Forecast Accuracy

Most people believe forecasting specific pricing trends, on a day-to-day basis, for the next 90 days is impossible. After all… you have heard and may well believe that the market is purely a “random walk”. When it comes to exogenous events, there is little doubt that almost by definition, these events which can skew the market or a security higher or lower, occur randomly. But… and this is important to understand… the vast majority of pricing trends are far from random and can be very accurately predicted if the proper quantitative analysis is performed on a sufficient amount of historical data.


With our proprietary software algorithms, it is certainly possible to perform accurate forecasts that are highly credible and very actionable; and do so consistently. To understand the concept behind our forecasting algorithms, we have written a document entitled, “The Math and the Theory Behind the CycleIntelligence Forecasts”, that is a very short but informative overview of how our forecasts are constructed. You are encouraged to read the article if you want to understand some of the math and science behind our app.


The concept is simple to grasp but the successful execution of the math to achieve the desired results has eluded mathematicians and scientists for centuries. Buried within historical price data, are virtually unrecognizable repeating pricing patterns. Some of these patterns (we call them “cycles”) repeat with a consistency that cannot be seen with the human eye. For example, we may find that every 82.03 days, the price of a certain ticker moves up 1.05% and that movement can be found to occur over 90% of the time throughout that ticker’s pricing history. This “periodicity” is one of the key elements that our algorithms use to identify these predictive phenomena. After all, if a specific pricing event has repeatedly occurred 90% or more of the time in the past, the odds are very high this event will continue to occur into the future.


But, just finding those cycles that repeat is not nearly enough. Anomalies occur in the data that must be filtered out, such as: noise, which are random movements that occur with a high degree of periodicity but have no discernible predictiveness; exogenous events, which are one-off events that skew the pricing trend abruptly higher or lower; embedded trends that are a function of the overall market and not particularly tied to the specific security; instability of cycles where the cycles tend to appear for a while and then disappear, only to appear again later; and a host of other anomalies that are removed from the data before a final set of cycles is identified.


Once the frequency and amplitude of each time-cycle is defined, each one must be tested to determine its degree of predictiveness. The quantitative analysis must find only those cycles that are mathematically predictive.


After starting with many millions of cycles, the analysis, refinement and filtering algorithms reduce this universe of ‘candidate’ cycles to a handful of cycles which, when combined together, prove to be amazingly predictive and absolutely tradable.


How To Read Time-Cycle Forecasts


Our time-cycle forecasts are 90-day projections of likely pricing trends and movements of any tradable security. Each forecast has the following:


  1. A 21-day, centered moving average, line that should be considered more for changes in trend direction than actual price.
  2. The actual daily forecast price.
  3. The estimated volatility range (Expected Move) of the security. This is called the EM Band.  The EM Band on the Forecasing App represents 1/3 of a standard deviation.
  4.  Bull/Bear Vertical Bands that denote changes in trend.  A green vertical line indicates the beginning of a bullish trend forecast and a red vertical line indicates the beginning of a bearish trend forecast.

Each of these elements of the time-cycle forecast is noted on the chart, below:


How To Use Time-Cycle Forecasts


CycleIntelligence is the world’s most advanced time-cycle forecasting application. The Forecasting App was developed to help identify special situations that can give traders an edge over other traders who rely on only fundamental and technical tools.  The algorithms take a traditional approach to finding cycles and use a ground-breaking technology to determine which of the cycles are the most prevalent and predictive. The final output gives the user an incredibly valuable, 90-day forecast of both trend and price which can be used to maximize profit, adjust risk, and help to eliminate bad trades.


