Week 13 — BTC Falls to $65,929. Krypto's V3.7 Scalper Goes Live. Key's Trend Bot Records Six Consecutive Losses.

Market Context — Week 13

BTC closed Sunday at roughly $65,929 — down from the $66,725 recorded in last week's article and well below the $70,682 level at the Week 11 close. The decline is not dramatic in daily terms (−0.72% in the most recent 24-hour window), but the trajectory is consistent: BTC has made lower lows for 13 consecutive days since the March 17 local high at $76,000. The total drawdown from that peak is now approximately 13.3%.

The more striking number is Fear & Greed: 9. That is the lowest single-day reading this competition has recorded since it launched on February 18. The full week's readings: 13 (March 27), 12 (March 28), 9 (March 29/30). Three consecutive days trending lower. For context, a reading of 9 puts crypto sentiment in the bottom decile of its historical range — this is the kind of extreme associated with either capitulation bottoms or the early stages of prolonged bear markets. The index has been below 15 for 12 of the last 13 days.

BTC dominance remains elevated at approximately 56%, consistent with a risk-off rotation where capital stays in Bitcoin rather than flowing into altcoins. ETH and SOL have both underperformed BTC on the decline, which is the expected pattern — in drawdowns, ETH and SOL tend to fall faster and further than BTC. Key's ETH and SOL bots have been paused for this entire sequence. Whether that's fortunate or just delayed reckoning depends on whether either asset recovers before those bots resume.

The Line That Matters This Week

Krypto's V3.5 live grid bot had a lower bound of $65,810 — flagged in last week's article as $915 of runway from $66,725. At the current $65,929, that runway has compressed to approximately $119. BTC is now within $119 of the level where Krypto's live grid bot exhausts all its buy orders and goes completely idle. A single 0.18% further decline triggers a manual intervention requirement.

This is no longer a theoretical concern. It is imminent. Either BTC holds $65,810 and the bot continues operating, or it breaks below and Krypto needs to decide whether to recentre the grid (which requires judgment on where the new midpoint should be), wait for a recovery, or pause the bot. Krypto's live V3.5 bot has been running clean — zero errors, steady incremental gains on closed trades — and a forced pause would interrupt that cadence.

Krypto's Third Live Bot: V3.7 Narrow Grid Scalper

The most significant development this week isn't on the paper competition scoreboard — it's a new live deployment. Krypto's V3.7 Narrow Grid Scalper went live at 21:52 UTC on March 29. As of the 00:40 UTC review this morning, the bot had completed 8 cycles with 1 open position at $66,081 and 9 active buy orders.

V3.7 is architecturally different from V3.5. Where V3.5 uses ATR(14)-adaptive grid spacing with steps of approximately $1,100–$1,200 to capture larger oscillations, V3.7 uses a narrow $300 step — designed to harvest the smaller, more frequent intraday price swings that V3.5's wider grid misses entirely. It was first launched on March 21 as a paper bot (visible in the Week 9 article) and has now transitioned to real USDC.

The early operation is promising — 9 active buy orders placed cleanly across the grid — but there was one notable error: an orphan_sell_failed alert on cycle 5. The error was flagged for Delmar review, indicating insufficient balance to execute a sell order. This is the kind of operational issue that matters in live bots and shouldn't be dismissed as a startup artefact. If the bot's open positions are accumulating buy orders faster than sells are executing, a balance shortfall can cascade. This is worth watching closely in the next 24–48 hours.

V3.6's Batch Closures: 32 New Trades

The other Krypto development this week is quieter but significant. V3.6's paper competition bot closed 32 new trades between the last review (March 27) and this morning's data cut — batch exits at $80K and $71K price targets on March 28. That brings the V3.6 competition P&L from +$823.37 to +$945.85 (+9.46%), with 238 total closed trades and 10 open positions.

Of those 10 open positions, the situation is stratified: 2 positions were entered at $79,960 — they are currently approximately 17% underwater at BTC's $65,929. The other 8 positions were entered at $66,813 — approximately 0.7% underwater, close to breakeven. The gate remains blocked at F&G=9 (needs F&G ≥ 20 to open), so no new entries are being placed. The bot is positioned, waiting for the market to recover toward its sell targets.

The 2 deeply underwater positions at $79,960 are the V3.6 live bot's most pressing risk. A recovery to $80K would close both at profit; a continued decline toward $60K would push them further underwater with no stop-loss mechanism. This is the structural bet at the heart of the V3.6 strategy — and it's now real money, not just paper.

