Exchange Flows Report — Week 26, 2026

Exchange Flows Report — Week 26, 2026

Week 26 of 2026 is one of those weeks that looks ordinary on the surface until you pull the tape. Across 1,486 tracked events and $8.80 billion in total volume, the story that emerges is not one of balanced price discovery — it is one of lopsided sell pressure, institutional-scale positioning on a decentralized venue, and a derivatives market that continued to dwarf its spot counterpart by every measure that matters.

The headline number is uncomfortable for bulls: sell pressure reached $4,127.0 million for the week, versus $2,271.4 million in buy pressure. That is a 1.82:1 ratio of sellers to buyers by notional flow — and it is not a number that reverses overnight. When the aggregate market structure leans this heavily to one side, it typically means either (a) a large cohort of participants has already made up its mind about direction, or (b) a squeeze is loading on the other side. Given the event distribution this week, the former interpretation deserves more weight.

The exchange that dominated volume — and dominated it emphatically — was Hyperliquid. With $3,122.1 million moved across just 202 events, Hyperliquid's average event size clocked in at approximately $15.46 million per event. For context, Binance Futures, which saw 633 events and $2,020.4 million in volume, averaged roughly $3.19 million per event. The gap is not incremental — it reflects a fundamentally different class of participant. Hyperliquid in Week 26 was not where retail traded. It was where size moved.

Beneath those top-line numbers, the week's market structure story is about fragmentation and consolidation simultaneously: fragmentation in event count (spread across ten venues), and consolidation in volume (three venues — Hyperliquid, Binance Futures, and Bitget — accounted for $6,174.7 million, or 70.1% of total tracked volume). That kind of concentration is a signal in itself.

Exchange Leaderboard

The leaderboard splits cleanly into two views this week: rank by event count, which tells you where activity is concentrated, and rank by volume, which tells you where money is concentrated. The divergence between those two lists is the most structurally interesting feature of Week 26.

• VOLUME RANK 1 — Hyperliquid: $3,122.1M | 202 events | $15.46M avg per event
• VOLUME RANK 2 — Binance Futures: $2,020.4M | 633 events | $3.19M avg per event
• VOLUME RANK 3 — Bitget: $1,032.2M | 571 events | $1.81M avg per event
• VOLUME RANK 4 — OKX: $1,016.2M | 297 events | $3.42M avg per event
• VOLUME RANK 5 — Bitunix: $560.4M | 435 events | $1.29M avg per event
• VOLUME RANK 6 — Binance Spot: $484.4M | 343 events | $1.41M avg per event
• VOLUME RANK 7 — OKX Spot: $183.0M | 156 events | $1.17M avg per event
• VOLUME RANK 8 — Coinbase: $153.9M | 386 events | $0.40M avg per event
• VOLUME RANK 9 — Gate Futures: $127.4M | 397 events | $0.32M avg per event
• VOLUME RANK 10 — KuCoin: $103.6M | 392 events | $0.26M avg per event

The activity leaderboard inverts the picture dramatically. Binance Futures sits at #1 with 633 events — 3.1 times Hyperliquid's event count — yet trails it by over $1.1 billion in volume. Bitget follows at 571 events, Bitunix at 435. The bottom of the activity leaderboard belongs to OKX Spot at 156 events, despite the combined OKX family (OKX + OKX Spot) posting $1,199.2 million in total volume. That tells you OKX's tracked activity was dominated by fewer, larger trades — a pattern consistent with institutional aggregation on a single venue before execution.

The average event size gradient is striking and deserves emphasis: Hyperliquid at $15.46M, OKX at $3.42M, Binance Futures at $3.19M, Binance Spot at $1.41M, Bitget at $1.81M — and then a sharp drop to Coinbase at $0.40M, Gate at $0.32M, KuCoin at $0.26M. There is essentially a $15M club (one member), a $1.5-3.5M bracket (three venues), and a sub-$0.5M retail tier. Market structure is never one market — it is at least three distinct markets operating on the same instruments at the same time.

Notable in the rankings: Bitunix's appearance as a top-five volume venue — $560.4M on 435 events — signals continued growth for this platform in tracked activity. Its average event size of $1.29M places it solidly in the mid-tier bracket, suggesting a maturing user base moving beyond purely retail-scale positions.

