The Composite Breadth Model has yet to trigger bearish as stocks recovered on Monday and closed strong. Despite this reversal, the technical damage has already been done to several big sectors and key industry group ETFs. Even so, the S&P 500 EW ETF (RSP) managed to hold its December low on a closing basis, as did seven of the eleven sectors. This is why the Composite Breadth Model has yet to flip. Today’s commentary will quantify reversal days and show that a recency bias may be in play now. We will then look at the new trend signal in SPY and the RSI dip below 30.
Value Sectors and RSP Hold Up
The chart below shows SPY and four sector SPDRs breaking their December lows, on a closing basis (red lines). The Technology SPDR (XLK), Consumer Discretionary SPDR (XLY) and Communication Services SPDR (XLC) are big sectors driven by large-caps (MSFT, AAPL, NVDA, TSLA, AMZN, GOOGL, FB). The Healthcare SPDR (XLV) also closed below the December low. These are the lagging sectors or the sectors leading the way lower. XLK, XLY and XLC account for 50% of the S&P 500. XLV accounts for 13.22% and is the second largest sector.
On the bright side, seven of the eleven sectors and the S&P 500 EW ETF (RSP) did not break their December lows on a closing basis (green lines). XLRE and XLB are very close, while XLI and XLF are just above. XLU, XLP, and XLE are well above their December lows and holding up the best. In general, these seven sectors represent the value side of the market. XLK, XLY and XLC represent the growth or high beta end of the market. Value is leading growth and the heavy weighting of growth is weighing on the S&P 500.
ETFs with Massive Intraday Reversals
SPY and several ETFs forged massive reversals on Monday. Dozens of ETFs were down over 5% at some point during the day and several rebounded to close positive on the day. The Semiconductor ETF (SOXX) was down 5.38% at its low and closed up 1.14% on the day. The Oil & Gas Equipment & Services ETF (XES) was down 5.37% and closed up 2.49%. The biggest intraday reversals were in four ARK ETFs (ARKK, ARKW, ARKG, ARKF). Each of these were down more than 7% and closed higher on the day. Their intraday swings were more than 10%. Hmmm, such big swings do not seem normal to me.
Reversal Days Through the Years
The S&P 500 SPDR (SPY) was down some 3.92% at its low and rallied back to close in positive territory (up .4% on the day). This is a MASSIVE swing. I looked at prior occasions when SPY was down more than 2% and closed positive on the day (35 times since 2000). The chart below marks these reversal days with the green arrows. There were quite a few during the advance in the late 1990s. There were a few during the two bear markets and big clusters when the market tried to find a bottom (late June to October 2002 and October 2008 to February 2009). These were some seriously chaotic months. Since the low in March 2009, these reversal days tended to marked lows. Note, however, that these reversal days required a 2-3 month “stabilization” process in 2010 and 2011.
The indicator window shows the percent decline from a high with the red bars showing when a decline exceeds 10% (peak to trough on closing prices). SPY has yet to decline more than 10% from its high. The long red arrows on the price chart mark a decline greater than 20%. The 20 plus percent declines in March 2001 and July 2008 led to bear markets, but the 20 plus percent declines in December 2018 and March 2020 led to spike lows. Thus, the 20% rule for a bear market worked from 2000 to 2010 and did not work since. Note that the S&P 500 declined 19.78% from high to low in December 2018.
There are clearly two market periods at work. Reversals days worked pretty well from 2010 to 2021, but did not work well from 2000 to 2010 (red zone). They also worked pretty well from 1996 to 1999 (a strong bull market).
A Recency Bias at Work?
Let’s put these reversal days to the test for the two periods. The table below shows results for the last ten years (2010 to 2021). There were just eight such reversal days (the ninth one was Monday).
As the results above show, the reversal trade worked well from 2010 until 2021. It did, however, take 2-3 months to stabilize after the May 2010 flash-crash reversal day and Aug 2011 European debt crisis reversal day. I am measuring the average gain and loss based on buying the open after the reversal day and selling N days later (5 to 65 days). There is a clear bullish bias throughout. The last line shows the averages with an 85% win rate. Notice that the win rate jumps when moving out to 15 days or longer.
Before getting too excited, let’s look at performance for the N day periods if we bought any time. In other words, simply buy the open, hold N days and sell the open after N days. This is the base case. A strategy with an edge should be able to beat this base case. The five day holding period was up 61% of the time. This number hit 70% when the holding period extended to 25 days and moved into the mid 70% area as we hit 55 days. Thus, there is already a positive bias during this period and this bias gets stronger with longer holding periods.
Now let’s look at the period from 2000 to 2010. The win rate ranges from 31 to 50 percent and averages just 41 percent (last line). This means trying to catch a reversal did not work 59% of the time. The last line also shows that the average loss was greater than the average gain. Simply put, reversal days did not work during this period.
We can also see that this period was difficult in general. Simply buying and holding for any N day period had a win rate greater than 50%, but the highest win rate was 57%. These win rates are much lower than the last ten years. The average loss for each N day holding period was often greater than the average gain. As the averages show on the last line, the average of the average gain was less than the average of the average loss.
SPY Trend Signal and Oversold Condition
The last chart shows the Trend Composite turning negative for SPY. This bearish signal reverses the bullish signal from 21-July-2020. This advance extended from 325 to 440 for a 35% gain. Not bad. Truth be told, bullish signals work much better than bearish signals when it comes to stocks. In other words, the odds or an advance are greater after a bullish signal than the odds of a decline after a bearish signal. Nevertheless, I view the current decline as an uncontrolled descent and atypical for a correction within a bigger uptrend. It looks more like a trend changing event. At the very last, the market has been broadsided and it could take a few weeks or month to stabilize.
A trend signal often triggers when price becomes extremely oversold. Note that RSI(14) moved below 30 for the 33rd time since 2000. Consecutive readings below 30 are not counted twice. An oversold condition, even in a downtrend, can give way to a bounce so I would not be surprised to see a bounce. Longer term, the inability of RSI(14) to hold above 30 is more bearish than bullish because it shows strong downside momentum that is more associated with downtrends.
