Where the Volume
Actually Goes
ATS aggregate rankings tell you which venues are largest. They do not tell you where a specific name's volume actually trades. The difference is 93x.
Using FINRA Rule 4552 issue-level data across 16.7 million records and 35 venues, Sapinover mapped per-ticker ATS routing patterns for 30+ securities spanning index ETFs, sector ETFs, precious metals, leveraged products, crypto ETFs, single names, and fixed income. The results reveal structural venue specialization that repeats week over week and carries real implications for execution quality and liquidity access.
01 / The Concentration Gap
DealerWeb ATS finished Q1 2026 with 0.44% of total NMS ATS share volume, ranked #22 among 35 reporting venues. By aggregate market share it is a minor participant, barely registering behind the names at the top of the league table.
Across Q1 2026, DealerWeb (MPID: DLTA) averaged 41.1% of all ATS-traded volume in SPDR S&P 500 ETF Trust (SPY) across 32 competing venues. That is a 93x concentration premium over its headline share. Across the same window it averaged 36.3% of QQQ and 24.3% of IWM. The pattern holds across every week in the quarter. It is structural.
This is the concentration gap: the distance between a venue's aggregate share and its share in the securities it actually specializes in. Every major ATS has one. Most market participants have never seen it quantified. Understanding it changes how you should think about liquidity access for specific securities.
The core finding
02 / Index ETF Anomaly
SPY, QQQ, and IWM are the three most liquid equity ETFs in the world by combined ATS volume. They represent three distinct indices (S&P 500, Nasdaq-100, Russell 2000) and attract broadly different institutional and retail constituencies. But in the ATS routing data, they share one thing: the same venue at the top.
~420M/wk shares · ~3.1M/wk trades
32 ATSs
+27 more ATSs
DLTA = DealerWeb ATS. KCGM = Virtu MatchIt ATS. Peer shares are representative Q1 snapshot.
~310M/wk shares · ~2.4M/wk trades
31 ATSs
+26 more ATSs
IATS = IBKR ATS. Retail brokerage conduit appears more prominently in QQQ than in SPY. Peer shares are representative Q1 snapshot.
| Symbol | Index | DLTA Share | DLTA Rank | INCR Share | DLTA vs. Market Avg |
|---|---|---|---|---|---|
| SPY | S&P 500 | 41.1% | #1 | 16.8% | 93x |
| QQQ | Nasdaq-100 | 36.3% | #1 | 18.3% | 83x |
| IWM | Russell 2000 | 24.3% | #1 | 20.5% | 55x |
Source: FINRA Rule 4552, Sapinover analysis. Q1 2026 average across 12 weeks. NMS Tier 1. Market average = DLTA Q1 overall share (0.44%). Concentration ratio = ticker share / market average.
IWM presents an interesting counterpoint. The Russell 2000's smaller-cap, higher-churn constituent base makes it more accessible to systematic and retail-facing venues. DLTA's lead narrows to 24.3% and Intelligent Cross closes the gap to ~20%, the tightest differential across the three indices. The concentration effect is real but attenuates as liquidity profile shifts away from the mega-cap core.
Why DLTA dominates the index ETF complex
03 / Sector ETF Map
Drilling below the index level to SPDR sector ETFs produces a different structural finding. DealerWeb's dominance retreats; Intelligent Cross (INCR) takes over. Across all 11 SPDR Select Sector ETFs -- from Technology (XLK) to Utilities (XLU) -- INCR ranked #1 in the observation window. No exceptions.
| Symbol | Sector | Tier | #1 ATS | INCR Share | #2 ATS |
|---|---|---|---|---|---|
| XLK | Technology | T1 | INCR | ~34% | UBSA |
| XLF | Financials | T1 | INCR | ~31% | UBSA |
| XLV | Health Care | T1 | INCR | ~30% | UBSA |
| XLY | Consumer Disc. | T1 | INCR | ~28% | EBXL |
| XLC | Communication | T1 | INCR | ~27% | UBSA |
| XLI | Industrials | T1 | INCR | ~26% | UBSA |
| XLE | Energy | T1 | INCR | ~25% | BLUE |
| XLB | Materials | T1 | INCR | ~24% | UBSA |
| XLP | Consumer Staples | T1 | INCR | ~24% | UBSA |
| XLU | Utilities | T1 | INCR | ~23% | LATS |
| XLRE | Real Estate | T1 | INCR | ~21% | UBSA |
Source: FINRA Rule 4552, Sapinover analysis. Observation period: weeks ending March 9 through March 30, 2026. NMS Tier 1. INCR shares are approximate based on observed range of 21-34%. LATS = Barclays LX ATS.
INCR's consistency across sectors is notable not just for its breadth but for what it displaces: DealerWeb, dominant at the index level, falls out of the sector ETF top five entirely. The institutional block flow that routes through DLTA in SPY does not disaggregate into sector vehicles the same way -- sector ETFs attract more systematic and algorithm-driven flows, which is precisely the order type that INCR's matching logic is optimized to handle.
The Energy sector (XLE) is the only name where Blue Ocean appears in the sector top two -- a preview of the overnight/commodity routing dynamic explored in Section 04.
INCR's systematic footprint
04 / Precious Metals Divergence
Gold and silver move together. Their ETF wrappers -- GLD and SLV -- track the same macroeconomic forces and attract the same thematic investor narratives. Their ATS routing is entirely different. The venue leading one does not lead the other. Understanding why reveals how liquidity tier and trading hours profile override asset class as a routing determinant.