Top 3 Exchange Deep Dives

Deep Dive #1 — Hyperliquid: $3,122.1M | 202 Events

Hyperliquid is the single most important venue to understand in Week 26, not because it had the most events, but because of what its volume-to-event ratio implies about the nature of the participants using it. At $15.46 million per tracked event, the activity profile on Hyperliquid is categorically different from any other exchange in this dataset. These are not retail swing traders sizing into $50,000 positions. These are accounts writing eight-figure tickets — and they chose a permissionless, on-chain perpetuals protocol to do it.

The implications for market structure are significant. Hyperliquid operates as a fully transparent order book — every position, every liquidation, every limit order is visible on-chain in real time. The fact that participants with $10M+ average event sizes are choosing this venue over private OTC desks or less transparent CEX dark pools suggests one of two things: either the cost-of-execution advantage on Hyperliquid is substantial enough to justify full transparency, or these participants are specifically using the on-chain venue to establish verifiable position records — a pattern seen in strategies involving proof-of-position for fund reporting or structured product collateral.

What's the directional read? With sell pressure dominating the week-wide aggregate at 1.82:1, and Hyperliquid representing 35.5% of all tracked volume, any sustained directional lean on Hyperliquid would move the aggregate materially. The fact that pump volume ($1,352.5M) edged out dump volume ($1,052.8M) in gross terms — even as total sell pressure ($4,127.0M) swamped buy pressure ($2,271.4M) — suggests that the larger individual events were net positive in price impact, while the aggregate flow of smaller events was relentlessly selling. Hyperliquid's large-ticket events likely account for a meaningful portion of those pump-side spikes.

Unique patterns: the ratio of volume to event count on Hyperliquid has likely been widening week-over-week as more institutional capital finds its way onto the protocol. The $15.46M average this week is a data point worth benchmarking against future weeks. If this number climbs further, it signals that large-account migration to on-chain perps is accelerating. If it drops, it may mean new retail entrants are diluting the average — a different, but also interesting, market structure signal.

Deep Dive #2 — Binance Futures: $2,020.4M | 633 Events

Binance Futures remains the engine of derivatives price discovery in centralized markets, and Week 26 reinforces that status. 633 events — the most of any venue in this dataset — combined with $2,020.4M in volume puts the average event at $3.19M. That is solidly mid-tier institutional: large enough to matter, frequent enough to suggest automated or algo-driven execution rather than single-ticket discretionary trades.

The event density on Binance Futures — 633 tracked events out of 1,486 total, representing 42.6% of all events — confirms what market participants have long understood: Binance Futures is where price is made before it travels to other venues. The bid-ask on most major perpetuals establishes itself here first, then gets arbitraged outward to OKX, Bitget, Gate, and smaller venues. Any serious analysis of cross-exchange flow has to anchor at Binance Futures.

The relationship between Binance Futures ($2,020.4M, 633 events) and Binance Spot ($484.4M, 343 events) is worth examining. The futures-to-spot volume ratio on the Binance family this week is approximately 4.17:1. This elevated ratio indicates that the dominant use of Binance's ecosystem in Week 26 was leveraged directional positioning rather than spot accumulation or distribution. The market was not quietly loading spot bags — it was making leveraged bets with high frequency. That is consistent with the broader sell-pressure imbalance: leveraged short positioning scales volume without necessarily moving the spot order book.

Notable: the gap between Binance Futures event count (633) and the next-highest venue — Bitget at 571 — is smaller than expected. Bitget is closing the activity gap. Over the past several months, Bitget has made aggressive moves in futures liquidity, and Week 26 shows that effort is translating into tracked event volume. Binance Futures is still dominant, but its lead in event count is not as commanding as its lead in raw volume would suggest.

