playbook

Bitkub Trade Playbook

Long-only swing trading on Bitkub THB pairs via the gokub-mcp tools. The tools are the hands; this is the brain. Follow the flow, respect the guardrails, never skip the validation gate.

Hard guardrails (never violate)

• Risk ≤ 2% of equity per trade. Size from the stop, not from a fixed amount.
• Long only, uptrend only. No shorting, no counter-trend entries.
• No entry unless validate_trade_setup returns can_trade=true. This is the gate.
• Take-profit ≥ 2R. If R:R < 2, skip the trade.
• Bitkub has NO native stop/OCO. Protect positions with client_side_stop_worker (poll it on a loop until triggered=true). Never assume a resting stop exists.
• Never lower a trailing stop. Move it up only.
• No averaging down on a short-term plan.

Key mechanic: indicators need candles

The analysis tools (check_market_regime, calculate_atr, calculate_rsi, detect_breakout_signal, detect_pullback_signal, calculate_ema, calculate_roc) take arrays (prices or OHLCV candles), not a symbol. So always:

1. get_historical_candles for the symbol+timeframe first.
2. Feed close prices (use extract_close_prices) or the candle array into the tool.

Use the big timeframe (1D / 240) for regime + relative strength, the entry timeframe (60 / 15) for ATR + entry signals.

The flow

Step 0 — Pre-session portfolio review (always run first)

Before screening for new trades, check what you already hold.

1. get_trade_history(status_filter=open) — list all open positions.
2. For each open position:
   • get_historical_candles(symbol, interval=60) → check_exit_signals(candles).
   • If result has should_exit=true or ≥ 2 exit reasons → exit immediately: place_limit_order(side=sell) → log_trade_exit(trade_id, exit_price, exit_reason).
   • check_trade_pnl(symbol, entry, qty, stop) — note current R to inform next steps.
3. get_my_open_orders — collect all symbols that already have a pending limit order on the exchange. Skip the entry step for any symbol already on this list. This prevents duplicate orders when the skill runs on a recurring schedule (e.g. every 2H).
4. If cash < minimum reserve (see skill overlay) after exits → do not open new positions this session. Review only.

Only after all urgent exits are handled, proceed to step 1 for new entries.

---

1. Shortlist — get_market_screener (volume, spread, depth) or get_market_overview for context. Keep 5–10 liquid THB pairs.
2. Regime filter — get_historical_candles (1D) → check_market_regime(prices). Trade only if UPTREND. (Confirm with calculate_ema: close > EMA200, EMA50 > EMA200.)
3. Relative strength — calculate_relative_strength_rank(symbols[], benchmark=btc_thb). Keep the top names.
4. Volatility fit — calculate_atr(candles) → keep ATR% in a sane zone (~2–6%/day).
5. Entry signal (entry timeframe) — detect_breakout_signal (close above 20-bar high
   • volume ≥1.5×) or detect_pullback_signal (pullback to EMA20 + RSI bounce 40–50).
6. Size — get_wallet_balance → calculate_position_size(balance, risk_percent=2, entry, stop). Stop from signal's suggested_stop (swing low or 1.5×ATR).
7. Gate — validate_trade_setup with the metrics above. Stop here if can_trade=false.
8. Round — get_symbol_rules → round_to_exchange_rules(symbol, price, qty). Reject if valid=false (below min notional).
9. (Optional) sanity — simulate_trade to confirm R:R ≥ 2 before committing.
10. Place + log — place_limit_order(side=buy) then log_trade_entry → keep the trade_id.
11. Protect — arm client_side_stop_worker(symbol, side=sell, trigger=stop, limit_price, qty) and poll it (e.g. via /loop) until triggered.

Managing an open trade

• PnL check — check_trade_pnl(symbol, entry, qty, stop) (fetches price; nets fees).
• At +1R — move stop to break-even (re-arm client_side_stop_worker at entry).
• At +2R — take partial and/or switch to calculate_trailing_stop(candles, atr_multiplier=2); re-arm the worker at the new level.
• Emergency exit — check_exit_signals(candles): STRUCTURE_BREAK or 2+ reasons → exit now (place_limit_order(side=sell) or let the worker fire).
• On exit — log_trade_exit(trade_id, exit_price, exit_reason) (computes net PnL + R).

Review (weekly)

• get_trade_history(status_filter=closed) and calculate_expectancy. Expectancy = win% × avgWinR − (1−win%) × avgLossR. If negative, change ONE variable and re-test.

Notes

• All symbols use lowercase btc_thb; the library normalizes to THB_BTC for the API.
• place_limit_order amount: buy = THB to spend, sell = coin quantity.
• Fees are charged both legs — the PnL/sizing tools already net them; don't double-count.
• The trade journal is a flat file (trades.json, override with env TRADES_FILE).