Daily, Weekly and Monthly Loss Limits: Building a Layered System

[FXS Prop Desk]

A single daily limit stops a bad day. A layered system also protects against a bad week or a bad month.

A practical variant is a trailing daily stop: once the day is up by a certain amount, the stop moves so that only part of that gain can be given back. It protects good days without forcing an early finish.

Example structure:

• Daily: -2% → stop for the day.
• Weekly: -4% → stop until next Monday, review journal.
• Monthly: -8% → reduce size by half for the next month or switch to demo.

Why layers help:

• Several "normal" bad days can add up to a serious drawdown if only daily limits exist.
• The weekly review forces a pause to check if the market or the trader changed.
• The monthly rule stops slow bleeding.

Do you use limits beyond the daily level, and what are your thresholds?

[Fairman]

Monthly: -8% → reduce size by half for the next month or switch to demo.

One thing worth checking is how the layers interact. With a 2% daily and 4% weekly stop, two bad weeks already reach the monthly -8%, so the monthly rule may trigger mid-month more often than expected. That might be exactly what you want, but it helps to run your past daily results through the full set of rules and see how often each layer would have fired. Also decide whether the weekly and monthly windows are calendar based or rolling. A rolling 5-day window catches a bad Thursday to Wednesday stretch that a Monday reset would split across two weeks and never flag.

[Shadow Trader]

Fairman, the interaction point is exactly where my old sheet failed. With a 2 percent daily and 4 percent weekly, two ugly weeks hit the monthly 8 percent by the 18th, and I was still taking full size because the monthly rule had not printed yet.

I now run a trailing daily lock once the day is up 1.5 percent: I will give back only 0.6 from that high water mark. It saved three strong London mornings last month that would have ended flat under a hard daily stop.

Weekly pause until Monday stays. The review is the point, not the calendar theatre. If the journal shows the same mistake twice, size drops before the monthly layer ever fires.

I disagree with switching straight to demo at minus 8. Halving size for ten sessions keeps the feedback loop live without digging deeper.

Where do you set the trailing lock relative to the daily hard stop? I keep both, and the lock sits inside the hard floor.

[Shadow Trader]

I took Fairman's suggestion and ran my last 250 sessions of daily results through the full set of layers. The surprise was how much the choice between calendar and rolling weeks changed the count.

img_13692_r2.png (48.88 KiB) Viewed 56 times

The chart shows how often each layer would have fired. The 2 percent daily stop hit 9 times, the calendar week rule at 4 percent hit 3 times, the rolling five session version hit 6 times, and the monthly 8 percent rule fired once.

Two of the three extra rolling triggers came from a bad Thursday and Friday followed by a bad Monday. A Monday reset split those losses across two weeks, so neither week looked like a problem on paper.

Going rolling does cost something, because each trigger meant sitting out until the window recovered. Over the year it added 7 days of no trading, which I can accept given what those stretches looked like.

The one tweak I made is to measure the rolling window from closed sessions only. Counting the current day's open loss tripped the weekly rule twice in the middle of a trade, and that is the job of the daily stop.

[Shadow Trader]

Daily limits address single sessions, but a string of bad days can cause serious damage that no single day suggests. A weekly loss limit provides a second layer of protection by pausing trading when cumulative losses exceed a threshold.

A reasonable structure might cap weekly losses at two to three times the daily limit. When triggered, trading stops until the next week or until a review has been completed. During the pause, analysis of what went wrong, whether market conditions shifted, whether execution slipped, or whether discipline eroded, takes priority.

The cooling-off period serves both financial and psychological purposes. It prevents compounding losses driven by frustration and gives time to restore a clear head. Returning at reduced size for a few sessions can rebuild confidence gradually. Tracking the frequency of triggered limits over months shows whether the strategy is deteriorating or whether performance remains within normal variation. Frequent triggers suggest that either the strategy or the risk settings need fundamental rethinking.

[FXS Prop Desk]

Also decide whether the weekly and monthly windows are calendar based or rolling.

The rolling point matters most on prop accounts, since nobody at the firm resets anything for a bad Thursday-to-Wednesday stretch. When running the old results through all three layers, one more output is worth looking at: how often the weekly rule fires in a week where the daily rule never did. If that almost never happens, the weekly layer adds nothing at those settings, and either the daily limit is too loose or the weekly one is too wide.

[Fairman]

That conditional count is a useful redundancy check, though I would not call the weekly layer useless just because it rarely fires in weeks with no daily stop. Its purpose may also be to interrupt a run of separate bad sessions before the account reaches the monthly threshold. I would replay the same dated results with each policy and record daily-limit hits, weekly triggers when the daily rule did not fire, total drawdown, and days sidelined. Comparing calendar weeks with a rolling five-session window would show whether the reset boundary is hiding clustered losses, as in the Thursday-to-Wednesday example. Also define the accounting consistently: closed versus floating P/L, commissions, time zone, and when a session becomes final. Then test whether the pause leads to a meaningful review or changes behavior, rather than treating the trigger count alone as success. If the weekly layer adds no protection in the replay, it may be redundant at those thresholds; if it catches damaging sequences the daily cap misses, the extra interruption may be justified even if it is infrequent. The objective is a coherent risk policy, not the largest possible number of limits.