Hi, I am trying to understand ways to predict large range bars - especially in stocks. Are certain fundamental or technical conditions more likely to result in large range days? For example, some ideas: - Narrow range bars: Given the cyclical nature of the volatility/range, is there is a period of narrow range bars, it would likely be followed by wider range bars - Earnings announcements: not sure about this one Any thoughts?
You cannot predict them. But you can do two things about that: 1) take all days you consider as being "large range" and look for common properties they have. This could be something as simple as making low of the day in the first x minutes. These facts will enable you to construct a trading system which is easy to test. 2) see what things precede such days, this is some data mining task, but be careful, because data mining gives garbage results, i.e. random. But don't look for prediction. Look for conditions where the chance of "large range" is higher. For example, you may see 40 large range days in your data of 500 days. So the "prediction" rate would be 40/500 = 8%. Let's say you found a condition that reduced trading days from 500 to 250 of which 37 were large range days. This would give you "prediction" rate of 37/250 = 14.8% which is much bigger than trying to trade every day, and also every attempt costs commissions. Hope this helps.