bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 3쪽 중 2쪽
virtual machines and,
through the exchange module, route trading into a common orderbook.
9 INJECTIVE WHITEPAPER OCTOBER 2026
4.3 Transaction admission and proposal construction
Before execution, Injective groups pending transactions into four ordered lanes: Oracle, Gov-
ernance, Exchange, and Default. Oracle updates receive the highest proposal priority, followed by
exchange-administrator transactions in the Governance Lane. Exchange-only transactions enter the
Exchange Lane, and other traffic enters the Default Lane. Validators apply lane rules to construct
and validate proposals.2
Lane routing preserves account sequence so a later transaction cannot bypass an earlier trans-
action from the same signer. Within the Exchange Lane, priority accounts for fee-discount tier and
transaction type, with liquidation-only transactions receiving the highest priority.2
Transaction admission also passes through the txfees module. It enforces a maximum gas re-
quest per transaction and minimum gas-price rules, including a separate threshold for high-gas
transactions. Governance-controlled parameters can enable an EIP-1559-style adaptive fee market,
define the target share of block gas, bound base-fee movement, and govern the rechecking of pending
transactions as the base fee changes. When adaptive mode is enabled, letbh be the base fee for
blockh,g h the gas used,g∗ the target gas, andαthe per-block adjustment rate. Before configured
bounds and periodic reset are applied, the update is:2,5,6
bh+1 =b h
(︃
1 +αgh −g ∗
g∗
)︃
(3)
Equation 3. Conditional adaptive base-fee update
Utilization above target raises the base fee; utilization below target lowers it. Lanes and fee
admission therefore solve different parts of the transaction pipeline: lanes determine functional
ordering within proposals, while txfees governs the economic and gas conditions under which trans-
actions enter and remain in the mempool.2,5,6
4.4 Block execution lifecycle
After proposal construction, every validator executes the same ordered transaction sequence
against the same pre-block state. Native module hooks perform scheduled work before and after
transaction execution. The exchange module’s EndBlock sequence is especially important because it
defines how orders submitted throughout a block become a single deterministic market transition.2
At BeginBlock, the exchange can process scheduled funding updates, expiry settlements, trading
rewards, and fee-accounting work. During transaction execution, messages create and cancel orders,
update oracle state, move balances, or call contracts. At EndBlock, the exchange determines fee
tiers, processes market orders against the beginning-of-block resting book, persists those results,
processes matching limit orders, persists fills and new resting orders, updates funding and rewards,
and emits the resulting events.2
10 INJECTIVE WHITEPAPER OCTOBER 2026
Deterministic block and market execution lifecycle
1 BeginBlock
Funding updates
Scheduled
settlements
Expiry processing
Rewards and fee
work
2 Transaction
execution
Orders and
cancellations
Oracle reports
Transfers and
contracts
State checks
3 Exchange EndBlock
Fee tiers
Market orders vs. opening
book
Persist executions
Limit-order batch clearing
Funding, rewards, parameters
4 Commit
State root
Precommit quorum
Deterministic
finality
Deterministic combine
Parallel work may improve execution, but every validator must derive the same final state before commitment
Figure 2: Deterministic block and exchange lifecycle
Some market groups are processed in parallel using a split, apply, and combine pattern. This
concurrency is an implementation strategy rather than a relaxation of determinism: markets can be
divided into independent work sets where their orderbooks and state updates do not conflict, then
recombined in a defined order before commitment.2
Scheduledexecutionisoneexampleofhowmodulelogicparticipatesinthislifecycle. TheWasmx
module allows governance-registered WASM contracts to receive abegin_blockersudo call under a
specified code identifier and gas configuration. Registered contracts may be pinned in memory and
can be deactivated or removed under defined conditions. Wasmx therefore supplies deterministic,
protocol-scheduled automation without introducing another virtual machine or relying on an off-
chain keeper, and its effects remain subject to the same execution and consensus rules as every
other state transition.2
5 Financial primitives as protocol modules
The financial layer is implemented as a set of modules within the state machine, which has three
direct consequences. Module state is available through common interfaces, execution is synchronized
with the block lifecycle, and protocol upgrades can evolve shared infrastructure without requiring
each application to deploy a replacement market stack of its own.
