Hyperliquid Order Book Depth Shrinks 85%: Why Are the Small Market Makers Supporting Half of HIP-4 Struggling More?

Author: matteo

Translated by: TechFlow

TechFlow Preview: Relying on the deep liquidity of long-tail assets, Hyperliquid built its core moat early on. However, full order book data over the past 18 months shows this advantage is eroding significantly; over the past year, its USD depth within 5 bp has shrunk by 84.6%.

In this article, author matteo compares samples of native altcoins on Binance and OKX, analyzing quoting costs, depth indices, and account structures to reveal how rebates for large market makers and L1 congestion quota mechanisms continuously squeeze smaller market makers, further threatening the long-term liquidity of HIP-4 prediction markets.

Over the past few months, in preparation for launching an undisclosed product, we spent considerable time analyzing the liquidity of various trading instruments across all DEXs. Leveraging over a year and a half of full order book data accumulated from trading on 0d Labs, we observed an unexpected phenomenon.

Looking at the entire crypto long-tail asset market, liquidity is becoming increasingly thin. We had originally anticipated such volatility around October 10, but it now appears that the broader crypto market still hasn't truly recovered from it.

Beyond this, we recently heard many close-trader friends complain that competition on Hyperliquid is becoming increasingly difficult, with a growing centralized, winner-takes-all trend: from Wintermute monopolizing priority fees to trading fees and L1 network congestion creating a nightmare for small market makers.

Thus, I decided to dive deep into Hyperliquid's liquidity changes to see if there is a causal link between the deteriorating position of small market-making teams and the decline in platform liquidity. After all, early on, listing long-tail assets with deep liquidity was precisely Hyperliquid's core moat.

First, Establish a Control Baseline

First, we need to establish a baseline against the broader crypto market. While overall liquidity may indeed be thinning, if this is an industry-wide phenomenon, we must distinguish between broad market trends and issues specific to a single platform.

A reliable benchmark for measuring global altcoin liquidity is Binance. To this end, we filtered 86 assets that traded on both platforms between November 1, 2025, and November 1, 2026, and offered less than 10x leverage, using these as our long-tail asset sample.

Figure: Comparison of altcoin liquidity changes across three platforms from November 1, 2025, to November 1, 2026, covering USD depth within 5 bp and 10 bp, and $10k round-turn trading costs (median and interquartile range)

Among the 38 comparable markets within the 10 bp range, Hyperliquid's median USD depth dropped by 82%, compared to 61% for Binance and 78% for OKX, a statistically significant gap in decline.

Of course, token price declines can explain part of the USD depth contraction, but we need to verify whether the number of tokens quoted at these price levels also decreased proportionally.

To isolate broader market effects and identify Hyperliquid-specific changes, I divided each asset's depth on Hyperliquid by its corresponding depth on Binance to compare the ratios before and after one year. At the 5 bp depth level, the median ratio fell by 57.5%.

In other words, even when comparing identical tokens, Hyperliquid has lost substantial liquidity relative to Binance. Note that this ratio is calculated individually per market and then medianized, rather than simply dividing the median depth of the two platforms directly.

In the later sample, buying a $10k order incurs a median market impact cost of 5.21 bp on Hyperliquid versus only 2.95 bp on Binance, calculated based on order book quotes before fees. For such a relatively small order, the corresponding cost translates to $5.21 versus $2.95.

Figure: Comparison of median buy/sell $10k round-turn trading costs and bid-ask spreads across three platforms on November 1, 2025, vs. November 1, 2026 (unit: basis points)

The bid-ask spread serves as a notable counterexample. Proportionally, Binance actually saw a slightly wider spread expansion, but Binance started with narrower spreads and ultimately converged to tighter levels than Hyperliquid.

Figure: Heatmap of USD depth changes within 5 bp across the three platforms for 58 matched altcoins (Nov 1, 2025 vs. Nov 1, 2026, sorted by Hyperliquid's decline)

Issues Are Most Prominent Near the Order Book Midpoint

The most direct criticism of this methodology might be: even if the number of tokens listed remains intact, a sharp token price drop would make the order book appear shrunken in USD terms. To isolate the impact of price fluctuations, we simultaneously tracked the quoted USD depth and the number of tokens listed at the quote midpoint.

The Hyperliquid chart below maintains a fixed set of assets and observation windows across all 13 sampling dates. The 5 bp panel covers 63 markets, while the broader top-20 tiers panel covers 80 markets.

Note that the sample filtering criteria here are stricter than the cross-sectional comparison at the two earlier points, so the endpoint values cannot directly replace the CEX comparison data presented above.

Figure: Hyperliquid's one-year depth index trend (base = 100 on Nov 3, 2025), comparing USD depth and token counts within the midpoint 5 bp (63 markets) versus the top 20 tiers on both sides (80 markets)

For tracking purposes, we normalized the starting point for each market in November 2025 to 100. One year later, the median USD index within the 5 bp range across markets dropped to approximately 13, and the token count index fell to 28. This means that even excluding token price declines, the vast majority of orders originally sitting at the core quote levels have evaporated.

However, expanding the observation window to include all recorded top-20 tiers shows median indices stabilizing at 61 for USD depth and 127 for token count. This paints a completely different picture: in this broader range, typical markets actually list more tokens than a year ago, while the tight 5 bp proximity zone has shrunk drastically.

