Monthly Archives: June 2015

Hour Compositions: Differences in hour minute start

One of the problems I encountered when I was first getting into research was the idea of the “daily” candle. We know that because the market is 24 hours, there are many ways to draw the daily candle and it varies from broker to broker. If you strictly trade candle patterns, this can be an issue. When I was tracking the hour of the day that the highs and lows occur, some hours were more favored depending on what hour I used as the “first hour” of the day.

In the long run, this problem kind of disappears as one focuses less on the candles and more on the overall price direction because it requires one to take enough candles into account that the overall picture is the same regardless of which view you use. But when one is looking specifically at a single candle, I can see this type of analysis making a difference.

The follow up with this is that when you look at a micro level, is this also true? Is 00:00-00:59 different from 00:20-01:19?

hour composition

Yes. The bin here is representing the 10s digit in the hour, i.e 00 means an extreme occurred between 00 and 09, or the first 10 minutes of the hour. Movement is movement and patterns that occur in higher time frames are present in lower time frames, but time does play a difference when one looks at things minute by minute. Within a span of 60 minutes, it is more difficult to create more “complex” structures. As a result, the time that the hour starts on is more likely to contain a price extreme because price is more directional in these frames.

Interestingly, a pattern that I noticed on altered charts makes a big difference depending on the hour reference. If one assumes random movement, the probability for a continuous 20 minute segment to contain at least one of the extremes to be about 33% (with each 10 minute span being 1/6 of the total, 1/6 + 1/6). Perhaps, given the one-dimensional movement of lower frames, it may not be crazy to see a number like 50%. However, the number can be pulled to quite the extreme, and the probability for a specific 20 minute segment to contain at least one extreme can be over 75% on some cases depending on the hour reference and the minute span.