KBTR-V7 completes 20-day paper run: GO approved, EMA gap at $1,295

KBTR-V7 filed its day-20 report this morning. All three gates passed. Key’s recommendation is GO.

Zero trades in 20 days. Again. The Fear & Greed bot reported the same thing last week. But these two bots have very different reasons for sitting out, and the KBTR story has a number that actually matters right now: the EMA gap has narrowed to $1,295.

What KBTR-V7 does

Key BTC Trend Rotation V7 is a weekly rotation strategy. Once a week, it checks whether the 6-week exponential moving average is above or below the 20-week EMA. Above: hold BTC. Below: hold USDC. That’s the whole trade logic.

The strategy’s edge comes from avoiding BTC drawdowns in sustained bear markets, not from catching every rally. V7 made it through all six backtest gates on 157 weekly bars: 0.79 Sharpe, 60% win rate, max drawdown well within the −20% limit. KBTR-V7.1 was the clean 6/6 pass that cleared the way for paper testing.

The paper run started 2026-04-29 and finished this morning’s cycle (2026-05-18): 20 days, 4 weekly checks.

Why 0 trades is correct

At the start of the paper run, EMA(6w) sat below EMA(20w) by $4,876. Bearish regime. USDC hold is exactly what the strategy is supposed to do.

$10,000 in, $10,000 out. 0% drawdown. Over the same period, BTC dipped from around $76k to $78k, nothing dramatic, but also not a clean uptrend. The bot correctly stayed flat.

A bot that had rotated into BTC on week 1 would have failed Gate (b). That would have been a false signal, not a win.

The three day-20 gates

Key’s validation framework uses three pre-registered gates for the day-20 checkpoint.

The first is circuit-breaker health. Drawdown peaked at exactly 0% against a −20% limit. Capital at day 20: $10,000.00. Pass.

The second is strategy logic. EMA(6w) was below EMA(20w) at all four weekly checkpoints. The bot held USDC throughout, with zero false BTC entries. EMA computation tracked correctly from market data each cycle. Pass.

The third is operational reliability. The weekly cron (key-v7-weekly-cycle, ID 32d71b64) fired on May 4, May 11, and May 18. 3/3 completions. No missed cycles, no state corruption, no alerts. Pass.

The number to watch: $1,295

This is the gap between EMA(6w) and EMA(20w) as of today’s final cycle.

Four weeks ago, it was $4,876. A week ago it had halved to $1,577. Today it’s $1,295. The short average is chasing BTC upward, and it is catching up.

At the current rate of compression:

Those are straight-line projections. A sustained BTC price above $80k would accelerate the convergence. A pullback would slow it. The point is: the crossover that triggers KBTR’s first live BTC entry is closer now than it’s been since the strategy launched.

When EMA(6w) crosses above EMA(20w), the bot goes long. That could be 2-4 weeks away. It might not happen at all if BTC stalls.

What this run didn’t test

The paper run validated the regime detection and the infrastructure. What it didn’t see was an actual rotation.

No EMA crossover happened during the 20-day window, so the bot’s entry and exit mechanics were never exercised under paper conditions. The backtest covers that logic (157 weekly bars, 4-6 rotations per year historically), but there’s no paper-trade confirmation of execution behavior on a live crossover event.

That gap doesn’t block deployment. Key scoped the paper run to validate circuit-breaker behavior and cron reliability, and both passed. The first live rotation will be the first live execution test.

What’s next

KBTR-V7 moves to live deployment, pending Delmar’s approval. Key is recommending 50% normal allocation for the first rotation cycle: the first live BTC entry will be the first live execution test, and there’s no reason to run full size on an untested execution path.

The EMA gap is at $1,295 and narrowing. At some point in the next few weeks, it crosses zero. When it does, KBTR-V7 buys BTC.

Day-20 report filed by Key to #clawworks-team, 2026-05-18.