A little more than a year ago, an old friend made a promise that wasn't kept.
Markets had settled into a congestion zone, and everyone was trying to figure out which way the pattern would break. On-balance volume, or OBV, whispered that the break would come to the upside. A Trader's Corner article found at this link included a chart showing how OBV had broken to the upside. Below is that chart, including the original annotations.
Annotated Weekly Chart of the SPY as of June, 2007:
In this instance, OBV didn't prove to be a trustworthy friend.
Annotated Weekly Chart of SPY:
Am I ready to throw away our friendship? No way. On-balance volume, like all indicators and overlays, can often provide useful guidance. Sometimes, the suggestions offered by OBV, like those of any other usually trustworthy friend, just don't work.
Why would OBV normally be trustworthy? OBV calculations require two types of input: volume and price. Many favored indicators require inputs related only to price action, so they're not truly independent of the price action seen on the chart. Volume provides an independent measurement. Volume can be high or low on up moves; high or low on down moves. Huge volume can result in a doji with a close rather near the open or in a large-range day.
Therefore, many market pundits such as those employing volume/price-spread analysis a la Tom Williams watch pure volume patterns closely. Many others look for indicators that include inputs related to volume and price action. The Money Flow Index (MFI) and OBV satisfy the needs of some of those seeking inputs independent of each other.
To calculate OBV, each day's volume is either added to or subtracted from a running cumulative total. If the closing price was higher than the previous day's, the volume is added; if lower, the volume is subtracted. Introduced by Joe Granville in a 1963 book, GRANVILLE'S NEW KEY TO STOCK MARKET PROFITS, the indicator operates on the theory that volume tends to lead price action.
Due to the way OBV is tabulated, a rising OBV indicates that more volume is produced on up days, while a declining OBV indicates that heavier volume occurs on down days. The actual value of OBV may not be as important as its direction. The general idea is that the OBV and price action should be headed the same direction. When they're not, the divergence suggests that big money might be either accumulating or distributing stock. Such divergences alert traders that price may break out in the direction the OBV is diverging.
Or so it works most of the time.
Annotated Daily Chart of the SPY:
So, what happened in June, 2007? Why did OBV give a false signal?
Perhaps it didn't. Prices did break higher for a time before the steep decline. Perhaps the fault lay with the default settings chosen for the OBV. There's no right or wrong setting for an indicator, but generally, the shorter the setting, the quicker the signal but the more likely than signal is to be false. Traders employing a signal built on five periods might see an earlier signal, particularly important to traders seeking a leading indicator, as many consider OBV to be, than one employing a signal built on twenty periods. The tradeoff for the earlier signal is more signals, some of which will be false.
Choosing between faster but less reliable signals and smoother but more reliable ones always presents traders with a conundrum. Since I tend to use any indicator as a warning to start preparing my what-if plans and not for trading signals, I tend to opt for the quicker but less reliable option. Those who trade based on indicator signals alone, not a wise choice in my view, might prefer the longer setting and the more reliable signals.
Perhaps the problem didn't derive from the settings chosen but from the interpretation of the signal. For example, two new green trendlines can be added to the above chart. Which one describes the real situation? The chart was originally snapped on 8/29/08, so that we could benchmark what was being seen at that time and compare it to what happened by the time this article was published. My charts show a red slider on the bottom when they've been scrolled back, so it's always easy to check whether the chart has been scrolled back or was snapped at the time being covered.
Annotated Daily Chart of the SPY as of 8/29/08:
Adding a second indicator such as the RSI can help clarify the interpretation. RSI also broke below the dark green trendline but has not yet broken above the light green one, hinting that the correct interpretation might be that of an OBV breaking below support rather than up through resistance. The interpretation that seems most likely here is that the OBV broke through support and has now risen into a possible "kiss goodbye" test, with the former support now holding as resistance. Perhaps something similar would have been found if we could go back to last June and watch OBV and RSI as the lines were originally played out and not as the subsequent action repainted them.
RSI happens to be one of those indicators derived from price alone. Combining OBV with volume might provide some clues.
Annotated Daily Chart of the SPY as of 8/29/08:
Some will argue that volume almost always decreases in the summertime, and decreasing volume during the summer months proves nothing. That's a view that's been espoused recently, for example, but other summer periods sometimes disprove that theory. So, these charts snapped as of 8/29/08 were suggesting some caution about what happens next. Depending on what happened with OBV, whether it turned down from that "kiss goodbye" test or continued higher, my old friend OBV was perhaps whispering that a downturn could be coming.
Annotated Daily Chart of the SPY as of Midday 09/05/08:
Who knows? Maybe last year, the OBV was telling the right story and I wasn't
listening closely enough. At any rate, I'm not ready to give up this friend, who
tells the tale independent of what price might be doing.