[Price-Volume Volatility]: Anyone ever seen historical testing of volume breakouts and the corresponding increase in price? Several indicators use a ratio of volume change and price change, in various combinations, to indicate the strength or driving force behind a price move. Directly or indirectly, these assume there is a linear relation between the volume change and the price change, only, intuitively, we know this is not the case. For example, if the volume was up 50%, which drove a $2 increase in price, then another time the volume was up 100%, which drove a $4 increase in price, we know that if the volume is up 2000% it's certainly not likely to drive an $80 increase in price-even if it was a completely bullish day (open=low, close=high). At some point DURING THE BREAKOUT DAY (and this is important), further increases in volume have a lesser impact on the price. However, on the next day, the same volume that occured during the period of price stalemate yesterday can indeed drive prices further up. I'm wondering, if we looked at the volatility change of the volume, versus the volatility change in the price (during individual trading days), would this be a better way to judge the staying power of a rally, and perhaps be a better indicator of an impending pause or pullback? Or maybe even a signal for an impending rally similar to Bollinger bands (which are mean deviation,price only). The prototypical breakout shows increasingly higher volume the first few days, then tapers off. The price change doesn't generally change in direct proportion though. If the price-volume characteristics followed a standard normal distribution (and this is a big "IF"), perhaps the relationship between the volume change and price change would be better understood, and therefore better predicted. Any volume change beyond 2.5 to 3 standard deviations during a given day would, in theory, have little further impact on the price for today's session (because this represents 90+ percent of the area under the standard normal curve). Since we're on a LOG scale now, it might? be reasonable to expect a 1:1 relation between the standard deviations change in volume versus the standard deviations change in price. No, it won't be exactly 1:1, and we can only estimate the amount of volume due to supply and the volume due to demand, but it might be worth looking into. If anyone's interested, try doing an independent experiment, and we'll consolidate our observations later.
dh |
