vault.base
Documentation for eth_defi.vault.base Python module.
Generic Vault adapter base classes.
Create unified interface across different vault protocols and their investment flows
Helps to create automated trading agents against any vault easily
Handle both trading (asset management role) and investor management (deposits/redemptions)
See
VaultBaseto get startedSee
RawVaultPriceRowfor the raw scanner parquet row schema andCleanedVaultPriceRowfor the enriched cleaned format
Functions
|
Read back a parquet file after writing and verify its integrity. |
Classes
Result of verifying a parquet file after writing. |
|
Schema for a single row in the uncleaned vault price parquet file. |
|
Describe assets vault can manage. |
|
Base class for vault protocol adapters. |
|
Manage deposit/redemption events. |
|
Vault share price and fee structure at the point of time. |
|
Support reading historical vault share prices. |
|
Vault-protocol specific intormation about the vault. |
|
Track assets and balances in a vault. |
|
VaultReadCondition() |
|
Unique id for a vault. |
Exceptions
Raised when a parquet file fails post-write verification. |
- exception ParquetVerificationError
Bases:
ExceptionRaised when a parquet file fails post-write verification.
This is a hard failure — the operator must investigate and restore from backup. Never catch and swallow this exception.
- __init__(*args, **kwargs)
- __new__(**kwargs)
- add_note(note, /)
Add a note to the exception
- with_traceback(tb, /)
Set self.__traceback__ to tb and return self.
- class ParquetVerificationResult
Bases:
objectResult of verifying a parquet file after writing.
Returned by
verify_parquet_file()on success.
- class RawVaultPriceRow
Bases:
TypedDictSchema for a single row in the uncleaned vault price parquet file.
This is the format produced by
VaultHistoricalRead.export()and written tovault-prices-1h.parquetby the historical scanner (scan_historical_prices_to_parquet()).The canonical columns are defined by
VaultHistoricalRead.to_pyarrow_schema(). Native protocol merges (Hyperliquid, GRVT, Lighter, Hibachi) may add extra columns (e.g.account_pnl,leader_fraction) that are preserved across schema migrations but are not part of this TypedDict.See
CleanedVaultPriceRowineth_defi.research.wrangle_vault_pricesfor the enriched schema produced by the cleaning pipeline.- __init__(*args, **kwargs)
- __new__(**kwargs)
- clear()
Remove all items from the dict.
- copy()
Return a shallow copy of the dict.
- fromkeys(value=None, /)
Create a new dictionary with keys from iterable and values set to value.
- get(key, default=None, /)
Return the value for key if key is in the dictionary, else default.
- items()
Return a set-like object providing a view on the dict’s items.
- keys()
Return a set-like object providing a view on the dict’s keys.
- pop(k[, d]) v, remove specified key and return the corresponding value.
If the key is not found, return the default if given; otherwise, raise a KeyError.
- popitem()
Remove and return a (key, value) pair as a 2-tuple.
Pairs are returned in LIFO (last-in, first-out) order. Raises KeyError if the dict is empty.
- setdefault(key, default=None, /)
Insert key with a value of default if key is not in the dictionary.
Return the value for key if key is in the dictionary, else default.
- update([E, ]**F) None. Update D from mapping/iterable E and F.
If E is present and has a .keys() method, then does: for k in E.keys(): D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
- values()
Return an object providing a view on the dict’s values.
- class TradingUniverse
Bases:
objectDescribe assets vault can manage.
Because of brainrotten and awful ERC-20 token standard, the vault does not know what tokens it owns and this needs to be specific offchain
- class VaultBase
Bases:
abc.ABCBase class for vault protocol adapters.
Allows automated interaction with different vault protocols.
Contains various abstract methods that the implementation class must override
Supported protocols include
Lagoon Finance:
eth_defi.lagoon.vault.LagoonVaultVelvet Capital:
eth_defi.velvet.vault.VelvetVault
Code exists, but does not confirm the interface yet:
Enzyme Finance:
eth_defi.enzyme.vault.Vault
Vault covered functionality
Fetching the current balances, deposits or redemptions
Either using naive polling approach with
fetch_portfolio()Listen to vault events for deposits and redemptions using
get_flow_manager()
- Get vault information with
fetch_info() No standardised data structures or functions yet
- Get vault information with
- Build a swap through a vault
No standardised data structure yet
- Update vault position valuations
No standardised data structure yet
Integration check list
Integration tests needed for:
☑️ read vault core info
☑️ read vault investors
☑️ read vault share price
☑️ read vault share token
☑️ read all positions
☑️ read NAV
☑️ read pending redemptions to know how much USDC we will need for the next settlement cycles
☑️ deposit integration test
☑️ redemption integration
☑️ swap integration test
☑️ re-valuation integration test
☑️ only asset manager allowed to swap negative test
☑️ only valuation commitee allowed to update vault valuations (if applicable)
☑️ can redeem if enough USDC to settle
☑️ cannot redeem not enough USDC to settle
For code examples see tests/lagoon and tests/velvet on the Github repository.