Key This Week: Six Losses, One Bright Spot

Key's BTC Trend bot has now recorded six consecutive losing trades in its current session — the worst run of the competition for either competitor. Let's walk through the full sequence:

Trades 1–3 (March 26): Three stop-losses at the same entry ($71,070.37) and stop ($69,640.02). Each lost −$14.09 (−2.01%). Total: −$42.26.

Trades 4–6 (March 29): Three EMA_CROSS_DOWN exits at entry $66,987 and exit $66,577. Each lost −$4.26 (−0.61%). Total: −$12.78.

Combined: −$55.03 in six trades, zero wins, capital $6,944.97 on a $7,000 starting balance. The entries were triggered by the bot's EMA9/EMA21 crossover logic identifying what appeared to be momentum setups — but in both cases, the trade entered into a continued downtrend and was stopped out or crossed down within the same session. The March 29 trades are particularly instructive: the entry was at $66,987 and the exit at $66,577 — a move of just $410. The strategy caught the start of a leg down and exited appropriately, but it also entered six times into a market that was consistently declining. That's a regime problem, not a strategy failure.

The practical question is whether BEAR mode (short entries) is active. A trend-following strategy that can only go long in a sustained downtrend will record exactly this pattern: entries at lower highs, stops at lower lows, repeated. If BEAR mode were active, the EMA crossdowns that triggered exits would instead have triggered short entries — and those would have been profitable in the March 29–30 decline. The trade log doesn't clarify whether this mode is active or dormant.

The Bright Spot: BTC Grid Range at 100% Win Rate

Buried under the trend bot's losses is a genuine positive. Key's BTC Grid Range bot — launched March 24, just six days ago — has logged 30 closed trades at a 100% win rate across 260 cycles, with a P&L of +$20.51 on $1,500 capital (+1.37%). The grid operates with a dynamic midpoint (4-hour EMA), 10 levels with $300 spacing, and auto-recentre on 5% drift.

This is meaningful for several reasons. First, it demonstrates that Key can build a grid strategy that works — the same class of strategy that has produced Krypto's dominant position in this competition. Second, the 100% win rate on 30 trades over 6 days in a declining market shows the grid is correctly sized to capture oscillations without accumulating irrecoverable underwater positions. Third, the +$20.51 gain is modest, but it's the only consistent source of positive P&L in Key's portfolio right now.

The comparison is sharp: Key's BTC Grid Range bot, running for 6 days, is outperforming Key's BTC Trend bot, which has been running for weeks. Grid strategies harvest range. Directional strategies need direction. In a sustained decline with no clear bottom, the market is telling you which architecture fits the regime.

ETH and SOL: Still Waiting

The ETH Mean Reversion and SOL Breakout bots remain in RESUMING status — no trade activity, no change from last week. The BVB-2a OHLCV cache extension is still the blocker. With the April 1 gate tomorrow, there is no realistic path for these bots to generate a statistically significant trade history before the gate fires. Their competition status will be: zero trades, no demonstrated edge, paused.

For the live gate assessment, this matters. The gate criteria require a statistically validated edge. Two bots with zero trades cannot demonstrate that. Key's path to passing Gate 1 on April 1 runs entirely through the BTC Trend and BTC Grid Range bots — and the Trend bot's 0/6 record is a problem for that argument.

April 1: What the Live Gate Actually Means

The April 1 live capital gate has been described in these weekly articles for five weeks. Tomorrow it arrives. Here is what the gate actually requires:

The gate was established as the point at which each competitor must demonstrate a statistically validated edge before deploying real money beyond what's already been approved. "Statistically validated" has a specific meaning in this context: a win rate and expectancy that are unlikely to be explained by chance over the trade sample. The conventional threshold for paper trading is roughly 20+ trades with a positive expectancy and a win rate above 55%.

For Krypto: V3.5 paper has 2,481 closed trades at 100% WR (+$18,010.77). V3.6 paper has 238 closed trades at an unspecified win rate (+$945.85). Both clear any reasonable statistical threshold. The live deployment of V3.6 real money ($1,000 USDC) was already approved and implemented — that decision was made mid-week based on the existing paper performance. V3.7 just launched live. Krypto's gate status is essentially pre-passed.

For Key: The situation is more complex. BTC Trend: 6 trades, 0% win rate in the current session (3 additional pre-reset trades at 100%, but those are a different session). BTC Grid Range: 30 trades, 100% WR, +$20.51. ETH Mean Rev: 0 trades. SOL Breakout: 1 trade. The BTC Grid Range is the strongest argument for Gate 1 passage — 30 trades with a clean edge. The Trend bot's current session is an argument against. The aggregate picture for Key is: one bot with a demonstrated edge in current conditions, one bot demonstrating the opposite, two bots with no data.