Deep Dive #3 — Bitget: $1,032.2M | 571 Events

Bitget's $1,032.2M in Week 26 volume across 571 events cements its position as the third-largest venue by volume and second-largest by activity. The $1.81M average event size places Bitget in a distinct competitive position: above the pure retail tier (Coinbase at $0.40M, Gate at $0.32M, KuCoin at $0.26M) but below the mid-institutional bracket of Binance Futures and OKX. Bitget is where semi-professional traders — funded accounts, prop desks targeting Asian session volatility, and active copy-traders with meaningful AUM — execute.

Bitget's 571-event count relative to its $1.03B volume puts it closest to Binance Futures in operating pattern: high frequency, mid-sized tickets. This is copy-trading infrastructure at scale. Bitget's copy-trading product is one of the most heavily used in the industry, and high event counts with consistent mid-range sizes are exactly what you'd expect when a platform has thousands of followers auto-executing behind a few high-volume lead traders. The aggregate can look institutional in volume while being retail in origin.

The buy/sell dynamics on Bitget likely tracked the broader market lean toward selling given the week's aggregate data, though Bitget's copy-trading architecture can create micro-bursts of coordinated buys when popular traders enter positions simultaneously. These buy bursts are often short-duration and high-amplitude — showing up as pump events in the data — before copy-traders close and the aggregate reverts. This pattern is worth monitoring on weeks where Bitget's pump-to-dump volume ratio diverges meaningfully from the market average.

CEX vs DEX Analysis

The CEX versus DEX breakdown for Week 26 is one of the most structurally significant data points in this report. In a week where total tracked volume reached $8,803.6 million across 1,486 events, the split breaks down as follows:

• Hyperliquid (DEX): $3,122.1M | 202 events | 35.5% of total volume | 13.6% of total events
• All CEX combined: $5,681.5M | 1,284 events | 64.5% of total volume | 86.4% of total events
• DEX average event size: $15.46M
• CEX average event size: $4.43M
• DEX volume-to-event efficiency ratio vs CEX: 3.49x

A single DEX capturing 35.5% of total tracked volume is not a curiosity — it is a structural shift. Hyperliquid did not reach $3.1B in weekly volume by being a novelty. It reached that number because it solved the custody problem and the transparency-versus-speed tradeoff in a way that made institutional-scale participation viable on-chain. The 3.49x efficiency ratio (DEX events generating 3.49x the volume per event versus CEX) is the data expression of that fact.

Why is volume flowing to Hyperliquid specifically? Three structural reasons stand out. First, funding rates on Hyperliquid have historically been competitive — sometimes more favorable than Binance Futures — which makes it cheaper to hold large leveraged positions for extended periods. Second, the on-chain orderbook provides full transparency that some institutional counterparties now require for risk management and reporting. Third, for accounts that have experienced exchange risk — frozen withdrawals, KYC enforcement actions, platform insolvency — Hyperliquid's self-custody model eliminates counterparty risk entirely. In a post-FTX market, that is not a small selling point.

The institutional versus retail split across the full dataset is visible in the average event size distribution. Hyperliquid at $15.46M per event is institutional by any definition. OKX at $3.42M and Binance Futures at $3.19M capture the professional-retail to small-institutional bracket. Binance Spot at $1.41M, Bitget at $1.81M, and Bitunix at $1.29M represent the sophisticated retail to semi-professional tier. Coinbase at $0.40M, Gate at $0.32M, and KuCoin at $0.26M are predominantly retail — smaller position sizes, higher event frequency relative to volume.

The fact that CEX still commands 86.4% of events despite only 64.5% of volume reinforces the point: retail activity is dispersed across many small events on many centralized venues, while institutional activity concentrates into fewer, larger events — with a growing preference for Hyperliquid as the execution venue of choice.

Regional Flow Patterns

Mapping the ten exchanges in this dataset to regional operational bases produces three distinct clusters, each with its own volume profile and market structure characteristics.

Asian Exchanges

The Asian cluster — OKX, OKX Spot, Bitget, KuCoin, Gate Futures, and Bitunix — combined for $3,022.8 million across 1,851 events when their combined totals are aggregated. This represents 34.3% of total tracked volume. The cluster is dominated by Bitget ($1,032.2M) and the OKX family ($1,199.2M combined), with Bitunix ($560.4M) establishing itself as the third pillar of the Asian exchange ecosystem in this dataset.

