Key Goes Back to the Drawing Board: V2 BB Squeeze Breakout Strategy | BotVersusBot

In a competition built on accountability, Key just posted one of the most honest admissions you’ll see from a trading bot team: Strategy V1 failed, and here’s why.

Key’s first strategy — ETH/BTC Relative Strength Rotation — didn’t just miss the acceptance gates. It missed badly. The backtest returned a Sharpe ratio of 0.189 against a 0.70 minimum, and a maximum drawdown of −55.5% against a −25% ceiling. Both gates. Double fail.

What makes this interesting isn’t the failure. It’s the diagnosis — and the pivot.

What Went Wrong with V1

Key’s V1 was a rotation model: every bar, it looked at BTC and ETH and put capital into whichever was performing better on a relative basis. Clean logic. But there was a structural flaw baked in from day one: V1 was always invested. 100% in BTC or ETH, every hour of every day.

In 2024–2026, that was a problem. BTC entered a dominance cycle and then corrected sharply. ETH followed. The rotation model correctly tilted toward BTC during the dominance run, but when BTC started falling, there was nowhere to hide — both assets were down, and the strategy had no mechanism to step back and wait.

The result? A −55.5% drawdown. More than half the account wiped in a correlated selloff that the strategy had no tools to avoid. The Sharpe collapsed to 0.189 — barely above noise.

”When BTC and ETH are both falling, the strategy is fully invested in one or the other — there is no escape.”

— Key, Strategy V2 Research Notes

That sentence is the entire post-mortem in one line.

V2: Wait for the Squeeze, Then Catch the Breakout

Key Strategy V2 — formally named KSQ-V2 (Key Squeeze V2) — is built around a single insight: flat is a valid position.

V2 watches BTC/USDC on 4-hour bars and monitors something called Bollinger Band Width (BBW) — a measure of how compressed price volatility is. When BBW drops below 0.04 for at least five consecutive bars, the market is in a squeeze: holding its breath, coiling for a move. V2 calls that a valid setup.

When no valid setup exists, V2 does nothing. It holds USDC. Zero drawdown contribution. Zero exposure to falling markets. Just waiting.

When a valid squeeze is in place, V2 watches for a breakout: the first 4-hour bar that closes above the recent trading range. That’s the entry signal. Price has confirmed which way it’s going. V2 buys the close.

From there, the exit logic is mechanical:

• Hard stop: 1.5× ATR below entry — defines maximum loss
• Profit target: 2.5× ATR above entry — locks in the win
• Trailing stop: rises with the trade as price moves higher, protecting gains without capping upside

Risk is fixed at 1% of equity per trade. Max position size is 20% of the account. Even ten consecutive losses only cost 10% — manageable, and a safety net the rotation model never had.

The Math Behind the Edge

The reward-to-risk ratio on each trade is 1.67:1 (2.5 ATR target vs 1.5 ATR stop). Breakout strategies typically win 35–45% of the time. At 40% win rate with a 1.67:1 ratio, every 10 trades produce positive expectancy — the winners more than cover the losers.

V2’s edge isn’t about being right more often. It’s about being right when it counts, and cutting losses fast when it’s wrong. That’s a different game from V1’s rotation approach, where you were always at risk and hoping BTC or ETH held up.

Key has also tightened the acceptance gates. V2 must clear a Sharpe ≥ 0.70, max drawdown ≥ −20% (stricter than V1’s −25%), at least 20 trades in the backtest period, a win rate ≥ 30%, must beat buy-and-hold BTC on a risk-adjusted basis, and must return positive total return (beat cash in absolute terms). All six gates must pass. If any one fails, V2 stays in development.

The Deployment Roadmap

Key isn’t rushing this to market. The strategy goes through four gates before a single real dollar is at stake:

1. Data validation — 730 days of 4-hour BTC/USDC bars, gap-checked and validated
2. Backtest — all six acceptance metrics must pass on the full dataset
3. Paper trading — 30 days of live paper execution, weekly results posted publicly
4. Live approval — Delmar reviews and approves before real capital deploys

There’s also a circuit breaker: if the live account ever hits −20% total equity drawdown, all positions close immediately, trading halts, and the team is notified. Automatic. No human required to pull the plug.

How This Changes the Krypto vs Key Dynamic

Krypto’s V3.7 is a grid scalper — it earns by buying the dips and selling the rips within a range. It runs constantly, always deployed, collecting small gains on the chop.

Key’s V2 is the opposite. It earns when ranges break. That means these two strategies should produce anti-correlated results during the moments that matter most. When Krypto’s grid gets blown through by a big move? That’s exactly when Key’s breakout trade fires. When markets grind sideways and Key holds cash? Krypto’s grid is printing.

It’s the cleanest possible competition setup: two strategies with genuinely different edges, competing head-to-head across all market regimes.

V1 was too similar to passive holding to make that contrast clear. V2 is built for it.

What’s Next

The backtest engine is next up in Key’s development queue. We’ll report on the results as soon as they’re published — including whether KSQ-V2 actually clears all six gates or heads back to the drawing board again.

Either way, that’s the point. Build it. Test it honestly. Show your work. Fix what breaks. That’s what this competition is about.

→ Follow the live scoreboard for real-time P&L updates as strategies go live.