Understanding the Indicators and Signals in the Index Breadth Model – Diving Beneath the Surface

Video was updated in November 2020 and includes Thrust Model.

This article covers the Trend and Thrust Breadth Models in detail. We start with an overview of breadth and explain why the S&P 500 is my preferred index for measuring broad market breadth.  Attention then turns to the individual indicators and the key levels for generating signals. These models use a “weight of the evidence” approach to broad market timing with signals that can be quantified and tested. As such, we will also show model performance since 2000 with charts for the equity curve and drawdowns. This is a long article with the first part covering the Trend Breadth Model  and second covering the Thrust Breadth Model.  

Trend Breadth Model Highlights

  • The Trend Breadth Models use five breadth indicators for each index
  • The 10-day EMA of AD% identifies thrusts in participation
  • Bullish with cross above +30% and bearish with cross below -30%
  • %Above 200-day SMA measures long-term trend participation
  • Bullish with cross above 55% and bearish with cross below 45%
  • %Above 150-day SMA measures medium-long term trend participation
  • Bullish with cross above 65% and bearish with cross below 35%
  • %Above 100-day SMA for medium-term trend participation
  • Bullish with cross above 75% and bearish with cross below 25%
  • High-Low% quantifies leaders (new highs) versus laggards (new lows)
  • Bullish with cross above +10% and bearish with cross below -10%
  • Indexes are S&P 500, S&P 100, Nasdaq 100, S&P 400, S&P 600
  • Models are bullish when at least 3 of 5 indicators are on bullish signals
  • Models are bearish when at least 3 of 5 indicators are on bearish signals
  • A 5/200 day SMA cross is added to $SPX model for trend confirmation
  • $SPX model is bullish when breadth is bullish and SPX 5-day > 200-day
  • $SPX model is bearish when breadth is bearish OR SPX 5-day < 200-day
  • The S&P 500 model is the most important of the five models
  • Model is designed to stay on the right side of the big trend
  • This is a “trend” model and signals will lag, just like any trend indicator
  • The model outperforms buy & hold with much better drawdowns

Introduction

Stock market breadth plays an important role in broad market analysis because it reflects what is happening under the surface. Price action, in contrast, shows us what is happening on the surface, which is often dominated by large-caps. Even though price action is the single most important factor, I would place breadth a close second when it comes to broad market analysis.

The S&P 500, for example, has 500 moving parts (stocks) and index levels do not always reflect what is happening inside. Breadth indicators allow us to look inside and take the vital signs for the stock market. In short, breadth indicators measure the degree of participation and can help us identify significant shifts within the market.

Breadth indicators are key to determining the overall market environment: bull or bear.  The Trend Breadth Models use five breadth indicators across five indexes to analyze and assess the broad market environment. I want to trade the long side and look for bullish setups when the broad market environment is bullish.  Conversely, I want to be defensive and selective when the broad market environment is bearish. More aggressive traders can consider short positions and bearish setups.

Picking the Benchmark for Breadth Indicators

Most breadth charts use indicators based on NYSE or Nasdaq stocks, not both. Breadth indicators based on just one exchange are seriously lacking. For example, NYSE breadth indicators exclude Apple, Microsoft, Google, Facebook, Cisco, Amgen, Celgene, Amazon, Starbucks, Comcast, Netflix, Ross Stores, Expedia, Costco and many other key Nasdaq stocks. Personally, I cannot use a broad market indicator that does not include these stocks.

As the name implies, “breadth” indicators should capture a wide swath of the stock market. This is why I derive my core breadth indicators from the broad indexes: S&P 500, S&P MidCap 400, S&P SmallCap 600, S&P 100 and Nasdaq 100. The first three are representative of the entire stock market because they capture large-caps, mid-caps, small-caps, Nasdaq stocks and NYSE stocks.  Nasdaq 100 breadth captures the performance for large-cap techs and high-beta names. Also note that some 82 stocks in the Nasdaq 100 are also part of the S&P 500 and these 82 stocks account for around 40% of the S&P 500.

The S&P 500 is the single most important index of the five. As such, it serves as the base-case for breadth and will be used for the following examples. The S&P 500 is the most widely followed index for the US stock market and the most use index for benchmarking performance. $SPX stocks are on the institutional radar, cover all eleven sectors and represent dozens of industry groups. Furthermore, around 2/3 of component stocks come from the NYSE and 1/3 from the Nasdaq. It covers all the key bases.

