Skip to main content

Automation Strategy Definition Template

Note:
This document defines the structure and current state of my automated strategies

🤖 Automation Candidate​

Status: ready | prototype | backlog
Complexity: low / medium / high
API needs: place/cancel conditional orders, OCO or emulated OCO, fetch funding, fetch VWAP/BB

🧩 Origin Metadata​

FieldDescriptionExample
Idea SourceWhere the concept originatedManual trade / Backtest result / Observation / Forum / Research paper
Discovery ContextWhat prompted the ideaRepeated manual success / QLIR anomaly / News-driven pattern
Initial ValidationHow it first showed promisePositive expectancy in N manual trades / Backtest > 1.2 PF / Visual edge
StageConcept → Code Written -> Backtested → Paper Trade (various position scales) -> Live → Scaling -> DeprecatedProtoype

  • note that not all strategies need to go all the way through the pipeline, this is just the full length process, confidence for live deployment can happen at any stage.

  • Paper Trade DIY Platform

Signal Definition (inputs)​

  • Timeframes: [execution_tf, confirm_tf]
  • Indicators: BB(20,2), VWAP(session or 1h), Volume(20)
  • Features: bandwidth_percentile, vwap_distance_pct, wick/body ratio, volume_ratio

State Machine​

  1. IDLE → detect setup preconditions
  2. ARMED → stage entries (and mirror side if straddle)
  3. FILLED → cancel opposite; attach stop + TP ladder
  4. MANAGE → tighten stop on rules (time or structure)
  5. EXIT → TP hit / invalidation / timeout

Risk Controls​

  • Max attempts per session: N
  • Cooldown after fail: X minutes
  • Kill switch on BTC whipsaw / VWAP slope breach
  • Slippage guard: reject if est. slip > Y bps

Telemetry​

  • Log: entry buffer used, MFE/MAE, fill slippage bps, funding P&L
  • Alerts: on fill, on stop move, on timeout
  • Tags: strategy:<name>, mode:auto

Backtest Hooks (QLIR)​

  • Label: regime flip, first midline test, second test, range edge
  • Metrics: win-rate by bandwidth bucket, TP ladder hit-rate, attempt index performance
  • Outputs: optimal buffer, stop model (fixed vs ATR), best hold window

💼 Live Wallet​

[View on Drift →](https://app.drift.trade/overview/portfolio?authority=<some_address>)

Maybe later create a component or iframe or something so user doesnt have to leave my site??

Example Spec​

Automation Notes — Range-Bound Bounce​

  • Detect range: rolling HH/LL clustering with min 2 touches per edge; VWAP slope ≈ 0; BB bandwidth ≤ 40th pct.
  • Entry rules: limit at edge ± buffer; require rejection candle or close-back-inside; volume_ratio ≥ 1.2 at edges.
  • Stops/TPs: stop beyond edge (buffer bps); TP1=VWAP, TP2=opposite edge; optional reload on re-test.
  • Invalidation: VWAP slope |s| > threshold or close + follow-through outside band.
  • Cooldown: after 3 attempts on same edge, pause until fresh liquidity builds (N bars).