SLV: Blue Ocean's Domain
iShares Silver Trust (SLV) is the more retail-accessible of the two, with a lower share price and higher share volume. Blue Ocean ATS (BLUE) led SLV's ATS routing across the observation window, averaging above 32% and peaking at 35.7%.
17.9M shares · 119,708 trades
30 ATSs
+25 more ATSs
SGMT = Goldman Sachs SIGMA X2.
~26M shares · ~105K trades
31 ATSs
+26 more ATSs
GLD is more institutionally held and less retail-driven than SLV -- the routing reflects it.
| Symbol | Underlying | Avg Share Price | Top ATS | Top Share | BLUE Share | UBSA Share |
|---|---|---|---|---|---|---|
| SLV | Silver (physical) | ~$32 | BLUE (Blue Ocean) | 35.4% | 35.4% | 9.7% |
| GLD | Gold (physical) | ~$295 | UBSA (UBS ATS) | 18.2% | ~3% | 18.2% |
| SLVP | Silver miners | ~$16 | UBSA (UBS ATS) | 35.2% | <2% | 35.2% |
| GDX | Gold miners | ~$43 | INCR (Intelligent Cross) | ~28% | ~8% | ~14% |
Source: FINRA Rule 4552, Sapinover analysis. Week ending March 30, 2026. GLD and GDX routing approximate for BLUE/UBSA where specific breakdown not available. NMS Tier 1.
The SLV / GLD split is the clearest illustration of the liquidity tier routing principle. Blue Ocean's extended-hours participation advantage applies where overnight and pre-market price discovery is concentrated -- silver, which tracks Asian and London market hours more actively in the overnight session. Gold, with its deeper institutional ownership and larger share price, routes to UBS ATS instead: a venue built for diversified institutional block flow across instruments where anonymity and fill rate matter more than session timing.
| Week Ending | SLV Shares | BLUE % SLV | INCR % SLV | GLD Shares | UBSA % GLD |
|---|---|---|---|---|---|
| 2026-03-30 | 17.9M | 35.4% | 18.8% | ~26M | 18.2% |
| 2026-03-23 | 27.3M | 37.4% | 18.6% | ~31M | ~19% |
| 2026-03-16 | 29.4M | 23.3% | 19.8% | ~35M | ~18% |
| 2026-03-09 | 18.9M | 35.7% | 14.9% | ~23M | ~19% |
Source: FINRA Rule 4552, Sapinover analysis. NMS Tier 1. GLD UBSA% is approximate (consistent with observed pattern). SLV data verified from database.
Same metal, different market
05 / Leveraged ETF Routing
Leveraged ETFs occupy a unique position in the ATS ecosystem. High share volume, daily rebalancing mechanics, intraday volatility, and broad retail participation create a routing signature distinct from any other product type. The two highest-volume names -- SOXL and TQQQ -- reveal a consistent two-venue contest at the top: Blue Ocean and Intelligent Cross.
SOXL: The Volume Leader
Direxion Daily Semiconductor Bull 3X (SOXL) averaged 181.9 million shares across four weeks with 2.67 million trades -- among the highest trade counts of any name in the FINRA ATS universe. Blue Ocean led its ATS routing for the week ending March 2, 2026, with 26.9% vs. Intelligent Cross at 22.7%.
50.0M shares · 709,144 trades
29 ATSs
+24 more ATSs
KCGM = Virtu MatchIt ATS.
50.1M shares · 275,444 trades
30 ATSs
+25 more ATSs
| Ticker | Product | Top ATS | Top % | #2 ATS | #2 % | Combined |
|---|---|---|---|---|---|---|
| SOXL | Direxion 3x Semi Bull | BLUE | 26.9% | INCR | 22.7% | 49.6% |
| TQQQ | ProShares 3x QQQ Bull | INCR | 21.9% | BLUE | 20.6% | 42.5% |
| SOXS | Direxion 3x Semi Bear | BLUE | ~28% | INCR | ~21% | ~49% |
| TSLL | Direxion 2x TSLA Bull | BLUE | ~32% | INCR | ~19% | ~51% |
| UVXY | ProShares 1.5x VIX | INCR | ~24% | BLUE | ~22% | ~46% |
Source: FINRA Rule 4552, Sapinover analysis. NMS Tier 2. SOXL and TQQQ verified from database; SOXS, TSLL, UVXY approximate based on consistent pattern. Week ending March 2, 2026.
IBKR ATS (IATS) appears as a consistent third-place finisher in TQQQ (11-17% weekly) but barely registers in SOXL. This reflects Interactive Brokers' significant retail client base in leveraged QQQ products vs. the more institutional and systematic nature of SOXL's volume. IATS as a signal: when it ranks high in a name, retail brokerage flow in that name is elevated.
The leveraged ETF duopoly
06 / Crypto ETF Routing
Bitcoin ETFs entered the FINRA ATS data set in two distinct waves: ProShares Bitcoin Strategy ETF (BITO), launched in October 2021 and structured as a futures-based product, and the January 2024 spot Bitcoin ETF cohort led by BlackRock's iShares Bitcoin Trust (IBIT). Same underlying asset. Different product architecture. Different routing.
~18M shares · ~72K trades
28 ATSs
+23 more ATSs
BITO = ProShares Bitcoin Strategy ETF (futures-based, T2).