We can also use a process of elimination to narrow our focus. The Nasdaq 100 and S&P 100 only have 100 stocks with a concentration of large-caps, which makes them a bit narrow for broad market breadth. The S&P MidCap 400 and S&P SmallCap 600 are broad enough, but mid-caps and small-caps trade more erratic than stocks in the S&P 500.  I could use the S&P 1500 or even all five indexes, but experience shows that small-caps tend to muddy the waters with their erratic price behavior.

Even though I am focused on the S&P 500 for breadth signals, I also monitor signals in the other four indexes. Signal differentiation can tell us when one index/group is leading or lagging. For example, S&P 600 breadth lagged for most of 2019 and Nasdaq 100 breadth led the charge off the April 2020 lows.

The following breadth indicators and models were developed using Norgate data and Amibroker software. There is nothing wrong with StockCharts’ breadth data, but I need data going back further than 2010 and signals that can be quantified using testing software. Furthermore, Norgate breadth indicators are based on data that is not adjusted for dividends, which is my preference. The numbers may be different than those seen on StockCharts because of calculation differences and dividend-adjustments. See this article for details.

Breadth Thrusts using AD Percent

The first indicator is based on Advance-Decline Percent, which is the percentage of net advances ((advances – declines) / total issues).  This indicator shows us the degree of participation, up versus down, on a daily basis. Readings above +80% show strong upside participation, while readings below -80% show strong downside participation.  

AD Percent is quite choppy because up and down days alternate on a regular basis. I apply a 10-day exponential moving average and look for breadth thrusts. A bullish breadth thrust occurs when the 10-day EMA exceeds +30%, while a bearish thrust triggers with a move below -30%. A move above +30% reflects broad upside participation over a short period of time. This breadth thrust is like a rocket lifting off and can signal the start of an extended advance. Conversely, a move below -30% shows broad downside participation in a short period of time and this can lead to further downside. 

The example above shows a bearish breadth thrust triggering in late February 2020 as the 10-day EMA of Advance-Decline Percent plunged below -30%. The bottom window shows the signal line hitting -1 for confirmation. A bullish breadth thrust triggered in late April with a move above +30% and this remains the active signals.

I chose ten days because I am looking for a powerful short-term move to signal a significant change in participation. 20 days is a bit too long and 5 days is too short.  Marty Zweig also used a 10-day EMA for the Zweig Breadth Thrust. Keep in mind that these settings are my personal preference and one size does not fit all trading styles. These are just the settings I have settled on after years of chart watching.

Measuring Strong Trends with High-Low Percent

52-week highs and 52-week lows are chart milestones that occur during strong trends. At the risk of stating the obvious, stocks forging 52-week highs are in strong uptrends and leading, while stocks hitting 52-week lows are in strong downtrends and lagging. The difference between the two tells us if more stocks are in strong uptrends or strong downtrends. This is valuable information on the market environment.

High-Low Percent is the percentage of net new highs ((new highs – new lows) / total issues). High-Low Percent is positive when new highs outnumber new lows and negative when new lows outnumber new highs. If there were 67 new highs, 5 new lows and 500 total stocks, High-Low Percent would be +12.40% ((67 – 5)/500 = .1240).

As opposed to breadth thrusts, which are often leading indicators, High-Low Percent acts more like a trend-following indicator because it takes time to reach such milestones.  High-Low Percent is less sensitive than AD Percent because we do not see big positive-negative shifts from day-to-day. New highs can dry up when the market falls, but we rarely see a sharp surge in new lows with a one-day decline.

The chart above shows High-Low Percent for the S&P 500 in the middle window and the signals in the lower window. High-Low Percent whipsawed in August 2019 with a move below -10% and above +10%, but stayed bullish as new highs remained fairly strong until February 2020. A bearish signal triggered in late February with an expansion in new lows and this signal remains active. It would take a move above +10% to reverse this signal.

Measuring Trend Participation with 3 Indicators

The humble moving average remains one of the most popular methods for trend identification and the breadth models employ three different moving averages. A 200-day SMA covers around 9 months and represents a long-term trend (at least in my book). A 100-day moving average covers three and a half months and represents a medium-term trend. A 150-day SMA sits in the middle and represents a medium to long term trend. Breadth turns bullish as more stocks move into uptrends (above these moving averages) and bearish as more stocks move into downtrends (below these moving averages).

The bullish and bearish thresholds widen in uniform fashion as the moving average period shortens. The Percent Above 200-day SMA turns bullish with a move above 55% and bearish with a move below 45%. The Percent above 150-day SMA turns bullish with a move above 65% and bearish with a move below 35%. The Percent above 100-day SMA turns bullish with a move above 75% and bearish with a move below 25%.

