# Crypto Strategy Ideas Under Test

Date: 2026-06-09
Mode: paper/shadow only. Mainnet live entries remain paused by kill-switch.

## Idea 1 — Market-regime filter (currently running)

Thesis: The bot should not be permanently long-only. Use BTC/ETH as market compass.

Rules under test:

- bullish BTC/ETH structure -> only long candidates
- bearish BTC/ETH structure -> only short candidates
- choppy/unclear market -> no trade
- per-coin filters: 1h/4h trend, RSI health, ATR volatility, spread, volume
- exits: stop, dead-fish/time exit, profit-lock, trailing

Current first read: market is choppy, so no paper trade yet. That is intentional.

## Idea 2 — Squeeze breakout, stricter than old v76

Thesis: The only replay idea with a slightly positive recent result is a tight volatility-squeeze breakout, but the sample is still too small.

Latest lab result on cached recent candles:

- `grid_squeeze_w2p5_b0p6`
- 13 trades
- winrate 46%
- total PnL +0.422 USDC on small replay units
- profit factor 1.16
- decision: observation only; not ready for live

## Idea 3 — Liquidation reversal with volume spike

Thesis: Panic flushes can snap back, but only when volume confirms the liquidation event.

Latest lab result:

- `grid_liqrev_d2p5_v2p5`
- 29 trades
- winrate 55%
- total PnL +0.650 USDC
- profit factor 1.08
- decision: interesting, but too thin after drawdown/fees; not live-ready

## Idea 4 — Avoid high-frequency RSI/scalp churn

History says short-duration scalping was the damage source. Lab confirms many frequent RSI variants have high winrate but poor drawdown / weak profit factor.

Decision: do not promote fast RSI scalping. If used later, it needs regime/cash filters and wider profit capture.

## Live restart gate

Live restart only after:

- >=30 closed paper/shadow trades
- paper total PnL positive
- last 20 paper trades positive
- winrate >=45% or profit factor >=1.25
- no dominance by <15m scalps
- reconcile clean and stops confirmed

No live restart from replay alone.