- Parameters
token_cache –
Token cache for vault tokens.
Allows to pass
eth_defi.token.TokenDiskCacheto speed up operations.require_denomination_token –
If
True, accessingdenomination_tokenwill raiseRuntimeErrorwhen the on-chain lookup returnsNone.Use for deployment scripts and operational contexts where a missing denomination token is always a hard error.
- __init__(token_cache=None, require_denomination_token=False)
- Parameters
token_cache (Optional[dict]) –
Token cache for vault tokens.
Allows to pass
eth_defi.token.TokenDiskCacheto speed up operations.require_denomination_token (bool) –
If
True, accessingdenomination_tokenwill raiseRuntimeErrorwhen the on-chain lookup returnsNone.Use for deployment scripts and operational contexts where a missing denomination token is always a hard error.
- abstract property address: eth_typing.evm.HexAddress
Vault contract address.
Often vault protocols need multiple contracts per vault, so what this function returns depends on the protocol
- property denomination_token: Optional[eth_defi.token.TokenDetails]
Get the token which denominates the vault valuation
Used in deposits and redemptions
Used in NAV calculation
Used in profit benchmarks
Usually USDC
- Returns
Token wrapper instance.
Maybe None for broken vaults like https://arbiscan.io/address/0x9d0fbc852deccb7dcdd6cb224fa7561efda74411#code
Note
Noneresults are not cached — the next access will retry the on-chain call. This avoids permanently caching a transient RPC failure.
- property deposit_manager: eth_defi.vault.deposit_redeem.VaultDepositManager
Deposit manager assocaited with this vault
- property description: Optional[str]
Human-readable vault strategy description.
Fetched from protocol-specific offchain sources (e.g. Euler GitHub labels, Lagoon web app API)
Returns None if the protocol does not provide descriptions or the vault is not in the metadata source
Override in subclasses that support offchain metadata
- fetch_available_liquidity(block_identifier='latest')
Get the amount of denomination token available for immediate withdrawal.
Only applicable to lending protocol vaults (IPOR, Euler, Morpho, Gearbox, etc.). Non-lending protocols should leave this method unimplemented.
Note: maxRedeem(address(0)) does NOT work as a proxy for available liquidity because it requires a specific address that has already deposited shares. For address(0), balanceOf is always 0, so maxRedeem returns 0 regardless of actual liquidity.
- Parameters
block_identifier (Union[Literal['latest', 'earliest', 'pending', 'safe', 'finalized'], eth_typing.evm.BlockNumber, eth_typing.evm.Hash32, eth_typing.encoding.HexStr, int]) – Block to query. Defaults to “latest”.
- Raises
NotImplementedError – For non-lending protocol vaults.
- Returns
Amount in denomination token units (human-readable Decimal).
- Return type
- abstract fetch_denomination_token()
Read denomination token from onchain.
Use
denomination_token()for cached access.- Return type
- fetch_denomination_token_address()
Get the address for the denomination token.
Triggers RCP call
- Return type
- fetch_deposit_closed_reason()
Get human-readable reason why deposits are closed.
Override in protocol-specific subclasses
Default behaviour: assume deposits are always open (return None)
- fetch_deposit_next_open()
Get when deposits will next be open.
For epoch-based vaults (Ostium, D2), return calculated window open time
For non-epoch vaults (Plutus, IPOR, Morpho), return None
Override in protocol-specific subclasses
- Returns
Naive UTC datetime when deposits will next be available, or None if:
Deposits are currently open
Timing is unpredictable (manually controlled)
Protocol does not support timing information
- Return type
- abstract fetch_info()
Read vault parameters from the chain.
Use
info()property for cached access.- Return type
Fetch the most recent onchain NAV value.
- Returns
Vault NAV, denominated in
denomination_token()- Return type
- abstract fetch_portfolio(universe, block_identifier=None)
Read the current token balances of a vault.
