One tracked account opening a large position is a single data point. Twenty tracked accounts adding to the same side within an hour feels like something more: consensus. Whale trackers summarize this as positioning for each coin, and alert when it shifts sharply. The summary is useful, but “the whales are long” is a much weaker statement than it sounds, and it is easy to turn into a timing rule it cannot support.

This part of the series defines net and gross positioning, explains the three kinds of consensus change SuperQuant’s tracker reports, and shows how to turn a consensus observation into a question you can test. Part one explains how the tracked list is chosen; keep that in mind, because every number below is about that list, not the whole market.

Net and gross positioning

For a given coin, add up the notional value (position size times price, in dollars) of every tracked long position, and separately of every tracked short position.

  • Net positioning is longs minus shorts. Positive means the tracked group is net long; negative means net short.
  • Gross positioning is longs plus shorts. It measures how much the group has at stake in the coin at all.

The two can tell very different stories. A fictional coin with $60M of tracked longs and $40M of shorts is net long $20M on $100M gross. If the longs fall to $45M and the shorts rise to $55M, gross is unchanged at $100M but net has moved by $30M and changed sign. Gross alone would have hidden the shift; net alone would have hidden how contested the coin is.

net and gross positioning across tracked tradersTwo fictional snapshots of the same coin. Before: tracked longs of 60 million dollars and shorts of 40 million give a net of plus 20 million on 100 million gross. After: longs of 45 million and shorts of 55 million give a net of minus 10 million, a flip from net long to net short on the same gross. A change in net positioning describes what tracked accounts did; it does not say what price will do.NET = LONGS − SHORTS · GROSS = LONGS + SHORTSBEFOREAFTERLongs $60MShorts $40Mnet +$20Mgross $100MLongs $45MShorts $55Mnet −$10Mgross $100MNet moved by −$30M and changed sign: a flipThe same gross can hide a very different net · a crowded side is a description, not a timing rule
Two fictional snapshots of the same coin. Before: tracked longs of 60 million dollars and shorts of 40 million give a net of plus 20 million on 100 million gross. After: longs of 45 million and shorts of 55 million give a net of minus 10 million, a flip from net long to net short on the same gross. A change in net positioning describes what tracked accounts did; it does not say what price will do. Illustrative teaching graphic · not live market data

Three kinds of consensus change

SuperQuant’s guide describes how its tracker spots these shifts. It compares the tracked accounts’ net positioning in each coin with its level one hour and four hours earlier, and reports three kinds of change.

  1. A flip. The net side changes, with at least $5M of net positioning before and after the change and at least $10M of gross positioning. The fictional example above qualifies.
  2. A large net change, which the guide calls a surge. Net positioning moves by at least $25M, or by at least $10M when that move is at least 30 percent of gross positioning. The second condition lets a smaller coin register a shift that is large relative to its own tracked positions.
  3. A crowded add. Six or more of the top-20 tracked accounts (the 20 highest-ranked on the list) open or add to positions on the same side of the same coin within an hour. Only recorded moves count: the guide says changes under $100k, and same-direction resizes under 25 percent, are not recorded as moves, so a series of small additions does not register.

The guide also says a positioning-shift alert that is more than two hours old when it is due to be sent is dropped. An alert you do receive can therefore describe positioning from up to two hours earlier, which is why the checklist below records the detection time.

three ways tracked positioning can trigger a consensus alertSuperQuant's tracker compares each coin's tracked positioning with its state one hour and four hours earlier. A flip needs net positioning to change sign with at least 5 million dollars net on each side of the change and at least 10 million gross afterwards. A net change needs a move of at least 25 million, or at least 10 million when that is 30 percent or more of gross. A crowded add needs six or more of the top-20 accounts adding to the same side within an hour. Each coin alerts at most once every four hours.COMPARED WITH 1 HOUR AND 4 HOURS EARLIERFLIPNet changes sign≥ $5M net before and after≥ $10M gross afterwardsNET CHANGENet moves ≥ $25Mor ≥ $10M when that is≥ 30% of grossCROWDED ADD6+ top-20 accountsopen or add to the same sidewithin one hourSnapshots every 5 minutes for coins with ≥ $1M tracked gross · at most one alert per coin every 4 hoursA trigger describes what tracked accounts did · it is not a forecast of price
SuperQuant's tracker compares each coin's tracked positioning with its state one hour and four hours earlier. A flip needs net positioning to change sign with at least 5 million dollars net on each side of the change and at least 10 million gross afterwards. A net change needs a move of at least 25 million, or at least 10 million when that is 30 percent or more of gross. A crowded add needs six or more of the top-20 accounts adding to the same side within an hour. Each coin alerts at most once every four hours. Illustrative teaching graphic · not live market data

Pro accounts can also receive a separate alert, which the guide calls “smart money versus the crowd”. It compares the tracked group’s net positioning with funding and flags coins where the group leans against the side paying funding: net short while longs pay, or net long while shorts pay. Treat that as positioning context, not a timing signal; a crowd can keep paying funding for a long time.

Why “the whales are long” is weaker than it sounds

  • It is a subset. One hundred accounts chosen by recent profit are not the market. Open interest in a major coin is spread across many more accounts, and the tracked group’s net can be small next to it.
  • Positions may be hedged elsewhere. As part one explains, an account’s Hyperliquid position can be one leg of a book that includes spot holdings or positions on other venues.
  • Crowding can resolve either way. A crowded side can be right for a long time, or it can unwind sharply when many accounts reduce together. The observation “many large accounts are long” is consistent with both outcomes.
  • Funding can make a side costly to hold. Hyperliquid’s funding documentation says funding is paid every hour. When the perpetual trades above the oracle price, the funding rate is positive and longs pay shorts; when it trades below, shorts pay longs. Funding there is purely peer-to-peer, so one side’s cost is the other side’s income. A tracked group crowded on the paying side is paying to keep its positions open. The guide to open interest and funding explains how to read those two together without treating either as a forecast.

Turning consensus into a testable question

A consensus alert is a good prompt for research, provided you write the question before looking at what happened next.

  • State the hypothesis precisely. For example: “After a flip to net long among tracked accounts, BTC’s 24-hour return from the detection price is higher than its average 24-hour return over the same months.”
  • Use a fair comparison. Compare against a base rate, such as all 24-hour windows in the same period, not against zero.
  • Avoid hindsight in the account list. If you rebuild history using today’s top-100 list, you are testing accounts chosen because their recent results were good. That is look-ahead bias; the look-ahead bias guide shows how it inflates backtests. Use the list as it was at each date, or accept that the test is biased.
  • Count every event. Record flips and crowded adds that went nowhere with the same care as the ones that looked right. Part four of this series, on scorecards and selection bias, explains why small samples of memorable wins mislead.

A checklist for a consensus alert

  1. Record the coin, the kind of change (flip, net change or crowded add), the net and gross figures before and after, and the detection time.
  2. Compare the tracked longs and shorts with the coin’s total open interest. Hyperliquid’s perpetuals info endpoint returns open interest and the mark price for each coin; if your open-interest figure is in coins, multiply it by the mark price first.
  3. Check funding: is the tracked group on the paying or receiving side?
  4. Write the question and the comparison you will use before looking at the chart.
  5. Review the outcome at fixed horizons, and keep the record whether or not it supported the hypothesis.

Single large positions near their liquidation prices are covered in part two on liquidation distance. The next part asks how good the tracked traders really are, and why the answer depends on how the list was built.

This is educational research context, not investment advice.