The week has a rhythm
Monday's trades earn nearly seven times what Friday's do — a clean fade across the week, visible in your own roster. But the real hazard of trading late in the week isn't the weak returns. It's the one thing your backtest is blind to.
1Monday's money, Friday's fumes
We sorted all 3,589 of the roster's backtested trades by the day they were entered. The pattern is not subtle:
| entry day | trades | win % | avg $/trade |
|---|---|---|---|
| Monday | 692 | 48.7% | $3.45 |
| Tuesday | 714 | 45.0% | $2.50 |
| Wednesday | 753 | 46.1% | $1.42 |
| Thursday | 734 | 44.1% | $0.85 |
| Friday | 696 | 44.5% | $0.50 |
Why? Early in the week, fresh money and fresh news set clean directional moves — the kind mean-reversion and breakout both feed on. By Friday afternoon, big players are squaring up for the weekend, volume thins, and moves get choppy and directionless. It's a real, well-known seasonality, and it shows up plainly in the roster.
2But Friday is not a losing day
Here's the trap the table sets: it's tempting to look at that $0.50 and think "ban Fridays." Don't. Friday is still profitable — just the least profitable day. Across the test it added +$349 net. Cutting it entirely would throw away real money to dodge a weakness that's modest.
This is the same discipline as everywhere else in the system: a weak edge is still an edge. You don't delete profit because it's small — you only cut something when it genuinely loses, or when it carries a risk the numbers can't price. Which brings us to the actual danger.
3The hazard the backtest can't see
The real reason to be careful late in the week isn't Friday's thin returns — it's what happens after Friday. The FX market closes around 21:00 UTC Friday and reopens 21:00 UTC Sunday. Over those ~48 hours the world keeps happening — elections, shocks, news — but you can't trade. When the market reopens, price can : jump to a new level with no trading in between.
Now here's the crucial bit. If you ask the backtest "is weekend holding dangerous?", it says no — it even looks slightly better. That answer is a lie, and it's a lie by omission. The backtest fills every stop at exactly the stop price. It has no concept of a Monday gap leaping over your stop. So the one risk that actually makes weekend holding scary — the oversized, gap-through-the-stop loss — is invisible to the very tool we normally trust.
Usually we act only on what the data proves. This is the exception that defines the rule: when you know a risk exists but your measurement is structurally blind to it, you don't wait for the data's permission — you act on the structural fact. Markets gap over weekends. That's not a backtest result; it's how the world is built.
4The rule: time as a risk filter, not a signal
So the fix is not to skip Fridays (you'd forfeit real profit). It's to stop opening new trades in the run-up to the close, so you never initiate fresh weekend-gap exposure. Trades already open are still watched; you just don't start new ones you'd be forced to carry across the gap. That's the — off by default, one switch to arm.
And notice the shape of the decision, because it generalises: day and time belong in the system as a filter — a rule about when not to trade — not as a signal. We checked whether the day of the week could predict direction; it can't, not cleanly, and trying would just invite overfitting. Time tells you when the ground is dangerous, not which way to point.
The campaign can refuse new entries from Friday WEEKEND_CUTOFF_HOUR (UTC) through the Sunday reopen — set WEEKEND_GUARD=true in .env. The dashboard's Time tab shows the live session clock and where the weekend gap sits; open positions carry a ⚠ held over weekend flag; and the Telegram card warns you when approving a trade late Friday would carry it across the gap. The tools exist — the guard is yours to switch on when you want it.
5The takeaway
1. The trading week has a rhythm. Monday is strongest, Friday weakest — but weakest still means profitable, so you filter, you don't ban.
2. Your best tool has a blind spot. The backtest can't see weekend gap risk, because it fills stops at the stop price. Trusting it here would be trusting a smoke detector with the battery out.
3. Some risks you act on without data's blessing. When you know a hazard is real and your measurement can't see it, structural judgement beats an in-sample number. Markets gap; guard the gap.