The chart above shows all three indicators with their bullish and bearish thresholds. All three turned bearish in October 2018, bullish in February 2019 and bearish in February 2020. All three triggered bullish in early June 2020, but these signals did not last as two of the three flipped back to bearish a few days later.

Putting it All Together

majority for a weight of the evidence approach. The 10-day EMA of Advance-Decline Percent is the early warning indicator that often triggers first, as it did in January 2019 and April 2020. The High-Low Percent and %Above Moving Average indicators often trigger later and represent follow through signals to the initial thrust.

The chart above shows the S&P 500 SPDR, the SPX Trend Breadth Model and the 5/200 day SMA crosses. The strategy goes long when the breadth model is net bullish AND the 5-day SMA is above the 200-day SMA. The strategy exits when the breadth model turns net bearish OR the 5-day SMA moves below the 200-day SMA. The red and green arrows on the price chart show some recent signals. There were some timely signals (March 2016, October 2018, February 2019, February 2020) and some whipsaws (June 2019, June 2020). 

Quantifying Performance and Adding Octane

Now it is time to test this model and quantify performance. The testing period runs from 1/1/2000 to 11/1/2020, which covers 20+ years, two full market cycles and some serious hiccups.  Buy & Hold for the S&P 500 SPDR (with dividends) is used as the benchmark. As such, the results when trading SPY also include dividends.

The start and end date for a backtest affect the overall results. I could have padded the results by starting at the end of a bear market, such as 1/1/2003. However, I think it is important to include two full market cycles (bull/bear) for more realistic results.

The first table shows the benchmark results for Buy & Hold and the 5/200 day SMA cross. Trading SPY, the Compound Annual Return was under 6% for both and the Total Profit Percent was around 225%. This means the portfolio was up 225% in total. Buy & Hold produced three big drawdowns: -55% in 2002, -47% in 2008 and -33% in March 2020. The Maximum Drawdown improved to -21% when using some basic market timing, such as a 5/200 cross.

The next table shows results when trading SPY,  without and with the 5/200 timing filter.  Trading without a timing filter means taking the breadth signals on their own. The Compound Annual Return was just 4.5% and the Maximum Drawdown was -43%, both of which are unacceptable.

Adding the 5/200 timing filter improved returns and drawdowns. Buy when the model is net positive and the 5-day is above the 200-day for $SPX. Sell when the model turns net negative OR the 5-day crosses below the 200-day for $SPX. The Compound Annual Return hit 6%, the Maximum Drawdown improves to -18.6% and the total profit exceeds 240%. Notice that total profit percent is higher than Buy & Hold even though this strategy is invested just 62% of the time (exposure). The 5/200 cross in the first table was invested 71% of the time.

Admittedly, the SPY returns are nothing to write home about. Nevertheless, keep in mind that timing strategies are designed to avoid the drawdowns of Buy & Hold and generate similar returns. Funds can also be deployed elsewhere when the model is bearish.

Chartists looking for more octane can consider QQQ, which has a much higher beta than SPY. In other words, buy/sell QQQ using the exact same signals from the SPX Trend Breadth Model. The table below shows the Compound Annual Return exceeding 8%, the Maximum Drawdown improving to -18% and the Total Profit Percent surging to 414%, despite a lower Win%.

I tested the Nasdaq 100 Trend Breadth Model signals with QQQ, as well as the Trend Breadth Model signals for the other indexes and their respective ETFs. The S&P 500 Trend Breadth Model with QQQ produces the best results (Compound Annual Return and reasonable Drawdown). In particular, the drawdowns were higher (>25%) when using the Nasdaq 100, S&P MidCap 400 and S&P SmallCap 600 models with their respective ETFs (QQQ, MDY, IJR/IWM). S&P 500 stocks are more stable and better suited for breadth indicators.

Trades, Equity Curves and Drawdowns

The future does not always look like the past, but backtest results can help set expectations going forward and prepare us. In addition to expected returns and drawdowns, I am also looking at trade frequency, time held, average gain and average loss. The next image shows all the long trades since 1/1/2000. Of the 14 winners, the longest trade lasted 648 trading days (2016 to 2018) and the average was 151 trading days. Of the 7 losers, the longest losing trade lasted 100 trading days and the average was 51 days.