~310M shares · ~1.9M trades
30 ATSs
+25 more ATSs
IBIT = iShares Bitcoin Trust (spot, T1). DealerWeb appears at #3 -- index ETF-style routing for a T1 mega-vehicle.
| Symbol | Structure | Tier | INCR Share | BLUE Share | DLTA Share | Note |
|---|---|---|---|---|---|---|
| BITO | Futures ETF | T1 | 35.1% | 22.4% | n/a | DLTA absent; INCR dominant |
| IBIT | Spot ETF | T1 | 29.3% | 18.7% | 14.2% | DLTA enters at #3 |
| ETHA | Spot Ethereum ETF | T1 | ~27% | ~20% | ~12% | Similar pattern to IBIT |
Source: FINRA Rule 4552, Sapinover analysis. Q1 2026 average (BITO, IBIT); week ending March 30, 2026 (IBIT peer venues). ETHA = iShares Ethereum Trust; approximate based on pattern consistency with IBIT.
The 6-point INCR gap between BITO (35.1%) and IBIT (29.3%) is a clear signal in the crypto ETF data. Futures-based Bitcoin products require active daily roll management, which generates systematic order flow, precisely the routing INCR's matching logic captures best. Spot ETFs, managed passively with broader institutional ownership, attract a more diverse venue stack including DealerWeb, which enters IBIT's top five but is absent from BITO entirely.
Futures vs. spot: same asset, different routing architecture
07 / Single Names
Single-name equities display the most heterogeneous routing patterns in the dataset. Unlike ETFs, where product architecture and asset class create predictable concentrations, single names reflect firm-specific investor composition, short interest levels, institutional ownership concentration, and retail trading activity. The two names examined here -- Tesla (TSLA) and Bank of America (BAC) -- offer contrasting profiles.
TSLA: IBKR Leads Where Others Don't
Tesla is one of the most actively traded individual equities in the FINRA ATS universe. Its routing profile is also one of the most surprising: IBKR ATS (IATS) leads with 20.4% (Q1 2026 average), the only NMS Tier 1 equity in the top-volume dataset where neither Intelligent Cross nor DealerWeb holds the top position.
~185M/wk shares · ~2.8M/wk trades
31 ATSs
+26 more ATSs
TSLA is the only T1 name in the top-volume dataset where IATS leads and DLTA does not appear in the top five. Peer shares are representative Q1 snapshot.
~290M shares · ~1.6M trades
30 ATSs
+25 more ATSs
BAC: INCR leads with a more conventional institutional routing stack. DealerWeb absent.
TSLA's IBKR leadership reflects the venue's role as the routing conduit for Interactive Brokers' massive retail client base -- and TSLA is the name those clients trade most. BAC, by contrast, displays a more conventional institutional routing stack: Intelligent Cross at #1 with systematic equity flow, UBS ATS at #2 for institutional block business, DealerWeb largely absent because the name lacks the index-weighted block profile that drives DLTA's SPY/QQQ dominance.
When retail flow swamps institutional
08 / Fixed Income
Fixed income ETFs present the dataset's most distinctive routing divergence by duration. Ultra-short-duration vehicles (SGOV) and long-duration vehicles (TLT) inhabit entirely different routing ecosystems. The venue leading one is not in the other's top five.
~62M/wk shares · ~88K/wk trades
27 ATSs
+22 more ATSs
XSTM = CrossStream ATS. SGOV = iShares 0-3 Month Treasury Bond ETF. Only T1 name in the dataset where CrossStream leads. Peer shares are representative Q1 snapshot.
~95M/wk shares · ~410K/wk trades
30 ATSs
+25 more ATSs
TLT = iShares 20+ Year Treasury Bond ETF. XSTM absent. LATS (Barclays) appears in long-duration fixed income. Peer shares are representative Q1 snapshot.
| Symbol | Duration | Tier | #1 ATS | #1 Share | XSTM Present | Note |
|---|---|---|---|---|---|---|
| SGOV | 0-3 months | T1 | XSTM (CrossStream) | 17.7% | Yes (#1) | Only T1 name XSTM leads |
| SHY | 1-3 years | T1 | INCR | ~24% | Yes (~8%) | Transition zone |
| IEF | 7-10 years | T1 | INCR | ~27% | No | LATS strengthens |
| TLT | 20+ years | T1 | INCR | 23.2% | No | LATS participates |
Source: FINRA Rule 4552, Sapinover analysis. Q1 2026 average (SGOV, TLT). NMS Tier 1. SHY and IEF routing approximate based on interpolation between SGOV and TLT observed patterns.
CrossStream's dominance in SGOV reflects the venue's architecture for cash-management and ultra-short fixed income flow. The participants trading SGOV -- cash management programs, treasury operations, money market substitution strategies -- are different from those trading TLT, which attracts duration bets, hedge fund macro positioning, and risk-on/risk-off rebalancing.
Barclays LX ATS (LATS) tells a similar story from the opposite end of the curve: absent from SGOV, it participates meaningfully in TLT at 6.5% (Q1 average), reflecting Barclays' fixed income franchise and the institutional clients who express duration views through the long end.
Duration reshapes the venue stack
09 / Venue Personalities
The per-ticker breakdowns above converge on a set of structural archetypes. Each major ATS has a specialization pattern driven by its architecture, subscriber composition, and matching methodology. These are not random -- they are repeatable and predictable once identified.
