forex-scalping.com

Daily pivots mean different things in different sessions

Pivot points are calculated from the previous day's high, low, and close and are plotted on many scalpers' charts. But the definition of the previous day depends on the platform's daily bar, which varies by server time.

A pivot computed from a Sydney-to-Sydney day differs from one computed from a New York close day. Prices may interact differently with each, and a scalper using one broker's pivots may see mismatched levels compared to a peer using another.

Also consider session-specific behavior. Pivots often matter more in the session where most participants use them. For European traders, London hours see reactions to those levels, whereas in the Asian session they may carry less weight.

A fair test is calculating pivots using a consistent day definition and checking how often price reacts near them in each session. Reaction can mean a bounce, a stall, or a sharp break.

If the reaction frequency is not above random in the sessions traded, the pivots may only be cluttering the chart.

Which session shows the most reliable reaction to pivots on the chart in use?

I tested this on my own charts using EURUSD and GBPUSD, with two versions of daily pivots, one from the platform's server time day and one from a fixed 5 PM New York close. On GBPUSD they sat about 12 pips apart on average, which is more than enough for a scalper to be on the wrong side of a level.

For counting reactions I defined a touch as price getting within 3 pips and then moving 8 pips away before returning. Over roughly four months, London session touches of the central pivot reacted around 55 percent of the time, which I'd call slightly better than coin flip. In the Asian session it was almost exactly random, so I removed pivots from my charts there.

The New York close version worked a little better for me, probably because most of the people I trade alongside use it. The lesson I took was to use the day definition that matches the crowd for that pair, then check reaction counts per session before trusting any level. For the sessions where the numbers were random, I just hid the lines.