Asian exchange activity is strongly futures-weighted. OKX's spot product ($183.0M) represents only 15.3% of the OKX family's tracked volume — the remaining 84.7% came from the OKX futures venue. Gate Futures similarly represents gate.io's tracked contribution entirely through derivatives. This reflects the structural reality of Asian crypto markets: leveraged futures trading dominates, spot accumulation happens but is harder to track at event-level granularity.

The time-zone pattern for Asian venues skews activity toward UTC+8 business hours — roughly 01:00 to 09:00 UTC. Events on KuCoin, Gate, and OKX tend to cluster around the open of the Asian session (midnight to 02:00 UTC), the lunch break (04:00-06:00 UTC), and the afternoon trading window (06:00-09:00 UTC). Bitget, with its substantial copy-trading base, shows a more distributed pattern since copy-trades execute whenever the lead trader is active, regardless of time zone.

Western Exchanges

The Western cluster — Coinbase and Hyperliquid — produced $3,276.0 million combined, but the composition of that number is extraordinary. Coinbase contributed just $153.9M on 386 events; Hyperliquid contributed $3,122.1M on 202 events. The Western cluster's volume story is almost entirely a Hyperliquid story this week.

Coinbase's $153.9M on 386 events at a $0.40M average event size confirms its role as a retail-to-institutional on-ramp rather than a high-volume trading venue. Coinbase events represent real spot buying and selling — not leveraged derivatives churn — which gives them outsized importance relative to their volume. When Coinbase events cluster in a particular direction, they signal genuine spot supply/demand shifts that are structurally different from futures positioning. The 386 Coinbase events in Week 26 carry informational weight disproportionate to their dollar volume.

Hyperliquid operates on a global 24/7 basis with no geographic restriction, but its participant base skews Western and professional. On-chain activity patterns suggest heavier event concentration during North American and European session hours (13:00-22:00 UTC), which is consistent with the fund and prop desk activity implied by its $15.46M average event size. Week 26's Hyperliquid volume, if confirmed to concentrate in Western hours, would suggest that institutional positioning on the bearish side of the market originated primarily from Western-timezone participants.

Global Exchange — Binance

The Binance family — Binance Futures and Binance Spot — combined for $2,504.8 million across 976 events, representing 28.5% of total tracked volume. Binance is genuinely global: its user base is distributed across every major time zone, which means its event stream does not cluster as sharply in any single session. This produces the highest sustained event frequency of any venue — 633 futures events alone — reflecting continuous price-making activity rather than session-specific bursts.

The Binance futures-to-spot ratio of 4.17:1 this week ($2,020.4M futures vs $484.4M spot) is a reflection of the market's leveraged character. During trending markets, this ratio tends to compress as spot accumulation catches up. During choppy or bearish periods, the ratio expands as participants prefer to express views through derivatives rather than commit spot capital. Week 26's ratio expansion — combined with the overall sell pressure imbalance — supports the bearish market structure read.

Arbitrage Routes Analysis

Cross-exchange arbitrage is most active on routes where (a) volume is high enough to absorb arb trades without moving price, and (b) price discovery is fragmented enough that meaningful spreads exist. Week 26's data highlights several active arbitrage corridors.

Primary Route: Hyperliquid ↔ Binance Futures

The most active arbitrage route in Week 26 was almost certainly the Hyperliquid-to-Binance Futures corridor. These are the two largest futures venues in the dataset, and their funding rate and mark price relationships create constant arb opportunities. When Hyperliquid's institutional participants build large positions, they create temporary funding rate imbalances that arb bots at Binance Futures can exploit by taking the opposing side and earning the rate differential. With $3,122.1M and $2,020.4M in respective volumes, the combined $5,142.5M flow between these two venues represents the primary price-setting mechanism for the week.

The typical arb execution on this route involves delta-neutral positioning: long perp on the venue with negative funding (to collect), short perp on the venue with positive funding (to pay less), with the spread profit being the rate differential minus execution costs. Given the sell pressure dominance in Week 26, the route likely ran with Hyperliquid funding leaning negative (rewarding longs) while Binance Futures funding stayed neutral-to-positive, creating a directional arb that favored going long Hyperliquid / short Binance Futures.

