Spent last night reorganising my whole process. Moving countries, two kids, stay at home dad, general life. It had got messy without me noticing. No new systems, no new code. Just clearing out. Feels like I can actually see what I’m doing again.

The automated testing is still doing it's thing

AI doesn't have inspiration. I asked it to design a slot car track for my son. Its first suggestion was an oval. It will do the same with your trading system. The most average answer, delivered with confidence.

Latest system is live on demo. Intraday breakout, one set of settings, five markets. Anyone can make a backtest look good by tuning each market separately. The test is whether the same settings hold up everywhere.

Automating my trading transformed how I understood the markets. Automating my backtesting has done it again. I got into trading for money and time. After five years, it’s actually starting to happen.

The automated testing is doing its job. Most systems fail. Hard gates first, then parameter sensitivity. Most ideas die at one of those stages. That’s the process working, not failing. The few that get through are the ones worth looking at myself.

Finally automated my robustness testing. One prompt and the whole thing runs in MT5 while I’m doing something else. IS, OOS, every market. I just read the report at the end. Years spent building systems that trade without me. Took me long enough to do the same with the testing.

Most trading courses teach you what to trade. Very few teach you how to know if you have an edge. Those are completely different problems. Only one of them matters long term.

Started a channel on systematic trading. No finance background, no coding background — the systems are AI-written MQL5, I just learned to prompt them properly. Are you discretionary or systematic? Genuinely curious.

I passed prop firm challenges not because my systems win more often. They don't. I passed because the risk model meant I could absorb losing streaks that would have blown most traders out. Edge without structure is just luck on a timer.

Ran a trail half marathon today. 1:40:41, 38th of 521. Top 7% if finishers. Not trading related. Except that it is... I can train for this because I'm not tied to a screen during sessions. Automation buys you a life outside the charts. That's the whole point.

Monte Carlo simulation doesn't tell you what will happen. It tells you the range of what could happen if your trade sequence was randomised 1000 times. That range is where your risk model should live, not the optimistic middle of the distribution.

4am start. Robustness testing on a new system. 10 years in-sample, 2.5 years out-of-sample, four markets. One question: does the edge hold on data it's never seen? Nobody posts this part. It's the part that decides whether anything goes live.

Newest addition to the portfolio analyser: correlation matrices. Daily P&L correlation shows if your systems win on the same days. Drawdown correlation shows if they lose on the same days. The second matters more. Systems that drawdown together aren't a portfolio.

The opening range on NAS100 tells a story every single session. Whether that story means a breakout, a fade, or nothing, that's what the data decides. Not intuition.

Failing a prop firm challenge doesn't always mean your system is broken. Sometimes it means you started during a bad stretch. A weak system with good variance passes. A good one with bad variance fails. The result tells you less than the years of testing behind it.