What does a market-data API actually give you?
A market-data endpoint does not define your coverage or rights. Understand feeds, timing, entitlements and the limits of API access.
Read articleBankster API
Find your next read.
Six ways into the systems behind finance.
Follow your curiosity
Explore by category.
Data & access ↗
What you can retrieve, what you can keep, and what the data can actually tell you.
A reproducible model needs reproducible data rights
Saved code does not guarantee a repeatable model run. Data retention, access and licensing rights can determine what can be reproduced.
Read article ↗Point-in-time data: the history a backtest was allowed to know
A backtest must use what was knowable at the time. Separate event dates, publication times, revisions and historical data availability.
Read article ↗LEI, ISIN, FIGI and MIC: identifying the entity, instrument and venue
LEI, ISIN, FIGI and MIC identify different things. Learn why matching an entity, security or venue does not establish permission.
Read article ↗Data lineage, quality checks and reconciliation establish different facts
Knowing where data came from does not prove it is complete or correct. Compare lineage, quality checks and reconciliation.
Read article ↗Lakehouse interoperability: files, tables, catalogs and writers
Shared file formats do not guarantee shared table behavior. Examine the catalogs, metadata and writer rules behind lakehouse interoperability.
Read article ↗Trading & execution ↗
From an order leaving the desk to the trade that actually gets done.
Execution benchmarks can reward unfinished orders
A benchmark based on filled orders can hide the cost of unfinished work. See how residual orders can reverse an execution ranking.
Read article ↗From order to settlement: the systems that change a trade’s state
An accepted order, a fill and a settled trade are different events. Follow the systems and handoffs that change a trade’s state.
Read article ↗Concurrent orders and shared risk limits
Two valid order checks can exceed one shared limit. Work through reservations, concurrent decisions and the state needed to enforce a cap.
Read article ↗Low-latency trading: separating delay from clock error
A timestamp difference can mix processing delay with clock error. See how synchronization changes what a trading latency measurement means.
Read article ↗Trading shutdowns have a completeness problem
Stopping one gateway does not prove all trading activity has stopped. Trace in-flight orders, acknowledgments and shutdown completeness.
Read article ↗FIX, ISO 20022 and FDC3: what each standard connects
FIX, ISO 20022 and FDC3 connect different parts of finance. Compare message exchange, business meaning and desktop context.
Read article ↗Payments & funding ↗
Moving money changes balances, obligations and the need for cash.
Continuous payments create a continuous funding problem
Continuous payment acceptance does not ensure continuous funding. Explore the timing mismatch between outgoing flows and replenishment.
Read article ↗Gross settlement can reduce waiting while increasing funding needs
Settling payments sooner can require more funding. Compare gross settlement, netting and the timing of incoming and outgoing cash.
Read article ↗Payment accepted, account posted, funds settled: three different states
Payment accepted does not mean funds settled. Separate an interface response, a customer ledger posting and the settlement event.
Read article ↗Counterparty exposure: from trades to netting sets and collateral
Trade values become counterparty exposure through agreement scope, netting and collateral. Follow the calculation boundaries.
Read article ↗Positions, accounting balances and risk exposures
A position, an accounting balance and a risk exposure describe different views of a trade. See where their numbers diverge.
Read article ↗How an approved loan becomes a serviced account
An approved loan still needs contracts, servicing records and ledger entries. Follow the handoff from lending decision to account operation.
Read article ↗Software & models ↗
The decisions between a calculation that works and a system people can rely on.
Exactly-once delivery and exactly-once financial effects
Delivering an event once and applying its financial effect once are different guarantees. Explore retries, deduplication and ledger boundaries.
Read article ↗Why financial firms use several programming languages
Why trading, pricing and operations use different languages—and how memory, latency and numerical workloads shape the choice.
Read article ↗From notebook to model service: preserving a financial calculation
Moving a calculation into production requires more than packaging code. Preserve inputs, versions, units and execution assumptions.
Read article ↗Decimal arithmetic does not choose a rounding policy
Decimal arithmetic cannot decide when to round. A worked example shows why line-item and aggregate rounding produce different balances.
Read article ↗Transaction databases, tick stores and search indexes
Ledgers, tick stores and search indexes preserve different facts. Learn why one database rarely serves every financial workload.
Read article ↗Where CRM and ERP sit in a financial institution
Client records and business accounts do not replace trading or settlement systems. Explore where CRM and ERP connect to financial workflows.
Read article ↗Risk & controls ↗
Who can act, what gets flagged, and the work that automation leaves behind.
More automated fraud screening can create more manual work
Better screening rates can still create a larger review queue. Work through how transaction volume and alert rates interact.
Read article ↗AI assistants connect permissions as well as data
Connecting an AI assistant to financial tools also connects permissions. Trace the authority behind retrieval, recommendations and actions.
Read article ↗KYC, AML and sanctions screening answer different questions
Customer identity, suspicious activity and sanctions restrictions need different controls. Follow the handoffs between KYC, AML and screening.
Read article ↗Workload identity and privileged access
A service identity and a privileged session grant different powers. See how machine identity, secrets and administrative access fit together.
Read article ↗Electronic recordkeeping: retention, legal hold and eDiscovery
Keeping records, preserving them under legal hold and retrieving evidence are distinct jobs. Map their storage and workflow requirements.
Read article ↗Infrastructure ↗
Where systems run, how they fail, and what it takes to recover or leave.
Two clouds can share one failure dependency
Two cloud deployments can fail through one shared dependency. Examine the identity, network and recovery systems behind apparent redundancy.
Read article ↗Recovery time includes transaction reconciliation
Restarting a service does not restore a trusted transaction state. See why reconciliation belongs inside the recovery timeline.
Read article ↗Vendor exit costs depend on reconstructing financial state
Exported files are only part of a vendor exit. Reconstructing balances, history and operating state can determine the real migration cost.
Read article ↗Cloud, on-premises and SaaS describe different choices
Cloud location, infrastructure ownership and SaaS delivery answer different questions. See how they intersect in financial systems.
Read article ↗Private connectivity, encryption and authorization protect different boundaries
A private connection is not encryption or permission. Trace the separate boundaries protected by network paths, cryptography and identity.
Read article ↗Why healthy services can produce incomplete financial reports
Healthy services can still omit financial events. Learn why operational monitoring needs separate checks for reporting completeness.
Read article ↗