Shogun_Kim

Correction on two numbers above (they got swapped): – Takeaway 2: "pumps +145%" should read +20%. That is the filter for the 88-event subset. – Takeaway 4: the VTHO bounce was +145% from the low, not +20%. The chart marks it correctly. Also, a -17.1% drawdown means liquidation from ~6x, not 5x. All stats unchanged.

Shogun_Kim

@Shogun_Kim

4 days ago

"Extreme negative funding triggers a short squeeze." We’ve all heard this rule of thumb. I ran the numbers across Binance Futures over the last 6 months. The data shows the exact opposite. Analysis: All instances of funding ≤ -1%/8h across 528 Binance USDT perpetuals (175 total events) (Consecutive prints on the same asset clustered as 1 event) • Total (175 events) – Median 48h return: -11.2% – Positive after 48h: 26% • After a +20% pump in preceding 24h (88 events) – Median 48h return: -14.1% – Positive after 48h: 24% • Without prior pump (87 events) – Median 48h return: -6.8% – Positive after 48h: 29% • Extreme funding ≤ -2%/8h (45 events) – Median 48h return: -12.3% – Positive after 48h: 36% Three out of four events saw prices trade lower 48 hours later. The idea that "crowded shorts must bounce" fails across almost every subset. Key Takeaways: The side paying the fee was winningA -1% rate doesn’t mean shorts are trapped. It usually means the market is collapsing so fast that paying a hefty premium to hold short exposure is still profitable. Negative funding after a pump is the deadliestWhen an asset pumps +145% and funding flips deeply negative, shorts were right three times out of four. Median 48h return: -14.1%. Liquidated before the bouncePrices can bounce—median max upside within 48h was +8.4%. The problem? Median max drawdown before that was -17.1% (at 5x leverage or higher, that's outright liquidation). Most long squeeze-chasers get wiped out long before any relief rally hits. Why we misremember: Survivorship bias Take VTHO in September: The first extreme negative funding hit right after a +55% surge. Price subsequently fell -27.7% in 24h and -17.0% in 48h. The +20% bounce I remembered on Upbit only happened after the floor had completely collapsed. We remember the violent green candle and forget the bleed that preceded it. Context & Caveats: – Not an edge to blindly short: extreme outliers like RAVE and ARIA appear in both the top 5 best and top 5 worst performers. – Sample limited to 6 months; trading/funding fees excluded; delisted coins omitted (meaning survivorship bias likely makes real-world performance even worse). #Crypto #TradingEdge #TradingHelper

Shogun_Kim

@Shogun_Kim

2 days ago

"Crowded longs get flushed." The mirror of last week's rule of thumb. Last time, extreme negative funding did not trigger squeezes (3 out of 4 events traded lower 48h later). So I ran the exact same test on the other side. Analysis: All instances of funding ≥ +0.1%/8h across 525 Binance USDT perpetuals, last 6 months (406 total events) (Consecutive prints on the same asset clustered as 1 event) • Total (406 events) – Median 48h return: -1.5% – Positive after 48h: 44% • After a +20% pump in preceding 24h (96 events) – Median 48h return: -8.1% – Positive after 48h: 33% • Without prior pump (310 events) – Median 48h return: -0.7% – Positive after 48h: 47% Crowded longs on their own? Basically a coin flip. Key Takeaways: 1. The funding sign barely mattered. The pump did. Median 48h return / % closed higher, side by side with Part 1: • After a +20% pump – Negative funding: -14.1% (24% positive) – Positive funding: -8.1% (33% positive) • No prior pump – Negative funding: -6.8% (29% positive) – Positive funding: -0.7% (47% positive) 2. After a pump, price tended to fade no matter which side was crowded Whether shorts were paying -1% or longs were paying +0.1%, the outcome leaned toward a fade. 3. Without a pump, crowded longs mostly went nowhere Across 310 events, the median was -0.7%. Funding alone gave almost no directional signal. Before you read funding, check the last 24h candle. Context & Caveats: – Thresholds differ (-1% vs +0.1%). Positive funding rarely gets as extreme as ±1%, so the threshold was set lower. Not an apples-to-apples comparison. – Events cluster in a few tickers (one asset alone accounted for 25 events). – Sample limited to 6 months; trading/funding fees excluded; delisted coins omitted. – Even after a pump, 1 in 3 positive-funding events closed higher. Not a signal to blindly short. #Crypto #TradingEdge #TradingHelper