How-To Guides

Six recipes for the jobs you will actually do with Warren, from an honest market read to publishing a strategy. Each is a plain-English conversation, not a configuration exercise.

Warren chat workflow

Everything below is driven by talking to Warren. You never call anything directly: you describe what you want, and Warren selects and chains the right capabilities behind a calibrated, honesty-gated answer.

1. Get an honest market read

Ask Warren: "What's BTC doing right now?"

You get back the regime, how strongly the timeframes agree, and the level that would prove the read wrong.

Ask for the read: "What's BTC doing right now?" Note the coverage verdict that comes with it.

Ask how durable it is: "How long is this likely to hold?" for the persistence forecast.

Ask for levels: "Where would I enter, and what's the invalidation?" Warren frames them as if/then scenarios.

If Warren abstains, ask "Why did you abstain?" The reasons are always concrete, thin, stale, or conflicting inputs, not a shrug.

An abstention is a correct outcome, not a failure. A platform that answers anyway is guessing with your money.

2. Find and validate a strategy

Ask Warren: "Find me a low-risk strategy for this market."

Then insist on the full discipline before any capital is involved.

Ask for a fit: "Find me a low-risk strategy for this market."

Stress-test it across conditions: "Backtest it across all three regimes and grade its robustness." Read the worst regime, not the average.

Reality-check it: "Now run it over the last year continuously." Only a result that survives both views deserves capital.

See Strategy Recommendations for discovery and Building and Backtesting for the honest testing discipline in full.

3. Deploy a paper bot with guardrails

Ask Warren: "Paper-trade it on BTC with a 5% profit target, 5% max loss, and stop after 20 trades."

State the deployment and its guards in one sentence, like the example above.

Review the confirmation card. It includes the pre-trade risk check. Approve it.

Monitor from the Live Strategies screen on the dashboard, or just ask: "How are my bots doing?"

Stop anytime with "Stop it" or "Stop all my bots." Resume later without losing history.

Paper is the default and needs no exchange keys. See Trading Modes for how paper and live differ, and Dashboard Overview for the per-bot health verdict.

4. Let Warren trade a position

Ask Warren: "Open a small long on SOL with a stop-loss and take-profit."

Describe the position: "Open a small long on SOL with a stop-loss and take-profit."

Warren proposes the exact position, size, leverage, stop, target, and liquidation distance, then waits for your approval.

Track it with "Show your open positions."

Scale out with "Close half of it." The book runs against the paper sandbox until you explicitly allocate live capital.

This is Warren's own discretionary book, kept entirely separate from your strategy bots. See Capital Allocation for how the two share one capital pool without ever double-spending.

5. Watch the market while you're away

Ask Warren: "Check the market every morning and message me what changed."

Create a schedule in plain English, like the example above.

On each run, Warren scans, compares against the last run, and sends a Telegram or in-app summary.

Read the summary. That is the whole job by default.

Scheduled runs are watch-and-report by default: Warren is hard-blocked from trading on an unattended run. Arming autonomous execution is a separate, deliberate step, nothing ever trades unsupervised by accident.

6. Publish or vet a marketplace strategy

Ask Warren: "Publish this strategy to the marketplace" or "Is this strategy actually any good?"

To publish, say "Publish this strategy to the marketplace." Its track record travels with it, tied to the exact code version.

Before cloning someone else's, ask "Is this strategy actually any good?"

Warren surfaces the worst drawdown and per-pair results behind the headline number, judge it on those, not the cover.

After cloning, re-validate it yourself: backtest across regimes, grade the worst one, then deploy paper first.

A "+180%" card with one catastrophic regime is a disqualification, not a buy. Marketplace numbers are computed by the platform from verified backtests, never typed in by an author.