Notes from the workbench
Posts
Writing about technical product craft, experimentation, automated trading systems, and lessons learned while building.
The Stop That Could Never Fire
My 0DTE bot opened a position late enough that its configured time stop landed after the market close.
Read post →The Bot That Walked Away
My 0DTE bot submitted an entry, spent about a minute failing to confirm the fill, logged Aborting, and went back to scanning for signals.
Read post →A Kill Switch That Doesn’t Abandon the Trade
An autonomous trading bot is convenient until I decide, mid-session and away from my computer, that it should stop.
Read post →A Backtest Needs Two IDs
When a strategy has been backtested hundreds of times, “which run was that?” stops being an administrative question. It becomes part of the evidence.
Read post →When My Backtest Changed Its Mind
My candle-based backtest liked a 0DTE options strategy. The live system did not.
Read post →Teaching My Backtester Which Trades Not to Take
Aggregate backtest results tell me whether a strategy deserves another question. They rarely tell me which question to ask.
Read post →The Trail Wasn’t the Only Problem
I started with a narrow question: was my trailing stop too tight for a fast-moving 0DTE option?
Read post →From Disk to S3: A Safer Trade-Journal Pipeline
Every completed trade in my bot becomes a row in a CSV journal. For a long time, those files lived on the trading machine and were copied through Git as an i...
Read post →One Strategy, Two Accounts, Zero Silent Fallbacks
Running one strategy against two brokerage accounts sounds like a configuration change. In a live trading system, it is an isolation problem.
Read post →Ticket Hierarchy in JIRA
Defining Ticket Hierarchy in JIRA
Read post →Product Development Lifecycle
Product Development Lifecycle
Read post →