A reader recently emailed to ask for an example of what I had called "relational reasoning" in trading. Here's a simple example from my own trading:When traders and investors are relatively optimistic about economic growth, they will favor Consumer Discretionary stocks (XLY) over Consumer Staples issues (XLP). Similarly, when there are concerns over economic slowdown and recession, fears regarding a pullback in consumer spending lead portfolio managers to seek the safety of staples over discretionaries.
Accordingly, we can view the relative performance of XLY and XLP as a kind of sentiment measure. As you can see in the chart above, drops in this ratio have tended to correspond to buying opportunities for SPY as negative sentiment about the economy is overdone. Similarly, peaks in the ratio of XLY:XLP have corresponded to market peaks in SPY.
Interestingly, we are at relatively modest levels of XLY:XLP despite the recent rise in stocks.
As a specific illustration, going back to April, 2004 (N = 860 trading days), when XLY:XLP has shown a positive change over a five-day period (N = 445), the next five days in SPY have averaged a gain of only .04% (239 up, 206 down). When XLY:XLP has fallen over a five-day period (N = 415), the next five days in SPY have averaged a gain of .33% (255 up, 160 down).
When the ratio of XLY:XLP has been relatively strong (top half of its distribution since April, 2004), the next 20 days in SPY have averaged a gain of .34% (262 up, 168 down). When the ratio of XLY:XLP has been relatively weak (bottom half of its distribution), the next 20 days in SPY have averaged a gain of 1.20% (306 up, 124 down).
When XLY:XLP has been < 1.4 (N = 287), the next 20 days in SPY have averaged a gain of 1.64% (220 up, 67 down).
In short, the XLY:XLP ratio seems to behave like many equity sentiment indicators: as an indication of excess optimism and pessimism that can be viewed in contrary fashion by savvy traders.
RELATED POSTS:
Sentiment and Short-Term Cycles
What Drives Sentiment
Trading With Sentiment Bars
.
Posts