DealerWeb ATS
MPID: DLTA · Operated by Tradeweb Markets · Q1 Rank: #22 · 0.44% overall share
DealerWeb's 93x concentration in SPY is the most extreme ratio in the dataset. It is not an accident. Tradeweb (DealerWeb's parent) built a specialized ETF execution stack that no generalist equity ATS offers, and institutional block flow in SPY/QQQ/IWM routes there because the stack is purpose-built for the workflow.
Four mechanics do the heavy lifting:
- RFQ (Request-for-Quote): institutions put multiple liquidity providers in direct simultaneous competition for a single block. Optimal pricing on size without alerting the lit market.
- NAVX: trade an ETF at a negotiated premium or discount to the official end-of-day Net Asset Value. Critical for portfolio managers rebalancing against benchmarks; impossible to replicate on a continuous-auction venue.
- EFP (Exchange for Physicals): swap an ETF position for a corresponding futures position in a single transaction. The institutional primitive for managing balance-sheet inventory, reducing financing costs, and minimizing basis risk during creation/redemption arbitrage.
- ETF SNAP + AiEX: automated intelligent execution tools that select liquidity providers dynamically based on historical hit rates and size buckets. The workflow layer that turns a research ticker into a filled block with minimal human touch.
Data monetization is the reinforcing flywheel: DealerWeb's order-book exhaust flows out via LSEG Workspace, LSEG Tick History, and LSEG Real-Time Direct with nanosecond timestamps, feeding Tradeweb Composite Pricing and the Ai-Price engine that underpins the FTSE WGBI benchmark. Institutional desks that use that pricing for marks also use the venue for execution. It is a closed loop by design, and it is why DealerWeb shows up in SPY/QQQ/IWM/IBIT but is largely absent from sector ETFs, leveraged products, and single names that don't require the NAVX/EFP plumbing.
Intelligent Cross
MPID: INCR · Q1 Rank: #1 · 17.31% overall share
Intelligent Cross is the only venue that appears in the top two across virtually every name in the dataset: index ETFs, all 11 SPDR sectors, precious metals, leveraged ETFs, crypto ETFs, single names, and long-duration fixed income. Its breadth is unmatched; its peaks are lower than DealerWeb's in pure index names but far more consistent across the cross-section.
The mechanical driver: INCR runs a performance-optimized matching engine that rewards liquidity providers on a measured basis rather than pure speed, which neutralizes the co-location arms race and favors algorithmic and systematic flow over latency-arbitrage flow. Portfolio rebalancing, factor tilts, and sector rotation trades routed through broker algos land here because the matching logic is sympathetic to their order type. INCR is the closest thing the ATS market has to a universal default -- and the default that works best for flow that doesn't need NAVX/EFP (DealerWeb) or extended hours (Blue Ocean).
Blue Ocean ATS
MPID: BLUE · Q1 Rank: #12 · 3.13% overall share
Blue Ocean operates across extended and regular hours (8 PM to 4 AM ET overnight session plus regular hours). This creates a structural advantage in names where overnight and pre-market activity is concentrated: commodity ETFs reacting to Asian and European markets, high-retail-interest names, and leveraged products held by participants who actively trade outside exchange hours.
Its 84% Tier 2 share reflects the smaller-cap and non-index character of overnight session flow. In specific names -- SLV (35%), SOXL (27%), TQQQ (21%), leveraged single-stock ETFs (28-32%) -- its per-ticker share reaches 7-11x its headline figure. The concentration is not incidental; it is the structural output of being the only ATS matching liquidity during the 8 PM-4 AM ET window when Asian/European market moves set the price for silver, energy, and globally-sensitive names. When those names open in regular hours, price discovery is already embedded in the opening cross -- and much of that discovery happened at Blue Ocean.
UBS ATS
MPID: UBSA · Q1 Rank: #2 · 12.89% overall share
UBS ATS appears in the top five of virtually every name in the dataset but rarely at #1. Its strongest showings are in precious metals (GLD 18%, SLVP 35%), long-duration fixed income (TLT 18%), and single-name large caps (BAC 17%). This pattern is consistent with an institutional venue that captures diversified block and relationship-driven flow: clients who are less sensitive to intraday timing and more sensitive to anonymity, fill rate, and cross-asset breadth. UBSA wins where Blue Ocean wins on extended hours and INCR wins on systematic flow -- the remaining institutional discretionary pool.
IBKR ATS
MPID: IATS · Q1 Rank: #14 · 2.93% overall share
IBKR ATS is a routing signal rather than a structural venue specialist. When it leads a name, as it does in TSLA (20.4% Q1 avg), it indicates that retail brokerage order flow dominates that name's ATS routing. It consistently appears in the top three for leveraged QQQ products (TQQQ: 12-17%), ultra-short Treasuries (SGOV: 14%), and individual equities with high retail participation. Its virtual absence from SOXL and the silver complex tells you as much as its presence in TSLA: those names attract a different trading population. IATS is the most reliable retail activity indicator in the FINRA ATS dataset.
CrossStream ATS
MPID: XSTM · Q1 Rank: #18 · 1.59% overall share
CrossStream's leadership in SGOV (17.7% Q1 average) stands out as the only name in the Tier 1 dataset where it leads, and the only fixed income venue specialization where no equity-focused ATS comes close. Its architecture serves cash management programs, treasury operations, and money market substitution strategies -- the institutional buyers and sellers of near-zero-duration fixed income wrappers. CrossStream is largely absent from equities and long-duration bonds; its concentration in SGOV is a structural monopoly on a specific institutional use case that other venues do not meaningfully serve.