SHould be supported by all implementations
- Parameters
universe (eth_defi.vault.base.TradingUniverse) –
block_identifier (Optional[Union[Literal['latest', 'earliest', 'pending', 'safe', 'finalized'], eth_typing.evm.BlockNumber, eth_typing.evm.Hash32, eth_typing.encoding.HexStr, int]]) –
- Return type
- fetch_redemption_closed_reason()
Get human-readable reason why redemptions are closed.
Override in protocol-specific subclasses
Default behaviour: assume redemptions are always open (return None)
- fetch_redemption_next_open()
Get when withdrawals/redemptions will next be open.
For epoch-based vaults (Ostium, D2), return calculated window open time
For non-epoch vaults (Plutus, IPOR, Morpho), return None
Override in protocol-specific subclasses
- Returns
Naive UTC datetime when withdrawals will next be available, or None if:
Withdrawals are currently open
Timing is unpredictable (manually controlled)
Protocol does not support timing information
- Return type
Read share token details onchain.
Use
share_token()for cached access.- Return type
- fetch_utilisation_percent(block_identifier='latest')
Get the percentage of assets currently lent out.
Only applicable to lending protocol vaults (IPOR, Euler, Morpho, Gearbox, etc.). Non-lending protocols should leave this method unimplemented.
- Parameters
block_identifier (Union[Literal['latest', 'earliest', 'pending', 'safe', 'finalized'], eth_typing.evm.BlockNumber, eth_typing.evm.Hash32, eth_typing.encoding.HexStr, int]) – Block to query. Defaults to “latest”.
- Raises
NotImplementedError – For non-lending protocol vaults.
- Returns
Utilisation as float between 0.0 and 1.0 (0% to 100%).
- Return type
- property flow_manager: eth_defi.vault.base.VaultFlowManager
Flow manager associated with this vault
- get_deposit_fee(block_identifier)
Deposit fee is set to zero by default as vaults usually do not have deposit fees.
Internal: Use
get_fee_data().
- abstract get_deposit_manager()
Get deposit manager to deposit/redeem from the vault.
- get_estimated_lock_up()
What is the estimated lock-up period for this vault.
- Returns
None if not know
- Return type
- get_fee_data()
Get fee data structure for this vault.
- Raises
ValueError – In the case of broken or unimplemented fee reading methods in the smart contract
- Return type
- get_fee_mode()
Get how this vault accounts its fees.
- Return type
- get_flags()
Get various vault state flags from the smart contract.
Override to add status flags
Also add flags from our manual flag list in
eth_defi.vault.flag
- Returns
Flag set.
Do not modify in place.
- Return type
- abstract get_flow_manager()
Get flow manager to read indiviaul settle events.
Only supported if
has_block_range_event_support()is True
- Return type
- abstract get_historical_reader(stateful)
Get share price reader to fetch historical returns.
- Parameters
stateful (bool) – If True, use a stateful reading strategy.
- Returns
None if unsupported
- Return type
- get_link(referral=None)
Get a link to the vault dashboard on its native site.
By default, give RouteScan link
- get_management_fee(block_identifier)
Get the current management fee as a percent.
Internal: Use
get_fee_data().
- get_notes()
Get a human readable message if we know somethign special is going on with this vault.
- get_performance_fee(block_identifier)
Get the current performance fee as a percent.
Internal: Use
get_fee_data().
- get_risk()
Get risk profile of this vault.
- Return type
- get_withdraw_fee(block_identifier)
Withdraw fee is set to zero by default as vaults usually do not have withdraw fees.
Internal: Use
get_fee_data().
- abstract has_block_range_event_support()
Does this vault support block range-based event queries for deposits and redemptions.
If not we use chain balance polling-based approach
- Return type
- has_custom_fees()
Does this vault have custom fee structure reading methods.
Causes risk in the vault comparison.
E.g.
Withdraw fee
Deposit fee
- Returns
True if custom fee reading methods are implemented
- Return type
- abstract has_deposit_distribution_to_all_positions()
Deposits go automatically to all open positions.
Deposits do not land into the vault as cash
Instead, smart contracts automatically increase all open positions
The behaviour of Velvet Capital
- Return type
- property info: eth_defi.vault.base.VaultInfo
Get info dictionary related to this vault deployment.
Get cached data on the various vault parameters
- Returns
Vault protocol specific information dictionary
ERC-20 that presents vault shares.
User gets shares on deposit and burns them on redemption
- property short_description: Optional[str]
One-liner vault summary.
Shorter version of
description()suitable for listings and tablesReturns None if not available
Override in subclasses that support offchain metadata
- class VaultFlowManager
Bases:
abc.ABCManage deposit/redemption events.