11 INJECTIVE WHITEPAPER OCTOBER 2026
5.1 Exchange state, subaccounts, and shared markets
The exchange module maintains spot and derivative markets, orderbooks, deposits, positions,
market parameters, fee settings, and settlement state. Applications using native, EVM, or WASM
interfaces can submit orders into the same module-managed markets, so each market has one pro-
tocol orderbook rather than an application-specific book. Liquidity is not produced by the module
itself. It is supplied by participants that submit orders directly or through smart-contract vaults and
other pooled strategies. These strategies may aggregate capital, but execution and price discovery
still occur through the shared orderbook rather than an automated market-maker curve.2
Trading balances are held in subaccounts that distinguish total deposits from amounts available
for new actions. Funds committed to resting orders, derivative margin, or other exchange obligations
remain part of the subaccount’s total balance but cannot be spent again while reserved. This
separation allows the protocol to prevent double use of capital and to maintain position margin
independently from a user’s unrestricted bank balance.
Derivative markets also carry initial-margin and maintenance-margin parameters. The former
constrainstheriskthatcanbeopened, whilethelatterdetermineswhenanexistingpositionbecomes
eligible for liquidation. Because the checks are performed against exchange-module state, every
interface admitting an order into a market inherits the same collateral and risk rules; Section 5.4
formalizes the resulting position-equity and maintenance tests.
5.2 Frequent batch auctions
Continuous matching creates a structural speed race because orders are processed one at a
time, allowing small latency advantages to determine who reaches a stale quote first. On a public
chain, transaction ordering can intensify that race by giving validators or sophisticated participants
an opportunity to insert or reorder activity. Frequent Batch Auctions (FBAs) were proposed in
market-design research as a response to this structural problem: discrete intervals and uniform-
price clearing replace competition on speed with competition on price. Injective implements this
model at the exchange-module layer, with each block serving as an auction interval.2,7
The design has three defining properties. First, each block acts as a discrete auction interval
rather than a continuous sequence of independent matches. Second, matching orders execute at
a uniform clearing price within the interval. Third, new orders are sealed at the orderbook level:
they are not published into the resting book until the interval closes and the EndBlock batch is
executed. In this context, sealed describes when orders enter the public orderbook rather than a
separate cryptographic commit-reveal scheme.2,7
These properties change the basis of competition within the matching process. Because orders
in the same batch are cleared together at a uniform price, being marginally earlier in the block
does not produce a superior serial execution price. The ordinary FBA path is therefore resistant to
12 INJECTIVE WHITEPAPER OCTOBER 2026
front-running, stale-quote sniping, and a major class of transaction-ordering maximal extractable
value (MEV). By reducing the value of latency advantages and the risk of adverse execution against
stale quotes, the design can also support tighter spreads and greater displayed depth. Experimental
comparisons of FBA and continuous double auction (CDA) markets report less predatory trading,
lower investment in low-latency technology, lower transaction costs, and lower volatility in spreads
and liquidity under FBA.7,8
Within the documented lifecycle, market orders are processed first against the resting orderbook
as it existed at the beginning of the block. Matching limit orders are then combined with that book
and cleared while the spread remains crossed, after which fills, cancellations, and newly resting
orders are persisted as part of the block transition.2
Crossed book before clearing
Uncrossed book after clearing
quantity
price
p
c
quantity
price
p
c
ENDBLOCK
BATCH
all matches
execute at one
price
bids
asks
matched quantity
Figure 3: Uniform-price batch clearing
Letp b andp s denote the prices of the last matched buy and sell orders. For a derivative market,
letp m denote the mark price. The documented clearing rule can be represented as follows, withpcross
denoting the price of the last crossing order when the match crosses the spread in that direction.2
pc =
⎧
⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎩
pcross if the final match crosses the spread,
pm ifp s ≤p m ≤p b,
pb +p s
2 otherwise.
(4)
Equation 4. Batch clearing-price rule
The lifecycle distinguishes market orders, matching limit orders, and atomic market orders.
