentNash account starts in Training mode — a safe, zero-risk environment where your AI agents run the full trading pipeline against real market data without ever placing a real order. This safety-first design applies across both supported exchanges, Kalshi and Polymarket, so you can evaluate strategies, learn the platform, and build confidence before committing real capital.
| Training | Live | |
|---|---|---|
| Real money at risk | No | Yes |
| Exchange orders | Simulated (paper) | Real orders placed |
| AI analysis pipeline | Full pipeline runs | Full pipeline runs |
| Rules engine | All safeguards enforced | All safeguards enforced |
| Trade status label | paper | executed |
| P&L tracking | Separate training P&L | Real P&L |
| Default for new accounts | Yes | Must be explicitly enabled |
| Supported exchanges | Kalshi + Polymarket | Kalshi + Polymarket |
Training Mode
Training mode is the default for every new account and every newly deployed agent. The complete pipeline runs end-to-end — markets are scanned, strategies analyze opportunities, the rules engine validates every decision — but no real order is ever sent to an exchange. Trades are recorded as paper trades, and P&L is calculated against live market prices from the target exchange (Kalshi prices in cents, Polymarket prices in USDC decimals).
Because the AI pipeline is identical in both modes, Training results are a reliable preview of how your agent would perform with real money. Use Training to compare strategies, tune risk parameters, and verify that exchange credentials work before flipping the switch.
Training mode is not a downgraded experience. Every safeguard, every analysis step, and every rule fires exactly as it would in Live mode — the only difference is that the final order is simulated instead of submitted.
Live Mode
Live mode places real orders on the target exchange using your connected account credentials. Enabling it is a deliberate, multi-step process:
- Go to Settings → Safeguards.
- Toggle the Live Trading master switch to enabled.
- Confirm the warning dialog acknowledging that real capital will be used.
- On the strategy detail page, switch the individual agent from Training to Live.
Live mode also requires valid exchange credentials for the exchange your strategy targets. Kalshi strategies need a Kalshi API key pair. Polymarket strategies need your proxy wallet private key and deposit wallet address. Configure both in Settings → Exchanges.
Live trades use real capital and are irreversible. Kalshi orders execute against their order book. Polymarket orders are cryptographically signed and submitted to the CLOB smart contracts. AgentNash cannot recall or undo an order once it reaches the exchange. Always test thoroughly in Training mode first.
Two Levels of Control
AgentNash gives you two independent safety layers so live trading can never be enabled by accident:
- Global toggle — A master switch located in Settings → Safeguards. When this is off, every agent on the account runs in Training regardless of its individual setting or which exchange it targets.
- Per-agent mode — Each agent can be set to "training" or "live" individually from the strategy detail page or the Agents page. This lets you run some agents in Training while others trade live.
Both levels must be set to Live before any real order is sent. The global toggle acts as a hard master switch — turn it off at any time to instantly revert every agent to Training. Training and Live environments are tracked independently: portfolio snapshots, trade history, and P&L are all segmented by environment. This segmentation also applies per-exchange, so your Kalshi training P&L and Polymarket training P&L never mix.
Environment Filter
The dashboard header includes a three-way filter pill — All, Training, and Actual — that controls which data you see across the entire platform. The same filter appears at the top of the strategy detail page.
- All — Shows combined data from both Training and Live environments.
- Training — Shows only paper trades, training P&L, and simulated positions.
- Actual — Shows only real trades, live P&L, and actual exchange positions.
Use this filter to cleanly separate your testing results from real performance. The filter is persistent within a session, so switching pages keeps your selected view. It is available on the Portfolio, Trades, Strategy, and Agents pages.
Guided Walkthrough
New users are guided through a 13-step interactive walkthrough that teaches the entire platform before any trading begins. The walkthrough is presented as a focused overlay that highlights one feature at a time and automatically navigates between pages.
- Welcome — Introduction to AgentNash and what the tour covers.
- Strategy page — Overview of AI trading strategies and how they work.
- Browse Strategies — How to explore strategy cards and pick one.
- Training vs Live — The per-agent mode toggle on the strategy settings panel.
- Configure Your Agent — Risk limits, position sizing, and market preferences.
- Deploy Your Agent — How to start an agent with one click.
- Trades page — Where every trade and its AI reasoning chain are logged.
- Trade History — Reading trade status, side, entry price, and P&L.
- Portfolio page — Your command center for tracking overall performance.
- Performance Chart — Cumulative P&L over time, filterable by mode.
- Open Positions — Active positions across all agents and exchanges.
- Settings page — Exchange connections, API keys, and safeguards.
- Completion — Confirmation that you are ready to start.
The walkthrough can be replayed at any time from the sidebar. For new accounts that have not yet been approved for live trading, the tour runs in a forced mode — it cannot be skipped — ensuring every user understands the platform before they trade.
During the walkthrough, the platform displays representative demo data so you can see how each page looks with active agents, open positions, and trade history. No real trades are placed during the tour.
Live trading carries real financial risk. Markets can move against your positions, and prediction markets may resolve differently than your AI agents forecast. Start in Training mode, review your agents' decisions, and only enable Live mode with capital you can afford to lose.