Looking for what we cover? Venues and markets that are live, in validation, or being mapped have their own page.
Transparency & Audit Center
Methodology & Transparency
Every metric on AfriFlux is derived from observable blockchain activity. This page explains what we measure, how we measure it, and where the limitations are.
Venues tracked
19
Chains mapped
13
Deposit addresses identified
703,919
Retail volume mapped
$973.90M
Retail ratio
80.40%
Last updated
2026-09-15
Section 02
What Counts As Retail?
The classification engine sorts deposit activity by size and behaviour. Only genuine retail flow reaches the headline metrics.
Tier 1Under $1KRetail
Tier 2$1K–$5KRetail
Retail Whale$5K–$25KRetail
Large Whale$25K+Excluded
BusinessHigh frequencyExcluded
Addresses classified
703,919
Retail wallets
703,569
Retail volume
$973.90M
Excluded volume
$237.67M
Section 03
Retail Filtering
Raw stablecoin volume is not retail volume. This is why retail classification exists — shown aggregated across every tracked venue, never broken down.
Observed · filtered venues
$825.61M
All stablecoin volume through identified deposit addresses at the 7 venues the engine is applied to.
Retail · filtered venues
$587.94M
After removing large-deposit and business flow. A SUBSET of market retail volume — the other venues are unfiltered and add to it.
Removed (non-retail)
$237.67M
Large-deposit, business and institutional deposit flow. All of it comes from these venues; the unfiltered ones remove nothing.
Retail ratio · filtered
71.20%
Retail share of deposit-address volume at these venues. Market-wide the ratio is higher, because the unfiltered venues are 100% retail.
A large share of observed onchain activity consists of liquidity operations, institutional settlement and business flows. AfriFlux removes these flows before calculating retail metrics.
Retail$587.94M
Non-retail$237.67M
Section 03 · venue ratios
Venue-Level Retail Classification
Retail is not a fixed percentage. Different venue models attract different activity — some are overwhelmingly retail, others carry heavy liquidity and settlement flow. A venue that serves individuals only reads 100%: the engine is not applied to it, so there is nothing to remove. These are classification ratios only; no infrastructure, addresses or relationships.
Category
Monica (Crypto offramp) settles ~100.00% retail, while Quidax (CEX) registers ~59.00% — composition varies by venue architecture. The framework is the same; the composition differs.
Section 04
How Deposit Addresses Are Identified
Every deposit address enters existence through one provisioning act — the onchain event that brings it under venue control. AfriFlux anchors on that act in whatever form it takes, then confirms each address through its sweep path and collection wallet. The provisioning act differs by architecture. The confirmation chain does not.
Layer AProvisioning anchors— varies by architecture
Fundedanchor on the funding source + bandDeployedanchor on the factorySponsoredanchor on the flow itself
A1Gas Pre-Funding
EVMfundedMost common
Addresses are funded with a precise native-token band from a common source before first use. The gas funder is the anchor; enumerating everything it funds within the band yields the full deposit-address set.
Observed signals
Gas source walletFirst-tx fundingFunding band/cadence
High94% coverage
A2Activation Band
Tronfunded
Tron addresses must be activated before use. An activator sends a precise TRX wake-up amount to each new deposit address. Enumerating every address that received the activation band gives the complete set; threshold-gated sweeps can leave a funded-but-unswept population, itself a signal.
Observed signals
Activator walletTRX activation bandFunding cadenceUnswept-balance population
Applicability: funded architectures
A3Fee-Payer Signing
Solanasponsored
No per-address funding exists on Solana. A fee-payer signs every sweep transaction instead. We paginate the fee-payer's signed transactions and extract transfers into the master/collection — grouping by owner, not token account, to count true users.
Observed signals
Fee-payer walletSweep-tx signerTransfers to masterOwner-level grouping
Applicability: sponsored architectures
A4Factory Deployment
Contract-deployeddeployedNo funding signal
Some venues never fund addresses — they deploy them from a single factory contract. The factory's deployment events enumerate every deposit address directly, each deriving deterministically from the factory.
Observed signals
Factory contractDeployment eventsDeterministic derivation
Applicability: deployed architectures
A5Sponsored Execution / Gasless
EVM · EIP-7702sponsoredNo funding signal
Some venues fund nothing. A relayer sponsors each sweep through an account-delegation contract (EIP-7702), so no funding event exists to anchor on. The anchor flips to the collection wallet: every distinct address that has swept into the collector is a deposit address — capturing relayer-swept and self-sweeping users alike.