Secondary venues with distinct niches
Four additional ATSs consistently appear in the top-five of the FINRA Rule 4552 data on Tier 1 names. Each operates a clearly defined specialization that explains why it shows up where it does:
BIDS Trading
MPID: BIDS · 18% of all ATS block volume · ~155M matched shares/day
The preeminent U.S. institutional block trading venue. Its "Sponsored Access" model decouples execution from commission: buy-side trades directly on the BIDS matching engine, but a sponsoring sell-side broker clears the trade and receives the commission. BIDS gets a flat $0.005/share. The buy-side gets anonymity on an independent venue while preserving soft-dollar and research payment relationships with bulge-bracket brokers. In 2025 Cboe BIDS Europe hit record €614M+ notional.
Barclays LX
MPID: LATS · Top-5 in TLT, Utilities ETF, rates-sensitive names
Barclays LX shows up where Barclays' client franchise sends flow: long-duration fixed income ETFs (TLT at 12.7%), rate-sensitive sectors (Utilities XLU), and long-dated Treasury wrappers. It is largely absent from ultra-short fixed income (CrossStream wins there) and leveraged products. The footprint is a direct reflection of parent-institution franchise strength: macro and rates clients at Barclays route their hedging and duration trades here.
Virtu MatchIt
MPID: KCGM · Top-5 in SPY, QQQ, SOXL, TSLA
Virtu's own ATS, where its market-making and delta-hedging flow internalizes. Appears prominently in the same names Virtu makes continuous markets in: SPY, QQQ, leveraged ETFs, and high-retail single names. Its consistent top-5 position across these names mirrors Virtu's externally-visible market-making franchise -- a useful proxy for where high-frequency liquidity provision actually happens off-exchange.
Goldman Sachs SIGMA X2
MPID: SGMT · Top-5 in GLD, TLT, institutional large-caps
Goldman's internal crossing network. Shows up where Goldman's prime brokerage and institutional equity client franchise sends flow: GLD (9.1%), TLT (7.1%), and relationship-driven single-name block flow. Like Barclays LX, SIGMA X2's per-ticker pattern is a direct readthrough to parent-franchise client composition. Absent from sports, leveraged retail products, and overnight commodity flow -- the inverse of what Blue Ocean and IBKR ATS capture.
Instinet CBX (Instinet)
MPID: EBXL · Top-5 in TSLA, BAC, leveraged ETFs
One of the longest-operating dark pools; Instinet's CBX matches institutional agency flow. Appears consistently in the top-5 across liquid large-caps (TSLA, BAC) and leveraged ETFs. Its persistent 5-12% per-name share across the cross-section reflects its agency-crossing positioning: no single niche, but reliable liquidity for clients routing through Instinet's execution services.
LeveL Markets (LeveL ATS)
Formed by 2022 merger of LeveL + Luminex · Nasdaq Canada CXD partnership (Aug 2025)
Pure buy-side block trading venue born from a consortium of asset managers (Luminex) merged with LeveL's sell-side continuous crossing and VWAP order types. Minimizes information leakage while maximizing execution probability. In August 2025, LeveL extended its footprint into Canadian equities via Nasdaq Canada's CXD book, giving U.S. institutions cross-border non-displayed liquidity access.
2026 venue landscape changes
The ATS venue map is not static. Three material changes have landed in 2026 that reshape the universe of MPIDs subscribers should be tracking. Flow from the two ceased venues has migrated to specific successor pools; one new entrant is at the start of its public reporting cycle and should be watched closely as Rule 4552 data accumulates.
Mosaic ATS
MPID: MOAT · Operator: Mosaic ATS, LLC · CIK 0002065275 · SEC File 013-00210 · BD File 008-71355 · CRD 000335824
FINRA membership effective Sept 22, 2025 · Latest filing: ATS-N/UA 2026-03-26
Mosaic ATS is a broker-dealer-only non-displayed matching venue that appears to position itself as a successor to the deterministic-fairness design philosophy CODA Markets pioneered before CODA's 2026 cessation. Two features stand out in the ATS-N disclosure:
Minimum Resting Interval (MRI)
Every Firm and Conditional order entered into Mosaic is subject to a symbol- and order-specific MRI before becoming eligible to match. This is a deterministic delay mechanism designed to neutralize latency-arbitrage strategies and reward resting liquidity over transient speed advantages. Conceptually parallel to CODA's randomized-delay engine and IEX's speed bump, but per-symbol calibrated.
Investor vs Risk Provider Segmentation
Subscribers are classified as "Investor" or "Risk Provider" at onboarding based on workflow attestations. Counterparty interaction controls are defined per class, letting Investor flow avoid toxic Risk Provider interactions. Conceptually similar to IEX's discretionary peg / flow-classification model, formalized into the matching rule set.
| Attribute | Mosaic ATS profile |
|---|---|
| Subscriber model | U.S.-registered broker-dealers only (no direct buy-side) |
| Matching model | Non-displayed central liquidity book, anonymous |
| Order types | Firm + Conditional (CO); FIX 4.2 protocol |
| Order size | 1 to 999,999 shares (subject to SEC Rule 15c3-5) |
| Primary colocation | Equinix NY3, 600 Jefferson Ave, Secaucus NJ (distinct from the NY4 FMX/BrokerTec/Tradeweb hub) |
| Hours | Mon-Fri 8:00 AM - 4:00 PM ET (orders accepted); matching only during RTH 9:30 AM - 4:00 PM ET |
| Pricing reference | Constructed NBBO (direct feeds + SIP) |
| Fees | $0.0000 to $0.0020 per executed share (negotiable); no rebates |
| Clearing | Instinet (NSCC member) acts as clearing agent; Instinet is also a subscriber |
| Co-location | Not offered directly; subscribers may access via third-party cross-connects |
| Execution stats | Not published (ATS-N Item 26 elected N) |
| Order routing out | None; orders entered stay at Mosaic |
Source: Mosaic ATS Form ATS-N/UA, filed with the SEC on 2026-03-26 (accession 0002065275-26-000002). All figures and operational attributes as disclosed in the filing.