The CycleIntelligence Forecasting App is not meant to be a stand-alone tool nor does it replace the fundamental and technical tools you already use. Our technology was developed to enhance the process you currently use to make your trading more effective and profitable. The following are recommendations on how to maximize your use of the CycleIntelligence Forecasting App:


  1. The Forecasting App is extremely effective at identifying an overall market bias using a top-down approach. For instance, when trading the US Equity markets, we recommend that you look at the major averages, i.e. S&P 500, DJIA, Nasdaq, Russell, etc.; and the time-cycle forecasts for each. If the forecasts for these averages are reasonably in agreement, a much higher level of confidence is achieved as to the likely direction of the markets over the next 90 days. After determining the likely direction of the markets, then run forecasts for the major sectors and industries. The goal is to find forecasts that are supported by the forecasts of the major averages. Finally, drill down and look for charts of individual stocks that mirror the forecasts of their sectors. This process does take time, but, when used with other traditional tools, this will greatly increase the likelihood of profitable trades, with better entry and exit points.

  2. Another method is similar to the top-down approach, but focuses on equities that trade similarly during periods of risk-on and risk-off. If the forecasts of these securities are similar to the “big picture” forecasts, then the potential outcome of trading these securities is greatly increased. Currency traders will want to look at several forecasts to see that a specific currency that shows strength or weakness against a number of different currencies. For instance, those trading the US Dollar will want to run forecasts on several US Dollar currency pairs to determine the overall trend. Forecasting only one currency pair will not accomplish this.

    If the “big picture” forecasts are mostly NOT in agreement, this is a warning that the risk of trading is higher than when the forecasts are more in agreement. The Forecasting App was not developed as a risk-free method for investing. It was developed to find special situations that give traders an edge over the majority of other traders who are using only traditional fundamental and technical tools. The Forecasting App cannot forecast exogenous events. However, after an exogenous event occurs that moves the market, the App can forecast the movement of the "return to normal".

  3. A significant component of the Forecasting App is its ability to predict pricing trend tops and bottoms over the next 90 days. These tops and bottoms are considered, “Turns”; which are the specific time frames that the underlying security is expected to change trend direction. Trends lasting three weeks or greater in length tend to be more reliable and more tradable than those lasting shorter periods of time. These longer trends occur when long, medium, and short cycles tend to coalesce and move in the same direction. The longer trends allow for easier entry and exit of trades by being able to determine in advance the best way to leg in or out of positions. Longer trends typically account for greater potential profit and a higher percentage of winning trades.


In summary, the best way to use this tool is to take your time and look at the market and your securities of interest from many different perspectives by incorporating several time-cycle forecasts with your current trading/investment strategies. The most effective trades are those in which all of the tools that a trader uses, support the trade. Better trades = more profit + less risk. The CycleIntelligence Forecasting App is integral for traders who want an edge and have the patience to wait for the right situation.


How Accurate are the Time-Cycle Forecasts


It is important to understand that forecasting the future is a science but it is not a perfect science. Think of it this way… Let’s assume you trade oil/gas futures and it is hurricane season. A large tropical depression is organizing half-way between the US and Africa. The tracking models show a 40% probability that the storm will go through the Gulf of Mexico oil/gas production region. The storm is approximately 10 days out to sea. Do you see the 40% projection and bet your or your client’s money that oil/gas will spike higher? Probably not… Why? The model shows a 60% probability that the storm will turn north and miss the US entirely. Is the risk too high to place the trade now, when the storm is so far out and not yet fully organized? Probably so.


So, what do you do? Do you stop looking at the forecasts each day or do you just assume the first forecast was all that was needed to make a decision? You probably keep looking at the updates as quickly as they are available from the Hurricane Tracking Center. At some point, the size of the storm and the track of the storm are sufficiently definable for you to make a trading decision. But, you do not trade just on the likely movement and size of the storm alone. No… you take into account the oil/gas demand, inventories and supply of product from sources other than the Gulf among many other considerations. There are likely many contributing factors that you will consider in addition to the forecast track of the hurricane.


This is not overly dissimilar from how you should ‘read’ the time-cycle forecasts. We recommend that you run the forecasts often (at least weekly) and consider the pricing ‘track’ along with all other pertinent information before making a final trading decision. Much like hurricane forecasts… our time-cycle forecasts are not written in stone and can change over time. We believe that using our time-cycle forecasts will significantly improve your win/loss ratio, but if you blindly follow these forecasts without due consideration to other pertinent factors, there will be times when the ‘storm’ changes course at the last moment.