Observed signals
Relayer / sponsor walletDelegation contractCollection fan-in
Applicability: sponsored architectures
A6Relayer Execution Events
EVM · EIP-7702sponsoredNo funding signal · no owned sink
Some venues fund nothing AND own no collection wallet — a relayer drives each user's delegated account (EIP-7702) and settles straight to shared external infrastructure. With no funding event and no owned sink to fan into, the anchor becomes the relayer's own execution log: every distinct account a relayer key drives is a deposit address. Enumerating every account executed by the relayer keys yields the complete set — including fund-and-hold and withdraw-only users a flow-based anchor would miss. An account at rest carries no fingerprint; it only becomes attributable once a relayer acts on it.
Observed signals
Relayer key (same across chains)Execution eventsDriven-account addressSettles to external/shared sink
Applicability: sponsored architectures
A7Pool Convergence
SolanasponsoredPool-anchored · two rails
Some venues run no per-user address at all — every flow converges on a single settlement pool that sits on one end of every transfer. The pool is the anchor: paginate its token accounts and keep every leg that touches it. Two rails then resolve from the one pool. A direct rail, where the customer sends straight to the pool — the sender is the user, captured natively. And a swept rail, where the customer funds an assigned deposit address that the pool fee-pays a sweep from; the funding leg one hop back is the user's deposit. Counting both rails — without double-counting the sweep against its funding — gives the true set.
Observed signals
Settlement pool + token accountsDirect senders (rail 1)Funded-then-swept addresses (rail 2)Pool fee-paid sweep
Applicability: sponsored architectures
Layer BConfirmation chain— common to all five
Provisioning anchorSweep pathCollection walletVenue infrastructure
B1Sweep Analysis
Deposits sweep to a collection wallet on a predictable schedule. The sweep path links addresses to infrastructure.
Observed signals
Sweep destinationSweep timingResidual balance
High88% coverage
B2Collection Wallet Anchoring
Multiple deposit addresses resolve to shared collection wallets, anchoring them to one venue.
Observed signals
Shared collectorFan-in patternReuse across chains
Medium79% coverage
Section 05
What We Observe vs What We Infer
Honesty about provenance is part of the brand. Directly observed data carries high confidence; modelled data is labelled as such.
Section 06
Metrics Library
The official source of truth for every number on the platform. Each metric carries its definition, methodology, confidence and limitations.
Retail Volume
High confidenceVolumeDefinition
Total stablecoin value deposited by retail wallets into tracked venue infrastructure over a period.
Methodology
Inbound deposits to discovered deposit addresses are summed, then passed through the retail classification engine to remove business, institutional and liquidity flows.
Limitations
Dominated by USDT and USDC; cUSD and PYUSD are broken out individually and other minor stablecoins are grouped under 'Others'. Reflects tracked venues only, not the entire ecosystem.
Section 07
Known Limitations
Stated plainly, not hidden. Trust is built by showing the edges of what the data can and cannot say.
Wallet ≠ Human
Cross-chain and intra-chain wallet reuse means address counts are not person counts.
Impact: Inflates wallet metricsHigh
Geography
Country is attributed at the venue level, not the individual depositor level.
Impact: Coarse geographyMedium
Outbound Flows
Withdrawal destinations are modelled from clustering rather than directly observed.
Impact: Lower net-flow confidenceMedium
Coverage
Only mapped venues, chains and assets are included; the universe expands over time.
Impact: Universe ≠ ecosystemHigh
USDT, USDC, cUSD & Others
USDT, USDC, cUSD and PYUSD are broken out individually; other minor stablecoins are grouped under 'Others' and remain in volume totals.
Impact: Broken out + groupedHigh
Settlement Infrastructure
Operational settlement structure is reconstructed, never published by venues.
Impact: Inference, not disclosureMedium
Section 08
Current Coverage
Exactly what AfriFlux covers today. Coverage expands continuously and every change is logged below.
Tracked categories
ExchangesLiveNeobanksLiveOfframpsLivePrediction MarketsLive
Tracked assets
BUSDPYUSDUSDCUSDTcUSD
Tracked chains
AptosArbitrumAsset ChainAvalancheBaseBSCCeloEthereumLiskOptimismPolygonSolanaTron
Tracked infrastructure3 protocols
Textile FX
On-chain FX rail · 5 networks
Paycrest
Settlement protocol · 7 networks
HyperFX
Cross-chain FX protocol · 3 networks
Measured from each protocol's own contracts, and not by the same method:
- Paycrest settles in contract state — amount minus residual, so a refunded order contributes zero by construction.