For some vault structures, we need to know how much redemptions there are in the queue, so we can rebalance to have enough cash
Create a replay of flow events that happened for a vault within a specific block range
Not implemented yet
- abstract fetch_pending_deposit(block_identifier)
Get how much users want to redeem from the vault.
- abstract fetch_pending_deposit_events(range)
Read incoming pending deposits.
- Parameters
range (Tuple[eth_typing.evm.BlockNumber, eth_typing.evm.BlockNumber]) –
- Return type
None
- abstract fetch_pending_redemption(block_identifier)
Get how much users want to redeem from the vault.
- Parameters
block_identifier (Union[Literal['latest', 'earliest', 'pending', 'safe', 'finalized'], eth_typing.evm.BlockNumber, eth_typing.evm.Hash32, eth_typing.encoding.HexStr, int]) – Block number
- Returns
Number of share tokens the users want to redeem from the vault.
Shares must be valued separately.
- Return type
- abstract fetch_pending_redemption_event(range)
Read outgoing pending withdraws.
- Parameters
range (Tuple[eth_typing.evm.BlockNumber, eth_typing.evm.BlockNumber]) –
- Return type
None
- abstract fetch_processed_deposit_event(range)
Read incoming pending deposits.
- Parameters
range (Tuple[eth_typing.evm.BlockNumber, eth_typing.evm.BlockNumber]) –
- Return type
None
- abstract fetch_processed_redemption_event(vault, range)
Read outgoing pending withdraws.
- Parameters
vault (eth_defi.vault.base.VaultSpec) –
range (Tuple[eth_typing.evm.BlockNumber, eth_typing.evm.BlockNumber]) –
- Return type
None
- class VaultHistoricalRead
Bases:
objectVault share price and fee structure at the point of time.
- __init__(vault, block_number, timestamp, share_price, total_assets, total_supply, performance_fee, management_fee, errors, vault_poll_frequency=None, max_deposit=None, max_redeem=None, deposits_open=None, redemption_open=None, trading=None, available_liquidity=None, utilisation=None)
- Parameters
vault (eth_defi.vault.base.VaultBase) –
block_number (int) –
timestamp (datetime.datetime) –
share_price (Optional[decimal.Decimal]) –
total_assets (Optional[decimal.Decimal]) –
total_supply (Optional[decimal.Decimal]) –
max_deposit (Optional[decimal.Decimal]) –
max_redeem (Optional[decimal.Decimal]) –
available_liquidity (Optional[decimal.Decimal]) –
- Return type
None
- export()
Convert historical read for a Parquet/DataFrame export.
The returned dict conforms to
RawVaultPriceRow.- Return type
- is_almost_equal(other, epsilon=0.001)
Check if the read statistics match.
Throttle with epsilon relative difference to get rid of small increment rows
- Parameters
epsilon (float) – Write changes with 10 BPS granularity
other (Optional[eth_defi.vault.base.VaultHistoricalRead]) –
- Return type
- static migrate_parquet_schema(existing_table)
Migrate an existing Parquet table to the current schema.
When new columns are added to
to_pyarrow_schema(), existing parquet files still have the old schema. This function adds missing columns as null arrays so incremental scans can write new data without losing columns.Non-canonical columns (e.g.
account_pnl,leader_fractionadded by native protocol merges) are preserved so that the EVM scanner does not destroy data written by Hyperliquid/GRVT/Lighter.- Parameters
existing_table (pyarrow.Table) – Table read from an older parquet file.
- Returns
Table with all canonical columns present (missing ones filled with nulls), legacy columns removed, extra columns preserved.
- Return type
pyarrow.Table
- classmethod to_pyarrow_schema()
Get parquet schema for writing this data.
Write multiple chains, multiple vaults, to a single Parquet file
Column semantics are documented in
RawVaultPriceRow
- Return type
pyarrow.Schema
- static write_uncleaned_parquet(df, path, compression='zstd')
Write a DataFrame to the uncleaned parquet using proper PyArrow types.
Native protocol merge functions (Hyperliquid, GRVT, Lighter) must use this instead of
pandas.DataFrame.to_parquet()to avoid type promotion (e.g.timestamp[ms]→timestamp[us]) that breaksmigrate_parquet_schema()on the next EVM scan run.Columns present in the canonical schema are cast to their canonical types. Extra columns (from native protocols) are kept with their pandas-inferred types. No pandas index is written.