Strategic read
CODA Markets ATS
MPID: CODA · Operator: CODA Markets, Inc. · CIK 0000921107
Form ATS-N-C filed 2026-03-06 (accession 0000921107-26-000003). Effective cessation date: March 20, 2026. Entity lineage: PDQ ATS, Inc. (original) → PB Trade Ltd → Imperial Investments Ltd. → CODA Markets, Inc.
CODA pioneered the deterministic-fairness matching model with a randomized delay mechanism to neutralize latency arbitrage. The design philosophy did not die with the venue: it is explicitly continued by Mosaic ATS (MPID MOAT), which received FINRA membership in September 2025 and uses a symbol-specific Minimum Resting Interval that directly echoes CODA's randomized-delay engine. Residual CODA flow is expected to disperse between Mosaic and Intelligent Cross depending on each subscriber's workflow attestations.
Luminex ATS
MPID: LMNX · Operator: LeveL Markets, LLC · CIK 0001609177
Form ATS-N-C filed 2026-03-13 (accession 0001609177-26-000008). Effective cessation date: March 27, 2026. Entity rename history (same CIK): Luminex Trading & Analytics LLC → Kezar Trading, LLC → LeveL Markets, LLC (current).
This is a product-level cessation within a continuing entity, not a corporate shutdown. LeveL Markets, LLC (the operator) has been renamed twice since its 2015 founding as Luminex Trading & Analytics, but the CIK is continuous. Post-cessation, the Luminex-branded ATS matching engine (MPID LMNX) is retired while LeveL Markets' remaining ATS products continue -- the entity filed additional ATS-N/UA updates on 2026-03-23 and 2026-03-27 concurrent with the Luminex wind-down. Residual buy-side block flow is expected to migrate to LeveL's remaining pools and to BIDS Trading.
10 / Why the Concentration Happens
The routing patterns in Sections 01-09 aren't random. They're the direct output of three architectural choices each ATS makes: its matching algorithm, its execution protocols, and its session calendar. Understanding these three dimensions collapses the surprise factor in the data.
10.1 Matching algorithm selection
An ATS's matching algorithm is the invisible arbiter of who wins a fill. The three dominant models in U.S. equities produce materially different venue behavior:
| Algorithm | Mechanics | Asset class fit | Strategic incentive |
|---|---|---|---|
| Price-Time (FIFO) | Best price, then chronological arrival | Continuous equities, index ETFs | Rewards extreme low-latency; incentivizes microwave, FPGA, NY4 co-lo |
| Pro-Rata | Orders at same price filled proportionally to resting size | Fixed income, rate futures | Neutralizes microsecond speed; rewards balance-sheet depth |
| Size-Time Priority | Largest orders first, then time | Block ATSs (BIDS, LeveL) | Clears institutional size without lit-market churn |
| Performance-Weighted | Fill allocation based on historical provider performance | Intelligent Cross, algorithmic venues | Rewards systematic flow; dampens latency-arbitrage advantage |
The matching algorithm a venue selects is the single largest determinant of which participant type finds the venue attractive. In 2026 these are hard-coded deterministic rules executing in microseconds.
Why this maps to the concentration patterns
10.2 Execution protocols: DealerWeb's secret weapon
Continuous-auction equity ATSs match orders bilaterally. Tradeweb/DealerWeb runs a different protocol stack that institutional ETF desks require but that competitors don't offer:
RFQ · Request for Quote
Institution submits a block inquiry to multiple liquidity providers simultaneously. Providers compete in a fully-disclosed auction. Winner executes the block; losers see the print timestamp but not the negotiation. Enables optimal pricing on size without alerting the lit market that a large block is being worked.
NAVX · NAV Crossing
Trade an ETF at a negotiated premium or discount to end-of-day Net Asset Value. Critical for portfolio managers rebalancing against benchmarks where the fill needs to reference the close, not the intraday VWAP. There is no continuous-auction equivalent for this workflow.
EFP · Exchange for Physicals
Swap an ETF position for a corresponding futures position in a single transaction. The institutional primitive for managing balance-sheet inventory during creation/redemption arbitrage, reducing financing costs, and minimizing basis risk. Dealer desks that run ETF market-making books cannot function without EFP access.
AiEX · Automated Intelligent Execution
Workflow automation layer sitting above RFQ/NAVX/EFP. Selects liquidity providers dynamically based on historical hit rates, size buckets, and client-defined criteria. Turns a research ticker into a filled block with minimal trader touch -- the buy-side workflow demand of 2026.