As with all accuracy claims, past performance does not guarantee future results; and, your results may vary considerably from the results of our testing, observations and trading strategies. Interpretation of the results can also be greatly affected by the specific type of trading performed. For example longer term trading strategies could draw a totally different conclusion, regarding the accuracy of the results, than a day-trading strategy.


When reviewing the "Performance Accuracy Observations", below, we use the following qualitative and quantitative criteria:


  1. We rate the performance of a particular forecast statistic as “Excellent” when our trades were very positively impacted by using the forecast either to open a position or close a position with regard to timing and price.

  2. We rate the performance of a particular forecast statistic as “Good” when our trades were positively impacted by using the forecast either to reduce the amount of loss that would have occurred without the reliance of the forecast or the amount of profit generated was better having reacted in accordance with the forecast versus relying on other exit strategies other than the time-cycle forecasts.

  3. We rate the performance of a particular statistic as “Fair” when our trades were not overly impacted, positively or negatively, from the reliance of the forecast.


Below are the opinions of our in-house traders regarding the statistical accuracy and reliability of our time-cycle forecasts. As mentioned above, your results could vary considerably based on your investment and trading methodology:

Statistic Definition Trader Score
First Trend This statistic measures how accurately the time-cycle forecasts accurately predicted both the direction and duration of the price trend from the current date through the next change in trend from either bullish to bearish or bearish to bullish. EXCELLENT
90-Day Trend This statistic measures how accurately the time-cycle forecasts accurately predicted both the direction of the price trend as well as the relative percentage change in price from the current price compared to the price of 90 days into the future. EXCELLENT
Daily Trend This statistic measures the accuracy of the difference between the forecast trend for each day throughout the forecast period, to the actual price movement for that day. EXCELLENT
Price Accuracy This statistic measures how close the daily price forecast came to the actual price throughout the entire forecast period. GOOD
EM Band Accuracy This statistic measures how well the forecast accurately predicted the price within 1/3 standard deviation of normal volatility throughout the forecast period. EXCELLENT
Timing This statistic measures the accuracy of the timing of forecast tops and bottoms throughout the forecast period. EXCELLENT
Trend Slope This statistic measures how accurate the forecast was in accurately predicting how quickly the price trend will change, how steep the trend will be and how long the trend will last. GOOD
30/60/90 Day
Price Accuracy
This statistic measures the accuracy of the forecast with regard to estimating the exact price at the 30th day, the 60th day and the 90th day. FAIR
Convergence and
Divergence Analysis
Many indexes and tradable securities can be identified that have strikingly similar price movement trends. It is important to compare the 90-day forecasts of these securities to determine when the forecasts exhibit more or less convergence (see “How to use time-cycle forecasts”, above.) EXCELLENT
Risk Assessment Whether you are more of a trader or an investor, the fact remains that owning tradable securities is not without risk. Weighing risk of owning or trading shares is always a critical component of any professional investment/trading strategy. When the preponderance of the time-cycle forecasts indicate weakness or strength in a security and/or its related securities and indexes, better decisions can be made regarding the amount of exposure to the security is warranted and/or when the better time to enter or exit a trade is indicated. The risk assessment aspect of the time-cycle forecast charts is an extremely important tool for quantifying the amount of risk tolerance is acceptable within any investment or trading strategy. EXCELLENT
Market Exposure If you tend to time when you expose more of your money or your client’s money to the market and times when you tend to stay more liquid, then our time-cycle forecasts can be an invaluable aid for determining when to be more or less aggressive in market exposure. When the forecasts tend to show upcoming long bullish or bearish trends that are coincident among multiple asset classes, the indication is that more exposure is warranted. When the forecasts are inconsistent and less correlated, the indication is that less exposure is warranted. EXCELLENT
Return versus
Security Price
Certainly, one way to measure the success of trading based on the time-cycle forecasts is to compare the actual return of the trader to the net gain in the security over the duration of the trade or holding period. Our traders report consistently better performance utilizing their trading strategies by incorporating the time-cycle forecasts, rather than taking a simple buy-and-hold approach. EXCELLENT