- Parameters
df (pandas.DataFrame) – Combined DataFrame to write.
path (pathlib.Path) – Output parquet file path.
compression (str) – Parquet compression codec.
- class VaultHistoricalReader
Bases:
abc.ABCSupport reading historical vault share prices.
Allows to construct historical returns
- __init__(vault)
- Parameters
vault (eth_defi.vault.base.VaultBase) –
- abstract construct_multicalls()
Create smart contract calls needed to read the historical state of this vault.
Multicall machinery will call these calls at a specific block and report back to
process_result()
- abstract process_result(block_number, timestamp, call_results)
Process the result of mult
Calls are created in
construct_multicalls()This method combines result of this calls to a easy to manage historical record
VaultHistoricalRead
- Parameters
block_number (int) –
timestamp (datetime.datetime) –
call_results (list[eth_defi.event_reader.multicall_batcher.EncodedCallResult]) –
- Return type
- class VaultInfo
Bases:
TypedDictVault-protocol specific intormation about the vault.
A dictionary of data we gathered about the vault deployment, like various smart contracts associated with the vault
Not standardised yet
- __init__(*args, **kwargs)
- __new__(**kwargs)
- clear()
Remove all items from the dict.
- copy()
Return a shallow copy of the dict.
- fromkeys(value=None, /)
Create a new dictionary with keys from iterable and values set to value.
- get(key, default=None, /)
Return the value for key if key is in the dictionary, else default.
- items()
Return a set-like object providing a view on the dict’s items.
- keys()
Return a set-like object providing a view on the dict’s keys.
- pop(k[, d]) v, remove specified key and return the corresponding value.
If the key is not found, return the default if given; otherwise, raise a KeyError.
- popitem()
Remove and return a (key, value) pair as a 2-tuple.
Pairs are returned in LIFO (last-in, first-out) order. Raises KeyError if the dict is empty.
- setdefault(key, default=None, /)
Insert key with a value of default if key is not in the dictionary.
Return the value for key if key is in the dictionary, else default.
- update([E, ]**F) None. Update D from mapping/iterable E and F.
If E is present and has a .keys() method, then does: for k in E.keys(): D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
- values()
Return an object providing a view on the dict’s values.
- class VaultPortfolio
Bases:
objectTrack assets and balances in a vault.
Offchain method to track what assets a vault contains
Takes
TradingUniverseas an input and resolves all relevant balances the vault holds for this trading universeBecause of brainrotten and awful ERC-20 token standard, the vault does not know what tokens it owns and this needs to be specific offchain
- __init__(spot_erc20, dex_hints=<factory>)
- Parameters
spot_erc20 (eth_defi.vault.lower_case_dict.LowercaseDict) –
dex_hints (dict[eth_typing.evm.HexAddress, list[str]]) –
- Return type
None
- get_raw_spot_balances(web3)
Convert spot balances to raw token balances
- Parameters
web3 (web3.main.Web3) –
- Return type
- property tokens: set[eth_typing.evm.HexAddress]
Get list of tokens held in this portfolio
- class VaultSpec
Bases:
objectUnique id for a vault.
Each vault can be identified by smart contract address by one of the contracts, related to its deployment. Usually this contract is vault contract itself.
We need both chain and address to specify vault we mean.
- __init__(chain_id, vault_address)
- Parameters
chain_id (int) –
vault_address (Union[eth_typing.evm.HexAddress, str]) –
- Return type
None
- verify_parquet_file(path, expected_rows=None, expected_schema=None, required_columns=None)
Read back a parquet file after writing and verify its integrity.
Performs a metadata read-back (not a full table load) to check:
The file can be opened and its metadata read without errors
Row count matches
expected_rowsif providedAll columns in
expected_schemaare present with correct types (extra columns are permitted — e.g. native protocol columns)All
required_columnsare present
Uses
pq.read_metadata()andpq.read_schema()instead ofpq.read_table()to avoid loading the full dataset into memory.This function should be called on a temp file before the atomic replace so that the previous good file is preserved when verification fails.
- Parameters
path (Union[pathlib.Path, str]) – Path to the parquet file to verify.
expected_rows (Optional[int]) – If set, assert the file contains exactly this many rows.
expected_schema (pyarrow.Schema | None) – If set, verify that all columns in this schema are present with the correct types. Extra columns are permitted.
required_columns (Optional[list[str]]) – If set, verify these column names are present.
- Returns
Verification result with metadata about the file.
- Raises
ParquetVerificationError – If any verification check fails or the file cannot be read.
- Return type