Secondary Route: OKX ↔ Binance Futures

The OKX-to-Binance Futures corridor is the most established CEX-to-CEX arb route in crypto. OKX's $1,016.2M in Week 26 volume versus Binance Futures' $2,020.4M creates a natural liquidity gradient: Binance Futures leads, OKX follows within milliseconds for major pairs. Statistical arb strategies on this route focus on the lag between Binance Futures price updates and OKX order book adjustments — typically sub-100ms but tradeable for automated systems.

OKX's $3.42M average event size is notably higher than Binance Futures' $3.19M, suggesting that the OKX events in this dataset are larger-ticket than average — possibly the result of arb bots sizing positions to be economically meaningful relative to the price differential being captured.

Tertiary Route: Bitget ↔ Gate Futures (Copy-Trade Arb)

A less obvious but structurally significant arb route in Week 26 involves Bitget ($1.81M avg, 571 events) and Gate Futures ($0.32M avg, 397 events). The average event size differential — 5.6x — suggests that copy-trading events on Bitget create momentum bursts that smaller Gate participants attempt to front-run or fade. This is not traditional statistical arbitrage; it is more accurately described as momentum-harvesting on a predictable liquidity event. When a high-follower Bitget lead trader opens a large position, the copy-execution creates a burst of identically-timed orders that temporarily moves price. Gate futures participants who identify the pattern can trade against the short-term overextension.

Spot-Futures Basis Route: Binance Spot ↔ Binance Futures

The $484.4M Binance Spot versus $2,020.4M Binance Futures volume creates ongoing cash-and-carry opportunities when the futures contract trades at a premium or discount to spot. In a bearish week with elevated sell pressure, the basis tends to compress or invert (backwardation), which discourages carry trades and may attract reverse-carry strategies. The 4.17:1 futures-to-spot ratio suggests the futures market was leading price discovery — consistent with directional short-sellers expressing views through leverage rather than spot.

Market Share Shifts

Without explicit prior-week volume data in this report, market share shifts must be inferred from the structural patterns visible in Week 26's numbers. Several signals point toward ongoing directional changes in the competitive landscape.

Hyperliquid's dominant position — 35.5% of total volume from just 13.6% of events — is not a one-week anomaly. Platforms with this kind of volume-to-event efficiency ratio are experiencing structural inflows from a specific category of participant: sophisticated traders who previously split their activity across multiple CEX venues and are now consolidating on a single, transparent, self-custody platform. If Hyperliquid's share is growing — and the weight of evidence in the crypto market structure narrative strongly suggests it is — then the venue most likely ceding share is not Binance Futures (which remains the price-discovery anchor) but rather OKX and the mid-tier Asian futures venues.

Bitget's close approach to Binance Futures in event count (571 vs 633) is a market share signal. Event count is a leading indicator for volume share because it reflects the number of distinct trading decisions made on a platform. More events mean more participants making decisions on Bitget, even if those decisions are smaller in size. If Bitget's event count continues to converge with Binance Futures while its average event size grows, it will challenge for the #2 CEX position within the futures landscape.

Bitunix's $560.4M on 435 events represents one of the more significant emerging market share stories in the dataset. A $1.29M average event size is not a retail-only platform — it's a platform where serious traders have started allocating meaningful capital. Watch this number. If Bitunix's volume-per-event continues to grow while its event count remains high, it is transitioning from a retail venue to a professional trading destination — a trajectory that Bitget completed over the previous two years.

On the losing side of market share, the data points toward KuCoin and Gate Futures. KuCoin's $103.6M on 392 events at $0.26M average — and Gate's $127.4M on 397 events at $0.32M average — represent the lowest volume-efficiency in the dataset. High event counts at low average sizes indicate active but small-scale retail participation. As retail participants mature and move to more feature-rich or lower-fee venues, these platforms face structural pressure. Neither $0.26M nor $0.32M average event sizes are consistent with platforms capturing the next wave of professional trading flow.