The next image shows the equity curve in green and the buy-hold SPY equity curve in blue. The strategy outperformed Buy & Hold almost all of the time, but hit a speed bump in early 2020. Total equity stands at 3,458  for buy & hold and 3,416 for the strategy. Zeros do not matter because these equity numbers are relative.

The path to these returns is what really matters here. The strategy driven by the model shows a much smoother path higher, while Buy & Hold shows at least three big setbacks.

The next image shows the drawdowns with the biggest in early 2020 (-18.6%). Prior to the March crash, maximum drawdowns were limited to the 12-14%.  

The next images show the equity curve when trading QQQ using the signals from the S&P 500 Trend Breadth Model. The equity curve surged from March 2016 until January 2020 and exceeded the 5,000 mark. Note that Buy & Hold for QQQ returned 6.3% per year during this period and the Maximum Drawdown was a whopping -83%. Higher returns entail more volatility and this is reflected in the drawdowns, which are more frequent.

Thrust Breadth Model Highlights

  • The Thrust Breadth Model uses three breadth indicators
  • The 10-day EMA of AD% identifies participation shifts
  • Bullish with cross above +30% and bearish with cross below -30%
  • %Above 50-day SMA measures trend participation
  • Bullish with cross above 85% and bearish with cross below 15%
  • %Above 20-day SMA measures short-term trend participation
  • Bullish with cross above 90% and bearish with cross below 10%
  • Model is bullish when at least 2 of 3 indicators are on bullish signals
  • Model is bearish when at least 2 of 3 indicators are on bearish signals
  • An SPX moving average cross is NOT used for trend confirmation
  • Model identifies sharp and material changes in participation
  • The Thrust model picks up signals that the Trend model might miss

The Thrust Breadth Model is designed to identify sharp changes in participation. These changes are considered “thrusts”, which can be on the upside or downside. A thrust reflects a material spike in participation that can signal the start of an extended move.

The Thrust Breadth Models are based on three short-term breadth indicators: 10-day EMA of AD%, %Above 50-day SMA and %Above 20-day SMA. We already covered the first indicator because it is part of the Trend model. %Above 50-day SMA triggers bullish/bearish with a thrust above/below 85/15 percent. %Above 20-day SMA triggers bullish/bearish with a thrust above/below 90/10 percent.

The Thrust Models turn bullish when at least two of the three indicators are bullish and bearish when two are bearish. For example, a 10-day EMA of AD% surge above +30% and %Above 50-day SMA thrust above 85% signals a material shift in upside participation. This is like a rocket lifting off and can lead to further gains.

The example above shows the S&P 500 Thrust Model turning negative in late February and positive in late April. The red ovals show the three bearish signals in late February. %Above 20-day SMA and 10-day EMA of AD% were first to trigger. These two also led the recovery thrusts in April and %Above 50-day SMA followed in May  (green ovals).

Performance for the Thrust Model

The next table shows performance when trading SPY and QQQ using signals from the S&P 500 Thrust Model. There is no timing filter for this model because thrust signals often occur when the 5-day SMA for $SPX is below the 200-day SMA. Trading SPY, the Compound Annual Return was 6.5% and the Maximum Drawdown was -25%. The two big drawdowns (~25%) occurred in March 2009 and late 2012. The drawdowns since 2014 did not exceed -12%.

As the last line shows, we can also add some octane by trading QQQ using signals from the S&P 500 Thrust Breadth Model.  The Compound Annual Return surges above 9% and the Total Profit Percent exceed 500%, thanks to the big run from March 2016 to January 2018. The Maximum Drawdown, however, increased to 29.66%. Live by volatility and die by volatility.

Conclusions

Identifying the broad market environment is the first step in the investment or trading process because getting the broad market environment right is more than half the battle. The Thrust Breadth Models can provide the early warning signals, while the Trend Breadth Models can capture big trends backed by broad participation. Model signals keep us on the right side of the market and dictate our trading bias. Look for uptrends, leaders and bullish setups when the models are bullish. Hunker down when the models are bearish.

The S&P 500 breadth models are the most stable of the five index-based models. In fact, the S&P 500 models can be used for broad market signals and other ETFs can be used for positions. Chartists can also consider buy/selling a broader basket of ETFs with these signals. For example, buy/sell four equal weight positions in SPY, QQQ, MDY and IJR on bullish and bearish signals. This would insure a horse in the race no matter which group leads. Looks like I have the next subject for a performance test.  

Thanks for tuning in and have a great day!

-Arthur Hill, CMT
Choose a Strategy, Develop a Plan and Follow a Process

Scroll to Top