Days under water: counting how long the account sat below its peak, from daily equity

[Fairman]

Days under water: counting how long the account sat below its peak, from daily equity

Maximum drawdown says how deep. It does not say how long you spent below the previous high, which is what the weeks feel like.

Daily closing equity over 20 days, indexed to 100:
100, 101.2, 100.4, 99.8, 101.5, 102.3, 101.1, 100.2, 99.5, 98.9, 100.1, 101.8, 103.0, 102.4, 101.6, 103.4, 104.1, 103.2, 102.9, 104.6

For each day, the running peak is the highest value so far. A day is under water if equity is below that peak.

Here the account was under water on 12 of the 20 days, 60% of the time. The longest stretch was 6 consecutive days, days 7 to 12, between the 102.3 peak and the 103.0 that finally cleared it. The deepest point was 98.9, which is 3.3% below the 102.3 peak.

An account that ends up 4.6% can spend most of the time below its high. That is normal, not a sign something is wrong.

Spreadsheet, equity in column A: column B =MAX(A$1:A1) for the running peak, column C =IF(A1<B1,1,0), then =SUM(C1:C20)/20.

[FXS HongKong Scalper]

Days under water is a measure I wish more traders tracked. It explains why many give up on working strategies: the time spent below the peak feels long even when the drawdown is small. Knowing the typical length in advance helps manage expectations.

[Fairman]

The longest stretch needs a counter that resets: in column D, =IF(C2=1,D1+1,0), and the longest run is =MAX(D:D). In the example the counter reaches 6, then resets when the account makes a new high at 103.0. Both numbers, share of days under water and the longest stretch, matter for planning. A trader who expects to be at a new high every few days will read a six day stretch as failure, while a trader who knows the typical stretch for the system is five to ten days will treat it as routine. Measuring it from the journal shows what routine looks like for your own account.

[Fairman]

Time under water grows with the length of the record, because a longer record has more chances to contain a long stretch. Over 20 days the longest is 6. Over 250 days it will usually be longer, so quote it for a stated window, such as the last 250 trading days, or comparisons between accounts mean nothing. And read it against the end result. If a year contains a three month stretch below the peak while the account finishes 20% higher, the three months are part of the shape of that strategy and not a malfunction, though only your own record can tell you how long a stretch to expect.

[Fairman]

A related figure is the ulcer index, which is the square root of the average squared percentage drawdown from the running peak. It rewards shallow and brief dips and punishes deep and long ones. For the 20 days above, the percentage below the peak is zero on the 8 peak days and runs up to 3.3% at the worst point, and the ulcer index works out at 1.34. Computing it needs one more column, the percentage below the peak, then squares, an average and a square root. It gives one number that combines depth and duration, which maximum drawdown alone cannot. It is less well known, so state the definition when you quote it to someone.

[Fairman]

The prop firm view is different. Many programmes reset or evaluate on a static balance and not on the running peak, so time under water relative to the starting balance matters, and a trailing drawdown rule works relative to the peak. Decide which peak you are measuring against and write it in the journal header. For a personal account the high water mark is the natural choice. For an evaluation the starting balance is the figure that matters, and days spent below it, rather than below the peak, are what the clock and the rules look at.

[Shadow Trader]

Every morning in Tokyo, banks set the day's reference rate for yen. Corporate and retail customers who want a fixed rate for the day have orders matched against it, and the banks end up with a pile of dollar buying or selling that needs to be offset.

That offset flow tends to appear in the minutes before 9:55 JST. On days when importers are heavy buyers of dollars, the pair often drifts higher into the fix, then stalls or slips right after.

The effect is not magic. It's inconsistent, and it's overshadowed by news. But on quiet days, it creates a faint pattern: a slow grind into the fix and a pause afterward.

For scalping, the useful part isn't predicting direction. It's knowing that volatility may spike briefly around the fix and then compress. Entering just before 9:55 with a tight stop has a different risk profile than entering at 9:40.

Fix-day variations exist too. The last business day of the month, and days ending in 5 and 10 in Japan (gotobi days), often see heavier importer flow.

Worth testing on actual tick history rather than trusting folklore.

Gotobi days versus regular days on USDJPY: any measurable difference in the data?