Market orders use the beginning-of-block resting book, while the limit-order phase applies the batch
clearing rule so that price and quantity determine execution without introducing a continuous
nanosecond priority race. Atomic market orders form a deliberate exception for smart-contract
13 INJECTIVE WHITEPAPER OCTOBER 2026
workflows that require immediate execution and therefore pay an additional fee rather than entering
the ordinary batch.2
5.3 Oracle accounting
Derivative risk and settlement depend on external prices. For each price feed, the oracle module
stores the latest normalized pricep(t), cumulative priceC(t), and timestampt. The cumulative
series permits a time-weighted average price (TWAP) over an interval without storing every inter-
mediate observation.2,9
TW AP(t1, t2) = C(t 2)−C(t 1)t2 −t 1 (5)
Equation 5. Time-weighted average from cumulative price state
Different providers follow different verification paths. In the off-chain reporting (OCR) path,N
oracle nodes exchange signed reports over a peer-to-peer network before a designated transmitter
submits an aggregate on chain. The module verifies the signatures and configuration, records the
median observation, and uses epoch and round identifiers to reject stale or duplicated reports, while
reward accounting can compensate contributors.2,10
TWAPs reduce sensitivity to individual observations, while median aggregation reduces sensitiv-
ity to individual reporters. Together with update cadence and provider diversity, these mechanisms
convert external market data into a deterministic state that the exchange module can use for fund-
ing, margin, and settlement.
5.4 Funding, margin, and liquidation
Perpetual markets use funding to connect the contract price to its reference market. The ex-
change computes a time-weighted premium from the relationship between execution prices and the
mark price, incorporates the configured interest component, applies a cap, and converts the result
into an hourly rate. The resulting payment transfers value between long and short positions without
changing aggregate open interest.2,11
For a simplified isolated linear position, letsequal+1for a long and−1for a short,Qbe
absolute quantity,p e be entry price,p m be mark price,Mbe posted margin, andFbe accrued
funding owed by the position. Equity before closing fees can be written as:
E(pm) =M+sQ(p m −p e)−F(6)
Equation 6. Simplified linear-position equity
14 INJECTIVE WHITEPAPER OCTOBER 2026
Ifr maint is the maintenance-margin ratio, a compact representation of the maintenance test is:
E(pm)< rmaintQpm (7)
Equation 7. Simplified maintenance-margin condition
Equations 6 and 7 provide a compact linear mark-to-market model rather than a complete repro-
duction of the implementation. The onchain check additionally incorporates market configuration,
funding state, order state, and applicable fees.
When a position breaches the protocol’s liquidation conditions, the exchange creates a reduce-
only market order that executes before ordinary matching. If closing the position leaves a negative
payout, the market’s insurance fund can cover the shortfall; if value remains after the forced reduc-
tion, the surplus is returned according to the module’s liquidation accounting.2
5.5 Insurance funds
Injective’s insurance module manages pools of capital that can absorb specified liquidation short-
falls in derivative markets. Underwriters contribute assets and receive share tokens representing a
proportional claim on the relevant fund. Redemptions follow the module’s notice process, reduc-
ing the risk that coverage disappears immediately during market stress. The module tracks fund
balances, share supply, and pending redemptions.2
Insurance capital forms the final layer of the market’s loss-management sequence: posted margin
absorbs loss first, forced reduction closes exposure, and available insurance capital may cover a
residual deficit under the module’s rules.
6 Permissioned assets and real-world asset infrastructure
6.1 TokenFactory
TokenFactory allows an account to create a native denomination in the namespacefactory/
{creator}/{subdenom}. The namespace administrator can mint, burn, transfer administrative
control, and update metadata subject to module rules. The resulting asset is a bank token, so it
can participate in the same canonical state used by native modules and MTS interfaces.2
TokenFactory answers how an asset is created and who controls its supply. It does not by itself
answer who may hold or transfer that asset. The Permissions module adds that policy layer.
15 INJECTIVE WHITEPAPER OCTOBER 2026
6.2 Role-based asset controls
The Permissions module applies role-based controls to TokenFactory assets, including minting,
receiving, burning, and sending tokens. An action is accepted only when the address’s combined
role permissions allow it, namespace policy permits it, no overriding blacklist applies, and any
configured receive hook succeeds. Receive hooks can apply deterministic contract logic using the
sender, recipient, action, and amount.2,12
The namespace can be sealed to make specified administrative settings irreversible. Issuers can
also opt into governance override, creating a public recovery path. The resulting roles, action masks,
hooks, and administrative configuration are represented directly in protocol state.
