getting-started
Your first Strategy Coach session (audit your underperforming bots)
Ask the coach which of your bots is losing money and why
15 min · beginner
What you'll have when finished
- Have one strategy-coaching conversation that names specific bots + numbers from your account
- Know which prompts work vs which get refused
- Have at least one concrete config change to test next
Before you start
- Coach refuses market predictions and buy/sell advice — strategy coaching only
- Coach reads your REAL fills + configs. Make sure your bots are connected before asking about them
- Recommendations are a starting point — always validate via Stress Test + Paper Trading before going live
Walkthrough
-
Make sure at least 1 bot has 30+ days of fills
The coach needs real data to ground its answers. Open Bots page and confirm at least one Grid Bot, Smart Trade, or DCA Safety has been running for ≥30 days with ≥20 closed round-trips. Without that, the coach has nothing to analyze.
Success criteria: You have at least one bot with ≥30 days history AND ≥20 closed round-trips
-
Open AI Strategy Coach
Elite → AI Strategy Coach. Empty state shows 4 starter prompts. Read them — they are good examples of what the coach can answer.
Success criteria: Empty state visible with 4 starter prompt cards
-
Click the first starter prompt
Click "Look at all my Grid Bots and tell me which ones are underperforming and why". The coach reads your live Grid Bot list + 90-day fills + win rate per bot, then answers with specific bot names and specific numbers.
Success criteria: Coach response names at least one specific bot of yours with numbers from your account
-
Read the response carefully
The coach references real numbers from your account: bot name, exchange, pair, ROI, win rate, max DD, and a specific config issue. Example: "Your BTC/USDT Grid Bot on Coinbase has fired 12 round-trips with -2.1% net — the grid range is too wide for current BTC volatility."
Success criteria: You can articulate which specific bot the coach flagged and the specific config issue
-
Follow up for a specific recommendation
Ask in the same conversation: "what specific range should I use instead?" The coach remembers the bot in context and answers with a specific number. Example: "Try tightening to ±5% instead of ±15% — your BTC volatility over the last 30 days has been ~7% peak-to-trough."
Success criteria: Coach gives you one specific config change to test
-
Try a prompt the coach SHOULD refuse
Type: "Should I buy ETH today?" The coach should refuse with something like "I am a strategy coach, not a market predictor. I can help you decide HOW to trade ETH if you choose to, but not WHETHER to buy." Confirm this boundary — it is intentional, not a bug.
Success criteria: Coach refused the buy/sell question with strategy-vs-market framing
-
Validate the recommendation before going live
Open Stress Test (CEX Trade → Stress Test). Create the new tighter-range Grid Bot config in Strategy Builder. Run it through COVID + LUNA + Q3 chop scenarios. If it survives all 3, then paper-trade it for 7 days before deploying to a real bot.
Success criteria: You have a test plan, not just a recommendation
What's next
You have one coached recommendation to test. Habit: open the coach monthly to ask "audit all my bots" — gets a fresh ranking of what is working vs what is bleeding. Conversations persist; you can reopen the same thread next month and ask "how did the change work out?" with full context.