forex-scalping.com

Minute-by-minute spread tracking reveals hidden patterns

Most traders know the general pattern of spreads: tight in London and New York, wider in Asia. A minute-level log shows far more detail. Spreads often flicker at specific minutes, widen before releases, and show spikes at the start of each hour.

Building the log requires tick data collected from the platform for a few weeks. Compute the median spread for each minute of the day, and for each day of the week. The result is a heat map of cost over time.

The heat map instantly reveals the cheapest and most expensive windows. A scalper may find that a seemingly good time, like 08:00 to 08:15 London, is consistently expensive because of the opening flurry, while 08:30 to 09:30 is far better.

With that information, trading hours can be tuned to avoid expensive minutes. The improvement is not dramatic per trade, but across thousands of trades it adds a noticeable amount.

The data is free to collect and the analysis takes a few hours.

What does the heat map of spread by minute look like for the broker in use?

Did this for about four weeks on EUR/USD with a little script that wrote the spread on every tick to a CSV. The heat map showed the obvious stuff, like rollover at 5pm New York being awful, but also a smaller spike at the top of every hour in the Asian session, and a slightly wider spread for a minute or two around 9:00 and 10:00 London on data days. The cheapest stretch on my broker was roughly 08:20 to 11:00 London, not the first minutes of the open.

I built the same kind of table but with a twist that exposed something my average-based view had missed. Instead of the median for each minute, I also kept the 90th percentile for each minute, and the picture was quite different. The median looked fine around 14:30 London, yet the 90th percentile was several times larger, because on a few days each month the spread jumped on a data release and dragged the entire window up.

That made me split days into ones with a scheduled release and ones without. On release-free days, the window was one of my cheapest of the session. On release days it was the worst. Having a calendar column in the same spreadsheet turned a vague "avoid 14:30" into a specific rule: trade it only when the calendar is empty.

The other detail is day of the week. Friday afternoons were consistently wider for me than Tuesday afternoons, so I trim my size after 15:00 London on Fridays. It's not dramatic, but it came straight out of the data.