HIP-4 by the numbers: volume grew 50×, traders did not

We pulled every HIP-4 fill tied to a front-end builder code from Hyperliquid's public data, 2 May to 14 Nov 2026. Volume is booming. The number of people trading is a few hundred a week.

Updated · Slay Predict

Disclosure: Slay is a HIP-4 front end. Method and limits are at the end; corrections are welcome.

Key numbers

MeasureValue
Outcome-code weekly volume$1.2M (May) → $61M (week of 9 Nov)
Human-like weekly active wallets349 → 357 over the same period
Human-like share of volume (since Oct)1–3% of volume, ~0.5% of fills
Best week for real traders1,209 human-like actives, after the $1M USDC rewards program (29 Aug)
Rewards cohort still trading at week 4 / week 818% / 8%
May launch cohort still trading at week 8 (no rewards)18%

Volume grew about 50×. Traders did not.

Outcome-code volume went from $1.2M a week in May to $61M in the week of 9 November. Human-like weekly active wallets over the same period: 349 → 357. Since 10 October, one taker wallet trades against three maker wallets, up to ~72k fills a day.

HIP-4 weekly volume vs human-like weekly active wallets, May to November 2026

The biggest "new user" days were one cluster

The two biggest new-wallet days (15 Jun: +412; 3–4 Aug: +596) came from one cluster: ~650–1,000 wallets that each place one order a day in the 10 minutes before the 06:00 UTC BTC daily market settles, filled by the same 3–4 taker wallets. In August that cluster was 55–60% of weekly active wallets, so any "HIP-4 daily users" figure that counts wallets includes it.

New HIP-4 wallets per day, with the pre-settlement cluster highlighted

Rewards brought people; few stayed

After the $1M USDC program (29 Aug), human-like actives rose from 843 to 1,209. Of ~950 new wallets in those two weeks, 1 in 3 traded on one day only, 18% were still active at week 4 and 8% at week 8; 2 in 5 lost money. Since October most new wallets are dust: a median of 2 fills, about $4, on one day. The cohort that held up best was the first: May launch users, 18% still trading 8 weeks later, with no rewards.

Front ends are smaller than they look

  • Liquidiction / HIP4.ag: 196 wallets ever, 14–22 human-like a week; since mid-October one market-making wallet is 93–99% of its fills.
  • Totalis builder code: 12 wallets.
  • Share: builder address not found, so not measured.
  • The one rising human curve is the Liquid wallet: 45 → 159 weekly actives from 31 Aug to 9 Nov, 50–89 new wallets a week, with no HIP-4 rewards. Distribution inside an app people already use beat incentives.

What it means

The real HIP-4 audience today is a few hundred humans a week across all front ends. That is not a knock on HIP-4; it is early. But volume and wallet counts are the wrong scoreboard. The right one is humans still trading at week 4.

Method and limits

Data: public Hyperliquid fills carrying a front-end builder code, 2 May to 14 Nov 2026. Human-like = at most 50 fills a day and a median fill of at least $5, after removing wallets with 80% or more of their fills between 05:50 and 06:10 UTC. It is a heuristic. Fills without a builder code are invisible here, so wallet counts are a floor.

Found an error? Tell us on X at @slaymoneyapp and we will publish the correction on this page. First published as a thread on X. Related: what HIP-4 is and how it works.

Polymarket and HIP-4 in one app.
Sign in with email or Google. Your funds stay in your own wallet.
See live markets