Shorts Hold 80% of Seven-Day Top-Trader Positions With $48M Unrealized Losses

Shorts Carry the Seven-Day Book

Shorts carry the seven-day line. Among the top 100 accounts ranked by seven-day returns on Hyperliquid, Lighter and dYdX, roughly 80% of position value now sits on the short side, about $2.50 billion against $608 million in longs, and the cohort as a whole is carrying approximately $48 million in unrealized losses. Hyperliquid (HYPE) positions account for about $56 million of that shortfall and Bitcoin (BTC) for roughly $33 million, so parts of the same book are in profit and the aggregate loss is narrower than the heaviest single bets. The snapshot comes from a daily position tracker built on derivatives records, refreshed hourly, which as of Friday, October 16 has for the first time folded the three venues into one aggregate instead of counting Hyperliquid alone; day-over-day comparisons are due from the next update. Its method is to rank the hundred most profitable accounts on each venue by realized returns over each window and sum their open positions, so the aggregate shows where the biggest recent winners still have exposure. The other end of the same table points the opposite way. Traders ranked by 24-hour returns were about 82% long, holding roughly $1.12 billion in long positions against $248 million in shorts. The 30-day cohort lands between the two at about 55% long, $1.336 billion against $1.084 billion, and it carries an unrealized profit near $82 million. The divergence tracks price direction across the windows the table covers: over the 24 hours to the snapshot, market data had BTC and Ethereum (ETH) each about 1.1% higher, Zcash (ZEC) up 8.4%, and NEAR Protocol (NEAR) up 10.4% at $5.32. Our live feed now puts BTC at $82,369 and ETH at $2,443, both marginally higher over 24 hours.

The coin-level split follows the same period-by-period flip. BTC positions in the 24-hour group sit 116 accounts long against 24 short, $221 million against $65 million, a 77% long share. In the seven-day group the accounts still lean long, 99 to 53, but the money does not: $656 million in shorts against $166 million in longs, an 80% short share by value. Over 30 days BTC reverts to 63% long, $533 million versus $317 million across 108 long and 41 short accounts. ETH shows the widest 24-hour skew of the majors, 113 long accounts against 14, with $315 million in longs and only $34 million in shorts, a 90% long share. The seven-day ETH book is 80% short at about $750 million across 52 long and 66 short accounts and carries roughly $17 million in unrealized losses, while the 30-day group is modestly short at 55%, $319 million against $258 million. NEAR Protocol (NEAR) is the most one-sided long in the sample: 55 accounts long against 5, $76 million against $0.9 million, a 99% long share, with about $29 million in unrealized profit, the largest of any asset tracked. Its seven-day group, at 54% long, is the only one of the ten largest positions where longs outweigh shorts, and its 30-day group is 74% long across 45 accounts with about $31 million in profit. ZEC splits cleanly by window: 97% long, about $142 million, in the 24-hour group, then 72% short and 64% short in the seven-day and 30-day groups respectively.

A Selection Effect, Not a Market View

None of this reads as a directional forecast, and a leaderboard of this construction selects for whoever called each window's direction correctly: on a day when the majors rose, long accounts climb the 24-hour ranking while the seven-day ranking retains the side that matched the week. A handful of large positions can also swing the totals, spot holdings and other venues sit outside the sample, and hedge status cannot be observed from position data alone. That separates this read from a sentiment gauge such as the Crypto Fear and Greed Index, which measures mood rather than realized performance. What it leaves standing is the imbalance itself: on the seven-day line, longs remain the lighter book at $608 million against $2.50 billion in shorts, and that loss-carrying posture stands until the weekly window rolls over.