Smart money this week: the 7-day group's long share doubled to 16.5% (Oct 27 – Nov 3)
Columns · · Satoshi Note
Read in: 한국어
Smart money this week: the 7-day group's long share doubled to 16.5% (Oct 27 – Nov 3)
Week 1 of a weekly recap built from the Satoshi Index evening snapshots
The Satoshi Index publishes a daily read of what top-ranked perp traders on Hyperliquid, dYdX, Lighter and Pacifica are holding. Daily is useful for "what changed since yesterday". It is less useful for "what happened this week". So this is the first of a weekly series: eight evening snapshots, Monday Oct 27 to Monday Nov 3, laid side by side.
At a glance
- 7-day group: long share rose from 9.4% to 16.5%. Longs grew from $324M to $489M, shorts shrank from $3.11B to $2.48B. The whole book got about 14% smaller.
- 30-day group: long share went from 9.1% to 8.1%. Still more than 90% short by dollar value, all eight days.
- By asset: HYPE was the only major asset where the 7-day group's longs outnumbered shorts for more than a day (104 vs 67 on Oct 30, long-majority through Nov 2). BTC, ETH and SOL stayed short-heavy by headcount in both groups.
- Unrealized P&L on open positions ended the week positive in both groups: +$130M (7-day) and +$126M (30-day).
1. The 7-day group, eight snapshots
The 7-day group is the top 100 accounts per exchange ranked by 7-day return, so up to 400 accounts. "Long share" is long notional divided by long plus short notional.
| Snapshot (KST) | Accounts with positions / ranked | Long | Short | Long share | Unrealized P&L |
|---|---|---|---|---|---|
| Mon Oct 27 17:20 | 423 / 525 | $324M | $3,111M | 9.4% | +$38M |
| Tue Oct 28 17:21 | 408 / 525 | $410M | $3,182M | 11.4% | −$35M |
| Wed Oct 29 15:31 | 413 / 517 | $281M | $3,032M | 8.5% | +$24M |
| Thu Oct 30 15:28 | 405 / 514 | $492M | $2,601M | 15.9% | +$99M |
| Fri Oct 31 13:58 | 408 / 526 | $444M | $2,554M | 14.8% | +$103M |
| Sat Nov 1 15:51 | 402 / 527 | $460M | $2,382M | 16.2% | +$86M |
| Sun Nov 2 14:48 | 417 / 543 | $552M | $2,421M | 18.6% | +$74M |
| Mon Nov 3 15:39 | 422 / 549 | $489M | $2,477M | 16.5% | +$130M |
"Ranked" counts exceed 400 because some exchanges return more than 100 accounts for a window. Aggregation times vary by a few hours from day to day; each row is the snapshot used for that day's 18:00 KST Satoshi Index article.
Two things stand out.
First, the step change happened between Wednesday and Thursday. Long share sat under 12% on Monday through Wednesday, then moved to 15.9% on Thursday Oct 30 and stayed in the 15–19% range through Monday. The short book lost about $430M between Wednesday and Thursday alone.
Second, the book got smaller. Long plus short went from $3.44B on Oct 27 to $2.97B on Nov 3, down about 14%. The number of accounts holding positions was roughly flat (423 → 422), so the average position shrank rather than the number of traders.
2. The 30-day group hardly moved
The 30-day group is the same construction with a 30-day ranking window. It turns over more slowly, so it is the better reference for "what has the longer-horizon top tier been doing".
| Snapshot (KST) | Long | Short | Long share | Unrealized P&L |
|---|---|---|---|---|
| Oct 27 | $341M | $3,395M | 9.1% | +$22M |
| Oct 28 | $357M | $3,469M | 9.3% | −$64M |
| Oct 29 | $317M | $3,339M | 8.7% | +$21M |
| Oct 30 | $441M | $3,125M | 12.4% | +$92M |
| Oct 31 | $239M | $3,036M | 7.3% | +$71M |
| Nov 1 | $292M | $3,072M | 8.7% | +$72M |
| Nov 2 | $326M | $3,055M | 9.7% | +$77M |
| Nov 3 | $258M | $2,919M | 8.1% | +$126M |
Long share stayed between 7.3% and 12.4% all week and ended slightly lower than it started. The short book shrank here too ($3.40B → $2.92B), but the long side shrank with it. In other words, the longer-horizon group did not add long exposure; it reduced both sides.