- HyperFX settles on a published fill event, with expiry and refund carried as separate outcomes of the order.
- Textile FX emits no amounts at all — its Fill event names only the parties, so value is reconstructed from the ERC-20 transfers of the same transaction, and only its African corridors are counted.
None of it enters the market total — it settles into venues the page already counts.
Section 09
Methodology & Coverage Updates
Every change to how AfriFlux measures, and everything it adds to coverage, recorded with its impact.
1–10 of 20
27 Aug 2026New venue
Monica added — Nigerian crypto-to-naira offramp
Monica, a Nigerian offramp that converts stablecoins to naira with instant bank payouts, is now mapped and live across the market, category and venue pages: $163.79M lifetime retail deposits from ~66,500 addresses across five chains (Tron, Ethereum, BSC, Solana, Base), led by Tron at 46%. Coverage starts March 2024. Offramps is now a five-venue category, and Monica alone is larger than every other offramp we track combined ($113.10M) — it is the second-largest venue in our coverage by measured stablecoin flow, behind only Quidax. It is classified retail-only, so the full deposit volume is reported as retail: the median deposit is small and the largest single depositor is roughly 1% of the book, so there is no business or liquidity flow to strip. Monica is inflow-only — it collects deposits and settles to treasury, so no retail withdrawal side is shown.
CoverageOfframps
27 Aug 2026Methodology
Solana deposits corrected for the SPL token-account gap
On Solana, an inbound SPL transfer credits the depositor's associated token account, not the owner address we hold in our deposit-address table. Matching on the owner address therefore misses inbound transfers and undercounts the chain.
For Monica this understated Solana by $2.50M — $5.35M observed against $7.84M actually received, a 32% shortfall on that chain. Monica's Solana flow is now measured on the sweep leg instead (deposit address to treasury), where the deposit address is the sender and matches exactly. The depositor set is identical either way — 4,591 addresses — so only the measured leg changes, not who is counted.
Monica's lifetime figure moves from $161.30M to $163.79M. No other venue's figure changes: Paj Cash already used a custom Solana shape for an unrelated reason, and the remaining Solana venues are being reviewed for the same gap. The durable fix is to index Solana token accounts at ingestion, after which this correction is retired.
OfframpsCoverage
25 Aug 2026New venue
Raenest added — Nigerian cross-border neobank
Raenest (consumer product Geegpay), a Nigerian fintech giving African freelancers, remote workers and businesses multi-currency USD/GBP/EUR accounts, cards and cross-border payments, is now mapped and live across the market, category and venue pages: $3.66M retail-attributed deposits from ~3,486 addresses across six chains (Tron, BSC, Ethereum, Solana, Polygon, Base), led by Tron at 39%. Coverage now extends to Morocco, Kenya, Ghana and South Africa. Neobanks is now a nine-venue category. Raenest runs consumer and business flows on shared infrastructure (not separable), so its retail figure is the size band and may include business flow; its ~$4.68M total observed flow is shown as a shared-rail segment excluded from retail totals. It is deposit-only onchain — settlement runs through external rails it is a small minority of, so no retail withdrawals are shown.
CoverageNeobanks
21 Aug 2026New venue
Daya Pro added — Nigerian crypto exchange
Daya Pro, a Nigerian B2C cryptocurrency exchange, is now mapped and live across the market, category and venue pages: $11.70M lifetime retail deposits from 161 deposit addresses across nine chains (BSC, Ethereum, Tron, Polygon, Arbitrum, Aptos, Solana, Optimism, Base), led by BSC at 31%. It is AfriFlux's first venue tracked on Aptos, and Exchanges is now a four-venue category. Daya Pro owns its deposit and settlement rails on every chain; volume is concentrated (the top 10 addresses carry 64%), reflecting its Pro-trader base, and it is classified retail-only so the full deposit volume is reported as retail.
CoverageExchanges
19 Aug 2026Coverage
Timon — BUSD and PYUSD added to coverage
Timon now accepts BUSD (on BSC and Polygon) and PYUSD (on Solana and Ethereum); both are ingested and tracked, and count toward Timon's stablecoin volume as deposits arrive. PYUSD is broken out individually in the public token split (see the composition update); BUSD is grouped under 'Others' and remains in volume totals. Volumes on both are nominal so far — USDT and USDC still carry effectively all of Timon's flow.
