Time-of-Day Filters: Matching Your Strategy to the Clock
Market behavior changes through the day, so the same strategy can have very different results at different hours.
Break your trades down by session and hour. Many traders find that their edge concentrates in one window. A breakout strategy may do best at the London open, when liquidity and volatility rise. A range strategy may do better in the quiet Asian session. A pullback strategy may work best during the morning overlap.
The reasons are structural. Spreads tighten and liquidity peaks when major centers are open. News releases cluster at particular times. Participants differ by session: Asian hours involve more regional players, London brings large banks, and New York adds US data and funds.
To use the idea, analyze your journal. Create a table of trades by hour with win rate and average R. Look for hours where performance is clearly different from the rest, with enough trades for the comparison to mean something.
Then test a rule that restricts trading to the productive window, and see whether results hold on new data.
Be careful with small samples. Splitting 60 trades across ten hours leaves six per hour, which says little.
Remember daylight saving changes, which shift the relationship between the clock and session opens twice a year.
Practical step: sort your trade history by entry hour in a spreadsheet and find your best and worst two-hour windows.
Time-of-Day Filters: Matching Your Strategy to the Clock
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
When people actually run this on their own journal, the last hour of their session often looks worst. Partly market, partly being tired by then. Cutting it can be the easiest improvement there is.
It’s Fairman 
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Use your broker's server time or a fixed zone consistently in the journal. Mixing local time and platform time across months makes the hour breakdown useless.
It’s Fairman 
-
FXS Prop Desk
- Posts: 1693
- Joined: Wed Oct 07, 2026 9:52 am
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Breaking results down by hour works best with enough trades in each bucket. With a few hundred trades split into 24 hours, many buckets have only a handful of trades, and the "best hour" may be noise. Grouping by session, or into three- or four-hour blocks, gives more reliable comparisons, and it's worth checking that the pattern holds in both halves of the data before adding the filter.
FXS Official Team | The Idea Lab & Strategy
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Spread is one reason an hour can look bad that has nothing to do with the setup. Say a scalper averages a 6 pip target and pays 0.8 pips in the London morning but 2.5 pips around the 5pm New York rollover, then the same entries give up 13 percent of the target in one window and over 40 percent in the other.
Splitting results by hour before and after costs separates the two. If an hour only turns negative once spread is included, the cost is the problem.
Logging in UTC is cleaner than local time but has a catch, since London opens at 07:00 UTC in summer and 08:00 in winter. A filter like 07:00 to 10:00 UTC quietly covers a different slice of the session after the clocks change, so hours counted from the open make a better column.
On sample size, the block idea from the last post can be put in numbers. With 60 trades in a block and a win rate near 50 percent, the 95 percent range is roughly 37 to 63 percent, so two blocks need to differ a lot before the gap means anything.
Has anyone tried filtering by recent volatility instead of the clock? It follows the same logic without the calendar problems.
Splitting results by hour before and after costs separates the two. If an hour only turns negative once spread is included, the cost is the problem.
Logging in UTC is cleaner than local time but has a catch, since London opens at 07:00 UTC in summer and 08:00 in winter. A filter like 07:00 to 10:00 UTC quietly covers a different slice of the session after the clocks change, so hours counted from the open make a better column.
On sample size, the block idea from the last post can be put in numbers. With 60 trades in a block and a win rate near 50 percent, the 95 percent range is roughly 37 to 63 percent, so two blocks need to differ a lot before the gap means anything.
Has anyone tried filtering by recent volatility instead of the clock? It follows the same logic without the calendar problems.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
The block approach from FXS Prop Desk raises a second question. A filter can raise the average R per trade and still lower the total, and it is easy to optimise for the wrong one.
Say 260 trades split into four three hour blocks counted from the London open. The first block has 90 trades at plus 0.20R, the second 70 at plus 0.05R, the third 60 at plus 0.15R, and the last 40 at minus 0.10R.
In the made up chart above, total R per block is the bars and the average per trade is the line. Dropping the last block is an easy call, since the total goes from 26.5R to 30.5R and the average climbs too.
The second block is the tempting one. Cutting it lifts the average from about 0.14R to 0.18R, but the total falls to 27R, so the cleaner looking stats come with less money.
Cutting it makes sense only if that quiet block costs something not on the sheet, like attention for the overlap that follows. A tired third block would be the sign.
How do others weigh this, by total R or by average? For a funded account with a drawdown limit, fewer trades at a higher average might still be the better choice.
Say 260 trades split into four three hour blocks counted from the London open. The first block has 90 trades at plus 0.20R, the second 70 at plus 0.05R, the third 60 at plus 0.15R, and the last 40 at minus 0.10R.
In the made up chart above, total R per block is the bars and the average per trade is the line. Dropping the last block is an easy call, since the total goes from 26.5R to 30.5R and the average climbs too.
The second block is the tempting one. Cutting it lifts the average from about 0.14R to 0.18R, but the total falls to 27R, so the cleaner looking stats come with less money.
Cutting it makes sense only if that quiet block costs something not on the sheet, like attention for the overlap that follows. A tired third block would be the sign.
How do others weigh this, by total R or by average? For a funded account with a drawdown limit, fewer trades at a higher average might still be the better choice.
-
gwhitlock64
- Posts: 60
- Joined: Tue Oct 20, 2026 11:01 pm
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Shadow Trader, your averages are off: it's 0.10R to 0.14R, tbh.Shadow Trader wrote:A filter can raise the average R per trade and still lower the total, and it is easy to optimise for the wrong one.
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Hi gwhitlock64,gwhitlock64 wrote: Wed Oct 21, 2026 7:30 amShadow Trader, your averages are off: it's 0.10R to 0.14R, tbh.Shadow Trader wrote:A filter can raise the average R per trade and still lower the total, and it is easy to optimise for the wrong one.
you have good eye, thanks for that.
@ShadowTrader, pls check it
Preserve your own money. Scale with the market's money. Exponential growth is the ultimate key.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Thanks gwhitlock64 and PTScalper, good catch, and fair point that the post skipped a step.PTScalper wrote: Wed Oct 21, 2026 9:59 am Hi gwhitlock64,
you have good eye, thanks for that.
@ShadowTrader, pls check it![]()
You're right about the first cut. With all four blocks the total is 26.5R over 260 trades, about 0.10R per trade. Dropping the last block gives 30.5R over 220 trades, about 0.14R.
The 0.14R to 0.18R in the post is the second step, cutting the middle block on top of that. That leaves 27R over 150 trades, or 0.18R.
So the average went 0.10R, then 0.14R, then 0.18R. Total R went up on the first cut and back down on the second.
I should have written out all three steps. Thanks for keeping the maths honest.
-
gwhitlock64
- Posts: 60
- Joined: Tue Oct 20, 2026 11:01 pm
Re: Time-of-Day Filters: Matching Your Strategy to the Clock
Ah, that makes sense now, I read it as one cut rather than two, so the 0.14R to 0.18R jump looked dodgy to me. A little running table (trades, total R, average after each cut) would make it obvious at a glance, imo. Im usually the one adding up on a beer mat at the pub quiz, so I appreciate seeing the steps laid out.Shadow Trader wrote:Thanks gwhitlock64 and PTScalper, good catch, and fair point that the post skipped a step. You're right about the first cut.