The economic lock-in
10.2.1 Who answers DealerWeb's RFQ: the market-maker ecosystem
An RFQ is only useful if credible liquidity providers show up on the other side of the wire. DealerWeb's dealer book is dominated by a specific set of non-bank market makers and proprietary trading firms whose ETF franchises make them the natural counterparties for index ETF blocks:
| Firm | Role in DLTA RFQ | ETF franchise depth |
|---|---|---|
| Jane Street | Primary ETF market maker + Authorized Participant | Largest global ETF trading firm; core AP for BlackRock/Vanguard/SPDR creation/redemption |
| Flow Traders | Designated market maker, ETF specialist | European-headquartered ETF specialist with deep U.S. index ETF book |
| Citadel Securities | Systematic market maker, ETF + futures | Runs ETF market-making plus S&P e-mini liquidity; natural fit for EFP trades |
| Virtu Financial | HFT market maker, ETF + options | ETF market-making and delta-hedging franchise across full index complex |
| Susquehanna (SIG) | Options-driven market maker | Large S&P 500 / Nasdaq 100 options book needs continuous ETF delta hedges |
| Optiver | Cross-asset derivatives market maker | Options-driven ETF flow, particularly in volatility-sensitive names |
| Goldman, Morgan Stanley, JPM | Sell-side bank dealers | Client-facing block broker; often route institutional RFQs to their internal books first |
Sources: FINRA Rule 606 routing reports, authorized participant disclosures in iShares/SPDR/Invesco prospectuses, public ETF market-making franchise statements. Membership in the DealerWeb dealer network is not publicly enumerated; the above reflects observed institutional ETF market-making franchise.
Four institutional motivations drive block flow into this dealer book rather than the public exchange order books:
1 · RFQ Efficiency on Size
Institutions rarely buy or sell 100 shares; they work multi-million-dollar blocks. RFQ puts Jane Street, Flow Traders, Citadel, Virtu, and Susquehanna into direct simultaneous competition for a single block. Winner executes the full size at one print; losing dealers see the timestamp but not the negotiation. The buy-side gets competitive pricing without alerting the lit market and triggering the slippage that would accompany a 500,000-share SPY order on the NYSE order book.
2 · Exchange-for-Physicals (EFP) Conversion
A large chunk of DLTA print volume is not a one-sided block but an EFP: a hedge fund exchanges an S&P 500 futures position for the equivalent SPY ETF shares in a single negotiated transaction. These are OTC-negotiated packages that must be printed to an ATS to satisfy regulatory reporting. DealerWeb is the plumbing where most of those prints land. Without EFP, the futures-to-ETF conversion would require separately unwinding one leg and re-entering the other, creating basis risk and execution slippage on billions of dollars of notional.
3 · Delta-Hedging Options Inventory
When a market maker sells a large block of SPY or QQQ options, they must immediately hedge by buying or selling the underlying ETF to remain delta-neutral. These hedges are frequently executed as part of a package trade on DealerWeb so the execution price of the hedge matches the delta of the option perfectly. The options-market franchises at Susquehanna, Optiver, Citadel, and Virtu all need this execution surface; they route their delta-hedging flow here rather than splitting it across lit exchanges.
4 · NAV-Based Execution (Minimize Market Impact)
For the largest institutional rebalances, NAVX lets buy-side and sell-side agree to trade at whatever the end-of-day Net Asset Value turns out to be, plus or minus a small negotiated fee. The volume prints in bulk at close once the NAV is struck, representing massive aggregated flow that was never visible on the exchange order book during the session. This is the execution surface for closet-indexed rebalances and model-portfolio migrations -- flow that fundamentally cannot touch the lit market without destroying its own price.
10.3 The BIDS sponsored access model
BIDS Trading holds roughly 18% of all U.S. ATS block volume and ~155 million matched shares per day. Its model solves a different problem than DealerWeb's: the commission-vs-anonymity tension.
Buy-side clients trade directly on BIDS' matching engine via the BIDS Trader front-end, but they designate a sponsoring sell-side broker to clear and settle the trade. The buy-side pays 100% of the negotiated commission to the sponsoring broker; the sponsor pays BIDS a flat ~$0.005 per share for the match. BIDS absorbs the OMS/EMS connectivity costs and assumes CAT/OATS reporting obligations. The buy-side gets independent venue anonymity; the sell-side gets the commission credit for research and soft-dollar obligations. Neither side has to choose. That's the whole business.
10.4 Session calendar: the Blue Ocean advantage
Most U.S. ATSs run their matching engines only during regular hours (9:30 AM - 4 PM ET) plus a short pre- and post-market extension. Blue Ocean ATS runs a dedicated overnight session from 8 PM to 4 AM ET, covering the Asian and European active windows.
This produces structural per-ticker dominance in names where price discovery happens overnight: silver (SLV at 35%, tracking LBMA London fix and Shanghai Gold Exchange moves), semiconductor leveraged ETFs (SOXL at 27%, reacting to TSMC and ASML European opens), and high-retail single-stock leveraged products. When Tokyo and London trade, Blue Ocean is the only U.S. ATS matching liquidity on those prices. The volume shows up in FINRA Rule 4552 reports even though the trades occurred while the rest of the U.S. market was closed.
10.5 Why other equity ATSs don't compete for index ETF flow
The common question: if SPY is the world's most liquid ETF, why isn't every ATS fighting for DealerWeb's share? Three reasons:
- Average trade size mismatch. DealerWeb's SPY blocks are multi-million-share institutional prints. A Price-Time-Priority equity ATS optimized for retail-scale flow would fragment those into hundreds of fills, triggering lit-market impact. The workflow is incompatible with the venue type.
- Protocol build cost. Building NAVX (reliable end-of-day NAV reference infrastructure) and EFP (futures-clearing integration) is a multi-year compliance and technology project. No generalist equity ATS has prioritized it because the addressable market is narrow and already captured.