CoverageNeobanks
19 Aug 2026Coverage
cUSD and PYUSD broken out of 'Others'
The public token split now breaks out cUSD (Celo) and PYUSD (Ethereum, Solana) as first-class stablecoins rather than grouping them under 'Others'. cUSD is the material addition — $1.48M of tracked deposit volume across ~120,000 transfers, concentrated on Celo-heavy venues; PYUSD is newly-appearing and small. USDT and USDC remain overwhelmingly dominant. Volume totals are unchanged — these tokens were already counted; only their presentation moves from 'Others' to their own labels.
CoverageComposition
18 Aug 2026New venue
Accrue added — Nigerian cross-border money app
Accrue, a Nigerian-founded cross-border payments and dollar-savings app, is now mapped and live across the market, category and venue pages: $13.35M lifetime retail deposits from ~4,500 wallets across six chains (Celo, BSC, Ethereum, Polygon, Base, Optimism), led by Celo at 47%. Neobanks is now an eight-venue category. Accrue runs consumer and business flows on identical rails (not separable), so its retail figure is the size band and may include business flow, and its $13.8M total flow is shown as a shared-rail B2B segment excluded from retail totals.
CoverageNeobanks
17 Aug 2026Methodology
Retail classification is now applied per venue
The retail classification engine sorts deposit addresses by size and frequency — Tier 1, Tier 2 and Retail Whale count as retail; Large Whale and Business are excluded. It exists to strip business, institutional and liquidity flow from venues that mix it with consumer flow. It has never been able to identify a business directly; it infers one from deposit behaviour.
That inference only makes sense where both kinds of flow share the same rails. On a venue that serves individuals only there is no business flow to remove, and applying the filter was excluding real retail — a large depositor at a consumer offramp is still a person. The engine is therefore now applied per venue rather than universally.
Ten venues are classified as serving individuals only and are no longer filtered: Spenda, Roqqu, Breet, Azza, Timon, Paj Cash, Bayse Markets, Evolution, SFx and Payrit. Five of them restate upward — Spenda $69.47M to $70.16M, Roqqu $57.39M to $58.22M, Breet $23.76M to $24.70M, Azza $5.02M to $5.19M, Bayse Markets $1.21M to $1.23M. The other five had nothing above the band and are unchanged. Total effect on tracked retail volume: +$2.65M, or +0.4%.
The engine still runs on Quidax, Busha, Chipper Cash, Grey Finance and Onboard Global. Every verification filter is unchanged for every venue — verified deposit addresses only, named exchange and OTC entities excluded, each venue’s own operational wallets excluded. Only the size band is now conditional.
The retail-ratio figure on this page is now computed across the venues the engine is applied to. Averaging in venues with nothing removed understated how selective the filter actually is.
All metricsCoverage
17 Aug 2026Methodology
Shared-rail venues — what we do not split
Grey Finance and Onboard Global run business and personal accounts on identical infrastructure: the same address type, the same movement pattern, no separate settlement path. A business account there is indistinguishable from a personal one, and the size band cannot separate them — a business often deposits retail-sized amounts and sits inside the band, while an individual moving a large sum sits above it.
We are not estimating a split. Their retail figure is the size band and may include business flow, and it is labelled as such. On the venues directory these venues also carry a total-flow figure, which includes retail and is shown only for context — it is excluded from any business-volume total or ranking, because a total is not a measurement.
Quidax, Busha and Chipper Cash are marked split under review: a business book is being measured for each and their retail figures may be restated when that work lands. No business-volume figure is published for any venue yet.
CoverageExchangesNeobanks
17 Aug 2026Correction
Paj Cash figure on this page corrected
The Venue-Level Retail Classification table on this page reported $1.10M lifetime volume for Paj Cash while the market, category and venue pages reported $2.60M. The market pages were correct.
Paj Cash uses a two-rail inbound structure that is resolved in a dedicated model rather than by the standard deposit-address join. That model was corrected upstream, but the shared daily aggregate every public figure rolls up from is incremental — it recomputes a three-day window each run, so partitions older than that window kept the superseded values. Rebuilding the full history brought them forward.
The corrected figure is $2.597M. No other venue was affected, and no other page carried the stale number.
Offramps