Hi Tim, the problem that I have with their method is they aren't applying a buying/selling scale factor to the total volume. Never will all volume be strictly buying OR strictly selling. Granted, you can't get the breakdown exactly. It can only be estimated from o-h-l-c data because these are the only data points we have available to us for TA. If you believe that, on average, the open, high, low, and close are representative samples of the trading distribution through the trading day (and I think we have to believe this--otherwise TA based on price data is almost meaningless), then several methods exist for estimating the net buying (or selling) volume on a given day. For example, ((c-o)/(h-l))*v is one estimate. It's a biased estimate because volume isn't uniformly distributed across all prices reached during the trading day. Some experts claim that public buying tends to be at the open whereas professional buying tends to be intraday, so some won't use the open price. Another thing, since they are using close to close data (ignoring daily range and open altogether), they are assigning volume significance to any gaps which exist over the observation window, which doesn't seem valid to me because by definition gaps are "no man's land" (no trades took place). But, as you know, what makes logical sense doesn't always translate into making an indicator better. Only real world trading can determine that.
dh |
