This morning our DAX EA looked at the market, measured the session, and refused to open a trade.
No error. No missed signal. It worked exactly as designed: the morning was already too volatile relative to the average daily range, and statistically that regime has been hostile to our entry model. So the EA sat out. It will sit out tomorrow too, if tomorrow looks the same.
Most EA marketing trains buyers to expect the opposite — activity as proof of life. Daily trades, daily screenshots, daily profits. But activity is the cheapest thing an algorithm can produce. Grid and martingale systems produce enormous activity right up until the day they produce a margin call.
Discipline is harder to sell, because discipline looks like nothing happening.
Here is what "nothing" costs and buys in our case. The volatility day filter skips roughly a third of trading days, and the skipped days come in clusters — in 13 years of data, quiet streaks of a week or more happened about four times a year, and twice the filter sat out for a whole month. In exchange, the backtest drawdown dropped by about a third and the average quality of each taken trade more than doubled. We knew these numbers before going live, so a silent week does not scare us. It is in the plan.
That experience taught us a checklist we now apply to any EA — including our own. If you are shopping for one, steal it:
1. Is there a live signal, and is it running right now? Not a screenshot of one. A link. If the seller's own money is not following the EA, ask why.
2. Is the maximum drawdown printed next to the profit number, in the same font size? Return without drawdown is not a result, it is half of one.
3. Does every trade carry a stop-loss, set before entry? Ask what the worst single day can lose. If the answer is vague, the answer is "a lot".
4. Does the seller tell you about losing streaks in advance? Our documentation says plainly: nine losses in a row is within normal operation. If you cannot sit through that, no EA will help you.
5. Was the backtest done on real tick data with spread and commission? A backtest at zero spread is fiction. We learned this the expensive way and re-ran everything on measured costs.
6. Does the EA sometimes do nothing? Strange as it sounds, this may be the strongest signal on the list. A system with no conditions under which it stays out has no model of when it loses.
We built Quartz DAX to pass this checklist, and the live monitoring signal went up this week so that every future trade is verified by the platform, not by our screenshots. The track is young — that is what "honest" looks like at the start. It grows one disciplined day at a time, including the days when the most profitable action is none.
Quartz DAX on the Market: https://www.mql5.com/en/market/product/192954
Live monitoring: https://www.mql5.com/en/signals/2389115
