Market Breadth Primer: Market Breadth Analog Forecasting Method
Classic approaches to the interpretation of market breadth data are mainly based on signal generation and recognition of outlier events. The reason for market participants to do that is obvious – the indicator got to produce something of immediate value to be considered as useful to the creator. What can be more convincing than having tradable signals generated from the breadth data to “prove” its usefulness?
The problem with the classic approaches is that it is not harvesting the most important information provided by the market breadth data – an objective reading of the state of the market through a specific measurement method. Individually, these market breadth data do not provide a lot of useful information for immediate application. However, when they are taken into account together within a model, they can be used to produce very powerful forecasting results similar to what we do with atmospheric data like barometric pressure, humidity and wind speed in weather forecasting.
The State of The Market As Described By Market Breadth
Price based analysis on a market suffers a potential problem that the scale of the data can be changing forever. A good example is the Dow Jones Industrial Average. It started at a very small number and over the year it expanded to more than 18,000. It is difficult to apply analytical method on the data and expecting to get consistent results. Some form of normalization has to be done to either the price data or the technical indicator so that you know you are not distorting your analysis results.
With market breadth data, we can design them so that they are range bounded for easy comparison across long period of time. A good example is the use of Advance Issues Percent over Advance Issues the raw data. This helps us to identify, historically, which trading days are having the same level of advance issues even though the price levels can be completely different.
Furthermore, it is possible to classify the market breadth data in a fuzzy way so that it is possible to look at them in a more abstract level. Using the same example Advance Issues Percent (AIP), we can define it with 5 different states:
- Very Bullish Day: AIP > 75%
- Bullish Day: AIP > 55%
- Neutral Day: 55% > AIP > 45%
- Bearish Day: 45% > AIP
- Very Bearish Day: 25% > AIP
Obviously, some trading days would be borderline between 2 states so they should be labeled as such. The goal is to capture the state of the market, not trying to optimize the readings for specific trading system or model.
The Changes Among The States Can Provide The All Important Context
Given any single trading day, we would have a collection of states from multiple market breadth data describing the trading day. That, however, lacks a key ingredient, the context, in a forecasting model. The context in this case is a description of how the market arrives at the current state.
For price analysis, the context is often driven by the description of the trend. For example, we may have a 3 weeks of rising moving average readings up to the current point in time. The way the moving average readings changed provides a context for us to understand or visualize the current situation.
Similar approach can be done with market breadth states. One can look at the trend of the market breadth data and use that as the context. For example, we can look at AIP and identify that it has been rising steadily over a 5 day period. This gives us the ability to compare this specific combination of the current state and context against more specific historical occurrences.
The Analog Forecasting Method
Given a collection of market breadth data, the combination of their currents states and the context, is a unique signature summing up the near term market conditions up to that point in time. In other words, we’ve found a way to represent the current market conditions with our market breadth based summary instead of the price data and its derivatives. With this signature on hand, we can use the analog forecasting method to generate useful projections based on the historical occurrences of the signature.
Analog forecasting is not in general applied as a trading model. Instead, the analog method gives us an idea of the expected outcome similar to what weather forecast is doing for us. Analog forecast can tell us if the overall near term is more bullish or more bearish. It can also figure out if a potential trend change is coming. But most important of all, it can tell us whether the risk level of the market is rising or falling.
By knowing ahead of time how the market is likely to behave in coming 3 to 4 weeks, you can incorporate the information into your investment decisions quickly without guessing what the market breadth data really means. Frankly, I am not a fan of having tons of squeegee lines on a chart and then making judgment calls on what the breadth data is telling me. If financial institutions need exact interpretation of this information, retail traders actually need this even more due to their limited resources and investment capital.
High Precision Forecasting
So is it possible to produce high precision forecasting models based on market breadth data? Of course.
To generate higher precise forecast suitable for day trading and very short term trading, we can no longer rely on end of day market breadth data. Specialized real-time market breadth data has to be collected on intraday basis so that much more refined analog can be produced. Due to the complexity in processing tick by tick data on thousands of symbols in real-time, let along the need to also summarize them at the same time, the technology for doing this is only available in highly specialized trading software.
Obviously, the unique ability to dissect the market data in real-time leads to the possibility of very powerful real-time market timing tools. The requirements and the types of market breadth data to be collected for real-time usage is very different from those we use for longer term planning. I will write about them in the future when I get the time.
This is just a short introduction to the concept of modernized market breadth analysis. It is what I work on over the past 20 years. Many of my research results were done for large institutions helping them to stay on the right side of the market. The strategic advantage of using market breadth analog method is very clear in terms of better risk anticipation from objective projections.
I choose to write this introduction in plain English instead of throwing in all kinds of formulas and equations in here because I would like to see more people understanding that the financial markets are not as unpredictable as they think. Market risk can be anticipated. By making some of the analog forecasting models available, I believe it can help many individual investors doing better with their investments.