6.3 Real-world assets and iAssets
Injective treats real-world asset infrastructure as a protocol capability rather than a single ap-
plication category. TokenFactory and Permissions support native issuance with configurable asset
policy. The exchange and oracle modules support a second route: markets that provide price
exposure to an external instrument through a cash-settled derivative.
These routes produce economically different instruments. A native permissionless token is a
bearer-style onchain asset with no protocol-level holder restrictions. A permissioned token can
represent a claim administered by an issuer and constrained by role policy. An iAsset is a cash-
settled derivative that tracks the price of an external asset without representing ownership of the
underlying asset.2,13
Table 1: Economically distinct asset models
Model Onchain object Primary authority What the holder owns
Permissionless
native token
Bank denomination Token administrator, if
retained
The token itself; any off-chain
claim depends on separate terms
Permissioned
token
TokenFactory
denomination plus
Permissions policy
Issuer roles, hooks, and
governance path
A token governed by the
issuer-defined claim and
redemption framework
iAsset Cash-settled
derivative position
Exchange rules, oracle,
market governance
Price exposure without title to
the referenced asset
iAssets solve a market-access problem rather than an asset-registry problem. They combine sta-
blecoin margin and oracle-based pricing within exchange-module markets to create programmable
price exposure without pre-funding or wrapping the referenced asset. Participants supply liquid-
ity by quoting into those markets, while applications can expose the same module-managed book
through different interfaces. The distinction between issued ownership claims and derivative expo-
sure lets builders select the asset model appropriate to the market they are creating.
16 INJECTIVE WHITEPAPER OCTOBER 2026
7 Interoperability and settlement assets
7.1 External assets and settlement paths
Injective connects its financial state to assets and messages from external networks through IBC,
Peggy, and Circle’s CCTP. These paths do not share a single trust model: each verifies a different
external event and creates an asset or message whose provenance remains identifiable after it enters
Injective’s state machine.2,14,15
IBC uses onchain light clients and packet commitments to verify state transitions between com-
patible chains. Relayers transport packets, but they do not determine validity, which instead derives
from the connected chains, client state, submitted proofs, and timeout rules.14
Peggy is Injective’s validator-secured, bidirectional bridge for ERC-20 assets on Ethereum and
has historically served assets including the legacy ERC-20 representation of INJ and Ethereum-
origin stablecoins such as USDT. When an Ethereum-origin asset moves to Injective, the ERC-20
is locked in the Peggy contract and Peggo orchestrators submit deposit claims; after more than
two thirds of Injective voting power attests to the event, the Peggy module mints a representative
denomination beginning with peggy. A withdrawal reverses the process by burning the Injective
representation and authorizing release of the ERC-20 from the Ethereum contract.2
Circle’s Cross-Chain Transfer Protocol, or CCTP, uses a different model for native USDC. A
source-domain burn is observed and attested by Circle’s service, and that attestation authorizes
native USDC minting on the destination. CCTP V2 support on Injective also enables optional
fast-transfer and hook patterns.15
The EVM and native execution environments on Injective are different. They share one chain
and one finalized state, so moving a token between their interfaces through MTS is not a cross-chain
bridge event.
Table 2: Settlement and verification models
Path What is verified Primary dependency Result on Injective
MTS Internal state transition Core modules and
mappings
One canonical balance
IBC Counterparty state through
a light client
Both chains, clients,
relayers
Voucher or packet state
CCTP Circle attestation of a
USDC burn
Circle issuer and attester Native USDC mint
Peggy Ethereum deposit event or
Injective withdrawal batch
Ethereum contract, Peggo,
and more than two thirds
of Injective voting power
peggy denomination or
ERC-20 release
17 INJECTIVE WHITEPAPER OCTOBER 2026
7.2 Settlement composition
Cross-chain composition can be understood as a sequence of finality domains. After a source
chain finalizes an event, a light client, attester, or bridge mechanism validates that event under its
own security model; only then does Injective accept the corresponding message and update its state
so that an application can use the resulting asset in a market or contract.
This architecture preserves provenance across different settlement paths. Native bank assets,
IBC denominations, bridge-originated assets, and issuer-minted assets can be r프로젝트로 돌아가기