Strategies Basics

First & Second Band

• Definition: A strategy uses two indicator series, first_band and second_band.
• Types: See categories: Moving Averages, Price‑Based, Volume‑Based, Trend/Other, Channels & Bands.
• JSON example:

{
  "first_band":  {"type": "ema", "factors": {"length": 21, "offset": 0}},
  "second_band": {"type": "sma", "factors": {"length": 55, "offset": 0}}
}

• Symmetry: Names do not force one to be faster. What matters is how they cross.

Cross & Direction

• Diff: Diff = second − first.
• Cross: The sign of Diff flips between two consecutive closed bars (e.g., + to −, or − to +).
• Direction: If first > second on the crossing bar → SHORT; else → LONG.
• Execution: Enter at the open of the next bar after the crossing bar (next‑open execution).
• No forming bars: Decide on closed candles only to avoid repainting.

Illustration

Blue is first_band, orange is second_band. The red dashed line marks the next‑bar open where execution takes place after the cross.

Axes and Timing

• Horizontal axis (Time →): Time flows left‑to‑right. One step along the x‑axis corresponds to one bar of the strategy’s Interval (e.g., 1m, 5m).
• Vertical axis (Price ↑): Higher positions on the y‑axis represent higher prices/indicator values.
• Closed vs forming bar: The cross is detected on a closed bar (the candle that has just finished). The red dashed line marks the next bar’s open, where orders are executed.
• Alignment: The bands are computed on the same bar timestamps as price. Use closed candles only to avoid repainting and to match the system’s evaluation.

Length (window/period)

• Role: Controls smoothing. Higher = smoother/fewer signals/more lag; lower = faster/more signals/less stable.
• Lag intuition: SMA lag ≈ (length−1)/2 bars. EMA/HMA/JMA reduce apparent lag at similar lengths.
• Guides by interval:
  • 1m: fast 7–21, baseline 34–55
  • 3m–5m: fast 10–21, baseline 55–89
  • 15m: fast 14–34, baseline 89–144
  • 1h: fast 21–55, baseline 100–200
• Channels: std (Bollinger) and scalar (Keltner) widen/narrow envelopes; combine with length to shape signal density.

Offset (shift)

• What it does: Shifts the indicator output forward by N bars (to the right). The current bar shows the value from N bars ago.
• Default: 0. Use 0 for parity and to avoid accidental look‑ahead.
• When useful: Rare niche cases (align a special anchored series, replicate a published method). Otherwise leave at 0.
• Caution: Non‑zero offset can give a false sense of responsiveness if misinterpreted. Always detect signals on closed bars.

Pairing Recipes

• Trend: EMA(21) vs SMA(55); HMA(34) vs JMA(89); Supertrend(10,3) vs EMA(21).
• Mean‑reversion: VWAP vs EMA(13); BBM(20,2) vs ZLMA(21); HL2 vs T3(8,0.7).
• Breakout: DCM(20) vs EMA(21); KCB(20,2) vs HMA(21).

Use higher multipliers/std/scalars and slower baselines to reduce whipsaw; lower to increase trade frequency.

From Idea to Playbook

1. Frame the hypothesis: Identify if you expect trend continuation, range rotation, or breakout. Choose bands accordingly.
2. Draft the JSON: Start with clean factor values (length, std, multiplier). Keep offset = 0 unless you have a documented reason.
3. Replay & annotate: Step through the last 30–50 signals on closed candles. Note false positives and restructure factors if needed.
4. Codify execution: Write when to enter, add, scale out, and invalidate. Store it alongside the strategy name so the desk can reference it fast.
5. Review cadence: Schedule a weekly review to see if volatility/volume changes require updating the parameters or retiring the play.