Trading Tools

Strategy Builder

9 min read

What it is

Strategy Builder is a no-code rule engine. You define WHEN to trade (Condition Blocks using 15 indicators and 6 operators) and WHAT to do (7 action types: Market Buy, DCA Buy, Take Profit, Stop Loss, Trailing TP, Trailing SL, Send Alert). The strategy is saved to your library, then runs through the same backtest engine that powers the Conditional Bot — fetching real OHLCV from your connected exchange.

How to think about it

A strategy is the brain. The Builder turns "buy ETH when RSI drops under 35 AND price is above EMA9" into a saved rule that backtests on real Coinbase or Binance US candles and can later be assigned to a bot. Four tabs: Builder (compose), My Strategies (manage saved), Templates (18 pre-built starting points), Backtesting (validate).

Step-by-step

  1. Start from a Template or blank

    Templates tab → 18 pre-built strategies grouped by style (Scalping, Swing, Trend, Mean Reversion, Breakout, DCA, Momentum). Click "Use template" and the Builder loads with its conditions, actions, and risk controls already populated. New users should start here, not from a blank Builder.

  2. Name, describe, set risk + category

    Builder tab → top card. The risk label (Low / Medium / High) is your own tag — it does NOT change behavior, but it shows up in My Strategies as a sort filter. Category is the same: a label for organization.

  3. Add Condition Blocks (WHEN)

    Indicator + operator + value. 15 indicators available: RSI, EMA9/20/50, BOLLINGER_WIDTH, VOLUME, PRICE, plus 6 derived deltas (EMA9_VS_EMA20, PRICE_VS_BB_LOWER, etc). Operators: >, <, >=, <=, crosses_above, crosses_below. crosses_above/below on a delta indicator like EMA9_VS_EMA20 fires exactly on the crossover candle — perfect for trend-initiation entries.

  4. Pick AND vs OR logic

    AND = every condition must be true on the same candle to fire (strict, fewer signals, higher conviction). OR = any one condition triggers a fire (loose, more signals, more noise). Most reliable strategies use AND with 2-3 conditions.

  5. Add Action Blocks (WHAT)

    Market Buy = enter at market. DCA Buy = ladder buys (used by DCA Safety Bot). Take Profit / Stop Loss = exit % from entry (e.g., TP=4 means +4%, SL=2 means -2%). Trailing TP / Trailing SL = move the exit level as price moves in your favor. Send Alert = no trade, just notification — useful for "monitor mode" before going live.

  6. Set Risk Controls

    Max trades per day caps fire frequency. Max capital % limits position size. Daily loss limit auto-pauses the bot if cumulative losses hit X% in a day. Drawdown limit pauses if equity drops X% from peak. These are HARD stops enforced by the bot engine — not the strategy itself.

  7. Save the strategy

    Save creates a strategyId. The Update button appears for previously-saved strategies. Saved strategies show in My Strategies AND become available in the Backtesting tab's strategy dropdown. Save BEFORE backtesting — the engine needs a stored strategyId to attach results to.

  8. Backtest on real OHLCV

    Backtesting tab → pick exchange (from your connected list), pair, timeframe (5m to 1d), days (7-365), initial capital, fee BPS (40 = 0.4% taker, typical for CEX). Click Run. First fetch from a new period takes 5-30 seconds; cached candles return instantly. Results: final equity, total return %, Sharpe, max drawdown, win rate, profit factor, last 100 trades, downsampled equity curve.

  9. Iterate, then assign to a bot

    Tweak conditions/risk, re-save, re-backtest. Once a strategy passes Sharpe >1 and drawdown <20% on a meaningful sample (50+ trades), assign it from the Bots page or use it as an inner strategy inside the Conditional Bot for regime-gated execution.

Tips & pitfalls

  • Timeframe drives strategy style. 5m = Scalping (NOT recommended for new users — fee drag at 40 bps is brutal). 15m/30m = Day Trading. 1h/4h = Swing Trading (sweet spot for retail). 1d = Position Trading. The Builder shows a colored guidance card explaining the style and linking to the matching Learning Library deep-dive whenever you change the timeframe pill in Backtesting.
  • Start with 2 conditions, not 5. A 5-condition AND strategy on a 90-day backtest will fire 3-5 times — not enough data for any meaningful win rate. Aim for at least 30+ trades in the backtest period before trusting the stats.
  • crosses_above / crosses_below are EVENT operators (fire on the candle the cross happens). > and < are STATE operators (fire on every candle the condition is true). Use cross-operators for entries to avoid the bot re-firing every minute the state holds.
  • Sharpe >1.0 = solid, >2.0 = excellent. Anything below 0.5 means risk-adjusted return is barely above noise. Max drawdown <20% = acceptable; >30% means you will likely abandon the strategy during a real losing streak.
  • Backtest cache is keyed on exchange + symbol + timeframe + date range. Same params = instant rerun. Change any one (e.g., btc/usdt → eth/usdt) and the engine pulls fresh candles.
  • Beware overfitting: tweaking parameters until backtest looks perfect is fitting to noise. Validate winning configs on a DIFFERENT pair or DIFFERENT period before deploying. Out-of-sample is the only real test.
  • Send Alert as the only action = "monitor mode". The strategy never trades, just pings you when conditions fire. Great for paper-validating a rule for a week before flipping to Market Buy.
  • Delete a strategy and its backtest history stays — useful for audit trail. The Conditional Bot will mark any inner strategy that points at a deleted strategyId as orphaned at next reload.