forex-scalping.com

Price action in the last five minutes of each hour

Something I keep seeing on M1 is how the last few minutes before the top of the hour behave differently from the middle of it. A lot of desks and algos work on hourly marks, and plenty of retail traders only look at the chart when the H1 candle is about to close. So you get a little burst of activity from about :55 onward, then a pause once the new candle opens.

The pattern I watch for is simple. If price has been grinding in one direction for most of the hour, those last minutes often see a push to tidy up the H1 candle, either extending the move to a clean high or low or pulling back so the candle closes off its extreme. Neither is reliable enough on its own to trade blindly. What it does change is how I handle a position I'm already in. Tightening a stop at :54 into a known burst of noise is usually a bad idea, because the wick that tags it is often gone by :02.

I also stopped taking fresh entries in that window unless the setup was already formed before :50. Entries that only appear because of the hourly shuffle tend to have nothing behind them once the new candle starts.

The effect is much weaker during Asia and stronger around London and the New York morning, which makes sense given who is active. On quiet days you can barely see it. I'd be interested to hear if anyone tracks this on the half hour too, since some data and fixings land on :30.

One thing to separate here is the hourly candle close versus actual scheduled events. A fair share of releases land exactly on the hour, so some of what looks like a top of the hour effect is really traders getting flat before a number. If you want to study it properly, strip out the hours that had anything on the calendar and look only at the clean ones. On my charts the effect shrinks once you do that, but it doesn't disappear. The remaining bit seems to come from people managing H1 candles and from automated rebalancing that runs on a timer. That's a smaller, messier edge, so I treat it as context and not as a setup.

For anyone wanting to check this on their own pair, the easiest way is a simple script that measures the range of each minute of the hour across a few months of M1 data. Group by minute number, average the high minus low, and plot it. You don't need anything fancy. If the bars around 55 to 59 and 0 to 2 stand out against the middle of the hour, you have your answer for that pair and that broker feed. Do it separately for London hours and the rest of the day, because averaging everything together hides the session differences completely.

Spread is the other thing that moves in that window, at least at some brokers. If liquidity providers are refreshing quotes around the hour you can see the spread flicker wider for a few seconds. On a 2 pip target that matters a lot. So even if the price behaviour gave you an edge, the cost of getting in right at :58 can eat most of it. I'd rather let the new candle open, see where the spread settles, and then decide. Missing the first move is cheaper than paying a wide spread into noise.

Last point from me. The candle close effect is strongest on the hours that also close a bigger candle. The 4 hour closes and the daily close get more attention than a random 11:00 bar, so the push into those marks is usually larger. If you trade off server time, check which hours your broker's H4 candles actually close on, because it varies by broker and by daylight saving. Knowing that one detail tells you which hourly marks to be more careful around and which ones you can mostly ignore.

Fairman, the last five minutes of the hour is an interesting window to study because so many orders are tied to round times. A simple test would be to line up the range and direction of those minutes with the middle of the hour over a quarter of data. If they differ consistently, that is a real effect; if not, the pattern may be something the eye notices more than the data shows.