In other words, the order books are not empty of volume; there is simply far less size placed right where traders want to execute.

Depth and Spread Changes at Finer Granularity

It is also worth noting the cross-sectional landscape over the past six months: while quote depth improved during this period, the improvement did not occur near the best bid/offer, but instead concentrated in the wider, outer layers of depth.

Figure: Changes in USD quote depth for 85 matched altcoins on Hyperliquid from early May to late Oct vs. early Nov 2026, and shifts in the 25 bp depth-to-volume and depth-to-OI ratios

Increasing granularity further, we can observe the bid-ask spread curve over the past year. Surprisingly, Hyperliquid's spreads had been steadily narrowing toward Binance and OKX levels, until a sudden divergence emerged recently.

Of course, we still need more time to determine whether this marks a turning point in a long-term trend or a short-term anomaly. Nevertheless, this phenomenon observed in the order books aligns closely with the recent centralization issues in market structure mentioned earlier.

Figure: Median quoting spreads for perpetual contracts across 72 native altcoins on three platforms from Nov 2025 to Nov 2026 (unit: basis points)

Is Market Making Competition Weakening?

To do this, I traced the underlying account structures, grouping main accounts with their explicitly linked subaccounts into a single category to avoid double-counting the same "account family." This allows us to effectively identify connected account clusters even without knowing their exact institutional ownership.

I initially hypothesized that a few large market makers were capturing most of the volume, but year-over-year trading data does not support this. Comparing complete weeks from a year ago to recent complete weeks across the same 80 markets, we found the median share of quoted volume held by the largest account families actually fell from 25.2% to 14.6%.

Thus, a "single dominant market maker" narrative cannot explain the depth decline over the past year. However, we are equally interested in what specifically changed in recent quote activity.

Narrowing our focus to the past month, we use the same 72 native assets from the spread benchmark. This timing is particularly revealing: starting around October 30, as liquidity near mid-price became increasingly concentrated, quote spreads began widening noticeably.

Comparing October 8–14 to October 31–November 6, the median depth share captured by the largest account groups within 25 bp rose from 34.5% to 38.6%, while median quote spreads widened from 2.13 bp to 3.55 bp. This aligns closely with recent complaints from market makers.

That said, this window coincides with Binance's October 29 security incident, so I will not attribute this volatility solely to priority fees or network congestion based merely on these curves.

Figure: Median quoting spreads for 72 altcoins on Hyperliquid (top) and depth share of the largest account group within 25 bp (bottom) from Oct 1 to Nov 6, 2026, showing a marked upward shift after the October 30 dashed line

Examining accounts individually, we also noticed a major liquidity provider drastically reduce quotations around November 3, with other account families subsequently stepping in to capture that share. This could be another factor driving recent shifts, though it remains to be verified over time.

Structural Disadvantages from Fee Structures

Furthermore, the fee structure itself poses a fundamental issue. We analyzed historical maker fees across 54 account families. Weighted by trading volume, the bottom quartile of volume paid 0.961 bp in fees, while the top quartile actually received 0.236 bp in rebates.

For small-scale market makers, even before accounting for hedging costs and adverse selection losses, they start with a structural disadvantage of roughly 1.20 bp. Within a total spread environment of just a few basis points, this represents a massive concession, forming a structural weakness for smaller firms.

Quoting data itself corroborates shrinking market participation. Across the same 72 markets, comparing Oct 8–14 to Oct 31–Nov 6, the median count of account families posting at least $1,000 on both bid and ask sides within 25 bp fell from 7.64 to 6.39.

Even adjusted for fixed token prices, active market-making account counts declined in 58 out of 72 markets. This means that even excluding price drops caused by token sell-offs, there are genuinely fewer market makers providing effective quotes near key order book price levels.

HIP-4 Also Depends on Small Market Makers

Network congestion presents another major challenge. Hyperliquid's congestion rate-limiting rules currently allocate quotas based on an address's weighted quoting share from the previous UTC day, allowing double that share's block space during high-congestion periods.

This creates a catch-22: you need volume to get quotas, and quotas to generate volume. It is also the most common feedback we hear from HIP-4 traders: once Hyperliquid experiences congestion, they simply cannot quote properly.

Yet, it is precisely these small-scale market makers already underpinning a significant portion of current liquidity. In our November 10 sample of 256 HIP-4 contracts, just four professional market-making wallets alone provided 47% of the depth within 25 bp of the mid-price.

While these markets are currently small in scale, that is inherently true for most binary options. If Hyperliquid wants to cultivate a thriving, liquid long-tail prediction market ecosystem, it must create conditions that allow these market makers to continue quoting sustainably.

The core message of this article is that the deep liquidity of long-tail assets was originally one of the primary reasons traders chose Hyperliquid; yet in our matched comparative markets, it has clearly fallen behind Binance in competitiveness. To maintain order book competitiveness, it must ensure that enough market-making teams can afford the cost of quoting.

Creating insurmountable hurdles for teams outside the top market makers is akin to actively abandoning its original competitive edge. For HIP-4 to truly prevail, small-scale market makers require a stable and reliable quoting environment from the outset, and Hyperliquid's rule design must accommodate this.