Notes on bots, edges and the gap between them.
Practical writing on futures algo trading — backtests vs live, prop-firm rules, signal logs and the analytics that keep an edge honest.
What Rithmic's API actually gives you — and the access trap
R|Protocol is a genuinely good execution and market-data API. The catch nobody warns you about is whether your account is even allowed to use it.
What you actually get from TopstepX — the data, not the marketing
TopstepX exposes a modern REST + realtime gateway. Here's the data surface a bot really sees, what's live, what you have to poll, and where it bites.
Why we won't put TradingView on our own site
A small, opinionated decision about third-party scripts — and why a trading-data product especially can't be casual about it.
What a good signal log tells you (and a bad one hides)
Every entry and exit your bot takes should carry a reason. Here's how to read them — and why 'NR7_breakout · vol z=2.1 · OFI+' beats a bare timestamp.
Design your strategy around the drawdown, not the upside
Everyone backtests for return. The traders who actually get funded backtest for the worst week.
Daily loss vs trailing drawdown: how prop-firm rules actually work
The two limits that blow most funded futures accounts, explained plainly — and how to track them before your bot trips a halt.
Topstep, Apex, MyFundedFutures: what actually differs
Every prop firm sells the same dream. The fine print is where they split — and where most evals quietly die. Field notes from someone who read all of it.
Why backtested edges die in live trading
The gap between a backtest and a live account is where most futures strategies quietly fail. Here's what actually causes it — and how to watch it open in real time.
Can you trust your backtest? Walk-forward and Monte Carlo, in plain English
Two robustness checks that separate a real edge from a curve you accidentally fit to history — explained without the math degree.
Bots don't predict the trend. They just react before you can.
Trend-following with a bot isn't magic. It's the discipline you can't fake at 2pm on a red day — automated.
Micros, minis, and Micro Bitcoin: sizing so you trade tomorrow
Trading MES instead of ES isn't 'going small' — it's how you test a real edge without betting the account on a hunch.
The live data problem nobody warns you about
Why your live feed and your backtest data never quite agree — and why 'free futures data' is mostly a trap.
Running a bot 24/7 is an ops problem, not a strategy problem
The strategy is maybe 20% of it. The other 80% is the bot still being alive, on the right clock, when the open hits.
Testing on data you didn't clean is testing on noise
Bad bars, roll gaps, timezone drift — the boring data problems that quietly invalidate a backtest, from the person who had to fix them.
Your trading bot needs a black box recorder
When a bot does something dumb at 14:32, 'it just did' is not an answer. Telemetry is.