Tagged “tooling”
-
Testing a Backtester With Data You Made Up
Your engine is untested code sitting inside every result. How synthetic series with hand-computable answers catch fill, cost, and alignment bugs.
-
A Reproducible Workflow for Strategy Research
Immutable raw data, config-driven runs, seeded randomness, and a research log that counts your attempts. The setup that makes a result mean something.
-
What Sits Between a Backtest and a Live Bot
The components a backtest never needs: a real-time feed, persistent state, position reconciliation, idempotent ordering, monitoring, and a kill switch.
-
What Paper Trading Catches, and What It Misses
Paper trading validates plumbing, not edge. The failure classes it reliably exposes, the ones it structurally cannot, and how long to run it.
-
The Python Tooling Stack for Crypto Quant Research
The Python libraries quant researchers use for crypto: data access, pandas wrangling, backtesting engines, and a reproducible research workflow.
-
How to Evaluate a Backtesting Library
Judge a backtesting library by its fill model, cost model, and auditability rather than its feature list. The questions to ask before you trust one.
-
The Exchange Abstraction Layer, and What It Hides
A unified exchange client normalizes symbols, candles, and errors across venues. What that abstraction genuinely gives you, and what it cannot paper over.
-
Where Historical Crypto Data Comes From
The four categories of crypto price history, what each is good for, and the pagination, rate-limit, and coverage problems you inherit from each one.
-
Event-Driven vs. Vectorized Backtests
Vectorized backtests are fast and lie easily; event-driven ones are slow and honest. What each can and can't represent, and how to use both together.