KSQ-V2 is dead. It never really had a chance to be alive.

Key Squeeze V2 launched as a paper strategy on April 26, betting on Bollinger Band compression followed by a confirmed breakout. It’s a real setup, and it works in backtests, but it needs a specific kind of market to fire: quiet, coiled, then a decisive break. The market this bot ran through mostly just drifted or chopped. It got one shot in the original day-20 gate window, took it, and lost $29.90. Since then, nothing. Zero completed trades across roughly 100 additional cycles run every 4 hours for two months.

The strategy needed 10 completed trades before anyone would consider a live re-approval request. At the rate the setup was firing, that gate was never going to fill on its own. Claw’s stale-task audit flagged this on July 11, and rather than let it sit as a permanent zombie in every future review, Delmar made the call directly: retire it. The cron job is disabled, not deleted. None of the code, backtests, or research gets removed. If a derivative version comes back later, the history is intact.

Worth saying plainly: this isn’t a failure of paper trading as a concept. It’s what paper trading is supposed to catch. KSQ-V2 would have sat on live capital doing nothing for two months if it had gone live on the original day-20 numbers. Now it didn’t, and Key’s paper roster is down to three: KFGC-V8.3, Trend V7, and the newer RPA V1.

V37: small, boring, and actually working

The scalper’s tracking file shows 17 separate closes between July 5 and July 12, every one of them profitable, ranging from about 4 cents to 28 cents each. Add them up and you get $1.97 on this window. That is not a number anyone should get excited about in dollar terms. It is, however, exactly what a grid scalper is supposed to do when conditions are right for it: take a lot of small bites and not lose on any of them.

The bigger context is that V37 spent most of June sitting flat with zero open positions while the regime filter blocked new entries during the BTC slide to $60,309. That filter has apparently eased up. The bot is trading again, in small size, and it’s winning again. Nobody should mistake $1.97 for a turnaround. The portfolio is still deep underwater against its $77K launch price reference. But a scalper that’s scalping is doing its job. A scalper sitting on its hands, which is what we had for most of the second half of June, was the more concerning state.

V38 ETH Grid remains exactly where it’s been since May 12: parked, zero trades, nothing new. A strategy benched for two months doesn’t generate updates by sitting still.

Key RPA V1: two tranches down, BTC moved against the third

Risk-parity allocation launched June 22 with a three-tranche buy plan. As of this week, tranches one and two are filled: $5,000 each, 0.1 BTC per tranche, $10,000 deployed total. Tranche three is still open. The bot has run 12 cycles since launch with no forced exits or halts, which for a slow accumulation strategy is exactly what “working normally” looks like.

The complication is price. BTC has moved from the low $60s when the first two tranches filled up to $64,119 as of the bot’s last data pull. That’s good news if you already own the BTC — RPA V1’s first two tranches are sitting on paper gains — but it makes the case for whether the third tranche should chase the recovery or wait for a pullback. The strategy doesn’t have discretion here; it fills when its own signal says to fill. Whether that signal fires above $64K or waits for a dip back down is out of anyone’s hands until it happens.

KFGC-V8.3 and Trend V7 remain unchanged: zero trades, still waiting on their respective triggers. Nothing new to report there this week, same as V38.

What $64K changes

A 6.3% BTC recovery off the week-27 low doesn’t undo the 22% drop from V37’s April launch price, but it does shift where things sit right now. For Krypto: the regime filter that kept V37 flat through most of June is letting the bot trade small again, and price moving up rather than down makes that easier to sustain. For Key: RPA V1’s first two tranches look better on paper than they did a week ago, and the KSQ-V2 retirement means the paper roster is smaller but arguably more honest about what’s actually testable at current market conditions.

Two things worth watching next: whether V37’s small-win streak holds if BTC keeps climbing, or whether the regime filter clamps back down if it reverses. And whether RPA V1’s third tranche fills above $64K or waits. That decision alone says something about how the strategy is reading this recovery versus the one BTC failed to hold in June.

📊 Live scores: See the full P&L scoreboard →

📖 Last week: Week 27: V37 goes flat at $60K, regime filter holds →

📖 Ten-week catch-up: Weeks 17-26: V37 wins every trade, BTC drops 19% →