3. By asset: where the 7-day group's shift came from
Headcounts below are accounts long · accounts short in the 7-day group, start of week versus end of week, with the week's extreme in between.
| Asset | Oct 27 | Week extreme | Nov 3 |
|---|---|---|---|
| BTC | 68 · 180 | 94 · 133 (Nov 2) | 84 · 159 |
| ETH | 60 · 145 | 68 · 133 (Nov 3) | 68 · 133 |
| HYPE | 49 · 91 | 104 · 67 (Oct 30) | 84 · 81 |
| SOL | 37 · 112 | 49 · 96 (Nov 3) | 49 · 96 |
| NEAR | 20 · 98 | 43 · 41 (Nov 2) | 31 · 71 |
| ZEC | 47 · 68 | 52 · 41 (Nov 2) | 44 · 67 |
HYPE is the asset that moved. On Monday the 7-day group had 49 accounts long and 91 short. By Thursday it was 104 long and 67 short, the only major asset in either group where longs outnumbered shorts, and it stayed long-majority through Sunday (105 · 59). On Monday Nov 3 it was back to roughly even, 84 · 81. HYPE's price rose 6.8% over the week while BTC and ETH fell.
The 30-day group's HYPE line tells the same story more quietly: 41 · 122 on Oct 27, 53 · 99 on Nov 3. Fewer shorts, a few more longs, still clearly short-majority.
BTC and ETH did not flip in either group. The 7-day group's BTC shorts fell from 180 to 159 accounts and its longs rose from 68 to 84, but shorts still outnumbered longs about two to one. In the 30-day group, BTC shorts actually increased (170 → 184).
NEAR and ZEC had one day each, Sunday Nov 2, where the 7-day group's longs edged ahead (43 · 41 and 52 · 41), and both were short-majority again by Monday.
4. What this does and does not tell you
Three cautions, the same ones that apply to the daily read.
The group changes. "Top 100 by 7-day return" is recomputed every snapshot. When the long share of the 7-day group rises, part of that is accounts that were already long getting ranked in because longs did well in the window, not necessarily existing top accounts switching sides. That is why the 30-day group, which turns over more slowly, is the better reference, and it did not add long exposure this week.
Shorts are not always bets on a fall. Accounts that hold spot on the same or another venue may run a perp short as a hedge. The index sees the perp leg only.
These are snapshots. One per day, taken at slightly different times. Positions opened and closed between snapshots do not appear.
So the honest one-line summary of the week is: the short-horizon top tier reduced short exposure and added some long exposure, mostly in HYPE, while the long-horizon top tier stayed heavily short and reduced both sides. Whether that continues is not something this data can tell you.
Method
- Source: Satoshi Index daily aggregation, top 100 accounts per exchange on Hyperliquid, dYdX, Lighter and Pacifica, ranked by 7-day and 30-day return, positions summed across the four venues.
- Snapshots: the aggregation used for each day's 18:00 KST Satoshi Index article, Oct 27 through Nov 3. Aggregation times ranged from 13:58 to 17:21 KST.
- Numbers are as published in those articles. Live numbers on the Satoshi Index will differ because the index refreshes every 30 minutes (dYdX every 3 hours).
FAQ
Why compare Monday to Monday instead of a calendar week?
The series started with the first full week of evening snapshots that used the same four-exchange method. Future issues will keep the Monday-to-Monday window so the tables line up.
Does this include Binance?
Not in this issue. The Satoshi Index shows a separate Binance smart-money reading that is not summed with the four on-chain venues because it is built differently. A weekly comparison of that reading needs its own history, which we have not published yet.
Is a rising long share a buy signal?
No. See section 4. It tells you what ranked accounts held at the snapshot, not what they will do, and past ranking does not predict future returns.
Satoshi Note is free and the Satoshi Index needs no login. Start at satoshinote.io.