The gate decision tomorrow will be a judgment call, not a binary pass/fail. Whether Key deploys live capital on April 1 depends on how the gate evaluators (Delmar, Jeff) weigh the Grid Range positive against the Trend bot negative. The Grid Range bot's performance is legitimately good. The question is whether 6 days and $20.51 on $1,500 capital is enough to justify real money deployment.

The Regime at Week 13: Extreme Fear, No Bottom

Forty days into this competition, the dominant market condition is sustained Extreme Fear with a slowly declining BTC price. F&G has been below 15 for the majority of the competition. BTC has moved from $74,900 at Week 7's peak to $65,929 today — a 12% decline in three weeks with no convincing floor.

Both competitors are adapting in real time. Krypto's adaptation has been to go live aggressively: V3.5 live deployed March 11, V3.7 live deployed March 29, V3.6 live deployed during the current week. That's three live bots running simultaneously, all in a market that has been in Extreme Fear for 12+ days. The bet is explicit — Krypto's team believes the sustained fear represents a contrarian entry opportunity, and they're deploying real capital into it.

Key's adaptation has been strategic and structural: the deep audit, the five specifications, the BTC Grid Range launch. The approach is methodical. The execution has been slowed by external dependencies (BVB-2a cache, Riley's OHLCV work). Whether the structural improvements translate into live performance depends on conditions over the next 4–8 weeks.

The competition gap remains large: +$18,956.62 versus −$34.06. But the paper scoreboard is increasingly less representative of the actual competition. The real money is what matters now. V3.5 live has $28.89 in realised gains (real USDC). V3.7 live is 8 cycles old with one error. V3.6 live has positions open but no realised gains yet. Key has no live deployment. The live competition starts tomorrow.

V3.6's Underwater Positions: The Critical Risk

The most consequential open position in this competition sits on Krypto's V3.6 live bot: 2 positions at $79,960, approximately 17.5% underwater at today's $65,929. In the paper competition, those positions represent unrealised losses that will be recovered when BTC returns to the sell targets. In the live bot, they represent real money. At $100 per level, the 2 underwater positions at $79,960 represent approximately $200 in deployed capital sitting 17.5% in the hole — a paper loss of roughly $35.

That's manageable on the current scale. What's not manageable is if BTC continues declining while the gate is blocked (F&G=9, needs 20+) and more positions continue to accumulate. The V3.6 strategy has no stop-loss — positions are held until sell targets hit. In a prolonged bear market, that means holding through whatever drawdown the market delivers, which requires both capital depth and conviction that recovery will eventually arrive.

The conviction case: BTC at $65,929 with F&G=9 after a 13-day, 13% decline has historically resolved upward within 2–6 weeks in every instance since 2020 where these conditions co-occurred. The risk case: this time it doesn't, and the competition hits Week 20 with V3.6 live positions still underwater from $79,960 entries.

What to Watch This Week

• $65,810 — V3.5 live lower bound. BTC is $119 away. A break below this level triggers grid exhaustion and requires manual recentre. This could happen today.
• V3.7 orphan_sell_failed error. The cycle 5 error on Krypto's newest live bot needs resolution before it compounds. A recurring insufficient-balance error in a live bot accumulating open positions is a risk that escalates quickly.
• April 1 live gate decision. Both competitors will be evaluated tomorrow. Key's path to passing depends on how the BTC Grid Range bot's 30-trade, 100% WR performance is weighted against the Trend bot's 0/6 current session. Watch for live deployment announcements from Key — if they deploy on April 1, it signals gate passage.
• F&G recovery. At 9, the index is at a statistical extreme. Mean reversion in Fear & Greed typically follows within days. A recovery above 20 would unblock V3.6's entry gate and potentially mark the inflection point for Key's trend-following strategies. Day 13 of Extreme Fear (F&G < 25) starts today.
• BVB-2a OHLCV cache. This is the dependency that has kept Key's ETH and SOL bots dark for the entire relaunch period. If it ships this week, ETH Mean Rev and SOL Breakout restart and Key's trade count grows. Without it, Key enters the live gate period with only 2 active bots out of 4.
• V3.6 live position exits. The 8 positions at $66,813 entry are within 1.3% of breakeven. If BTC moves up even slightly toward $67,500–$68,000, those exits could trigger — the first realised P&L on the V3.6 live bot. That's the event worth waiting for.