The Coinbase anomaly is worth flagging separately. $153.9M on 386 events at $0.40M average is low volume for a regulated U.S. exchange with Coinbase's brand and institutional access. However, Coinbase's role in this ecosystem is qualitatively different from a pure trading venue: it is a fiat on-ramp, an institutional custody solution, and a regulated spot market. Its tracked event volume understates its actual importance to the market structure because the largest Coinbase activity — OTC block trades, prime brokerage flows, ETF basket creation/redemption — occurs off the event-driven flow captured in this report.

Next Week Watch

The structural setup heading into Week 27 is defined by three things: an unresolved sell pressure imbalance (1.82:1), an emerging institutional preference for Hyperliquid as the primary large-ticket venue, and a mid-tier exchange layer (Bitget, Bitunix, OKX) that is consolidating activity at the expense of the retail-tier bottom.

• WATCH: Hyperliquid average event size. If it moves above $18M next week, institutional flow acceleration is confirmed. If it drops below $12M, new retail entrants are diluting the average — a different signal worth tracking separately.
• WATCH: Binance Futures event count vs Bitget event count. The gap narrowed to 62 events this week (633 vs 571). If Bitget closes to within 30 events, that is a meaningful market share headline.
• WATCH: The sell-pressure ratio. Week 26 closed at 1.82:1 sell-to-buy. Any move toward 1.50:1 or below would signal buying interest absorbing the supply. A move above 2.0:1 would suggest accelerating distribution.
• WATCH: OKX average event size versus Binance Futures. OKX averaged $3.42M vs Binance Futures' $3.19M — OKX's tracked activity was slightly larger on a per-event basis. If this premium expands, it may indicate institutional arb desks are routing more flow through OKX as a primary venue rather than secondary.
• WATCH: Bitunix's event count trajectory. If it breaks above 500 events, it becomes the #3 venue by activity — a position currently held by Gate Futures at 397 and KuCoin at 392 (both well behind Bitunix's 435).
• WATCH: Total tracked volume. $8.80B across 1,486 events is the Week 26 baseline. A compression below $7B would suggest broad risk-off and position reduction. An expansion above $10B would indicate leverage is being added, not removed — particularly meaningful given the current sell-pressure context.
• WATCH: The pump/dump ratio. Week 26 saw $1,352.5M in pump volume versus $1,052.8M in dump volume — a net pump-positive split despite the sell-pressure imbalance. This divergence (large individual events are bullish; aggregate continuous flow is bearish) is unstable and will resolve in one direction. Which side capitulates first is the key market structure question for Week 27.

Potential structural catalysts for Week 27: any macro print affecting risk sentiment will propagate through Coinbase first (as the regulated spot venue most sensitive to traditional finance flows), then Binance Spot, then into futures across the board. The sequence matters — a Coinbase spot move that is not confirmed by Binance Spot within minutes is likely a local liquidity event, not a true directional signal. If both confirm, watch for Hyperliquid to see the largest absolute volume response given its institutional participant base.

One scenario worth monitoring specifically: a short squeeze on the Hyperliquid-Binance Futures corridor. With sell pressure dominating and large institutional participants active on Hyperliquid, if funding rates turn sufficiently negative, long carry trades become economically attractive to the same institutions currently sitting short. A rapid funding-rate-driven unwind could produce a high-velocity squeeze that shows up in Hyperliquid's event data as a sudden spike in both event count and average event size. That combination — more events AND larger average size — would be the clearest possible signal that the squeeze is institutional in origin and meaningful in scale.

Sign Off

Week 26 told a clear market structure story: sell pressure at 1.82:1, institutional money consolidating on Hyperliquid, and the mid-tier exchange landscape quietly repricing its relevance as volume efficiency becomes the differentiator that matters. The $8.80 billion moved across 1,486 events was not random noise — it was a structured rotation with identifiable fingerprints on every venue.

The market is not broken. It is repricing. The question heading into Week 27 is whether the buyers who drove $1,352.5M in pump volume can scale their conviction to match the $4,127.0M in sell pressure that came at them this week — or whether that asymmetry holds, deepens, and becomes the defining feature of this quarter's market structure. Hyperliquid will tell you first. Binance Futures will confirm it. Everything else is commentary.

Exchange Flows — Week 26.