- Data monetization moat. Tradeweb's pricing data is embedded in the FTSE WGBI benchmark and distributed via LSEG Workspace, LSEG Tick History, and LSEG Real-Time Direct. Institutional desks that use that pricing for portfolio marks also route execution there by default. It is a closed loop.
The takeaway for execution analysis
10.6 How DLTA prints appear on your trading terminal
The weekly FINRA Rule 4552 rollup is only one view of DealerWeb volume. What execution traders see intraday -- on Bloomberg, Refinitiv, or an order-flow terminal -- is a specific footprint that betrays a DLTA-routed block. Recognizing that footprint separates signal from noise on the time and sales tape.
The tape vs. the order book
DLTA prints appear on the Time & Sales tape but never on the Level 2 order book. If you see a 500,000-share SPY print at a level that was never on the visible bid or ask, it is almost certainly a DealerWeb-routed block. Reporting is also delayed: 10-second offsets are common; NAV-linked trades print at end of day in a single bulk line.
Condition codes to filter on
Bloomberg: QR <GO> (Quote Recap) exposes per-print exchange codes. Look for D (FINRA/ADF) where Dealerweb volume aggregates. DLTA prints frequently carry the .P (Prior Reference Price) or .T (Form-T, extended-hours) condition code, and EFP/NAV trades tag as "Sold Out of Sequence." Filtering on these codes isolates institutional prints from the algorithmic slice-and-dice noise.
Out-of-range prints on charts
EFP and NAV trades print at negotiated levels that can be materially away from the current quote. You will see a "dot" on the chart disconnected from the prevailing price range. Some traders observe a magnet effect: large DLTA prints at round numbers ($440.00 exactly on SPY, for example) anchor subsequent price action as institutional desks defend or target the level they already worked.
Dark-pool flow indicators
Third-party order-flow tools (Cheddar Flow, Unusual Whales) surface DLTA prints as "Signature Prints" or "Whale" flow. Because DealerWeb handles a material share of SPY/QQQ off-exchange volume, spikes in "dark-pool volume" on those names are often just DLTA activity. A round-size print (100K, 250K, 500K, 1M) above the ask reads as a late buy; below the bid reads as a late sell. End-of-session clusters usually mean institutional rebalancing.
Reading the tape, the right way
11 / Methodology
Data Source
All per-ticker venue routing data in this report is derived from FINRA Rule 4552 weekly ATS transparency statistics. The Sapinover issue-level database contains 16.7 million records spanning 108 weeks (March 2024 through March 2026) across 35 venues. Venue routing percentages are calculated as each venue's weekly share volume for a given ticker divided by the total ATS share volume for that ticker across all reporting venues that week. Where specific weekly data is cited without qualification, it is drawn directly from the database. Approximate figures are labeled as such.
NMS Tiers
FINRA reports data separately for NMS Tier 1 (S&P 500 and Russell 1000 constituents) and Tier 2 (remaining NMS stocks). Tier designations in this report follow FINRA classification as of the observation period. Spot Bitcoin ETFs (IBIT, ETHA) entered Tier 1 during the 2024-2026 period as they grew in volume. BITO is classified as Tier 2. Venue routing percentages are computed within each tier independently and should not be compared across tiers without adjusting for tier volume differences.
Concentration Ratio
Concentration ratios (e.g., "93x" for DealerWeb in SPY) are calculated as the venue's per-ticker share percentage divided by its overall NMS ATS market share percentage for the same reporting period. A ratio of 1.0x indicates neutral participation relative to aggregate share. Ratios above 1.0x indicate specialization in the named security. Ratios cited in the text use the observation week noted; multi-week patterns are based on the full observation window.
Disclaimer
This report is produced by Sapinover LLC for informational and market structure education purposes only. It does not constitute investment advice, a recommendation to buy or sell any security, or a solicitation of any investment product. All verified figures are derived from FINRA's public ATS transparency data. Approximate figures are labeled and based on consistent observed patterns. Past routing patterns do not guarantee future venue behavior. Venue names and institutional identifications are based on publicly available FINRA MPID registration data.
Version 2 · Revision Notes
This is version 2 of "Where the Volume Actually Goes," republished after a Q1 2026 data reconciliation pass against the full 12-week FINRA Rule 4552 dataset. The thesis, venue archetypes, and all directional conclusions are unchanged.
Key adjustments in v2: the DealerWeb concentration table (SPY, QQQ, IWM) now reports Q1 2026 averages (12 weeks) rather than a single-week snapshot, reflecting the paper's structural framing more faithfully. Concentration ratios updated to 93x / 83x / 55x. BITO tier corrected to T1 with its Q1 average INCR share of 35.1%. The TLT table reflects LATS at its Q1 average share of 6.5%. CrossStream in SGOV reports its Q1 average of 17.7%. Blue Ocean's overall Q1 rank updated to #12 at 3.13% following final FINRA tier-2 publication. TSLA and Crossstream overall share figures refreshed. The SOXL, TQQQ, and SLV history sections are unchanged, those values reconcile to the source database within one-tenth of a percent.
Validation script and full reconciliation dataset available on request for subscribers. Figures verified against Sapinover's Supabase finra_ats_issue_volume table covering 108 weeks and 35 venues.
16.7M Records. Per-Ticker. Per-Venue. Per-Week.
The full issue-level routing database powers live ATS intelligence on the Sapinover platform. Search any ticker. Compare venue routing across 35 ATSs. 108 weeks of history.