PAL is one of the tools we use to analyze market behavior patterns. When we perform our scans, a considerable amount of data is automatically collected. We have created a model/algorithm that makes use of the collected data to derive inferences, projections, and probabilities. The advantage of using a computer-driven algorithm to gather data, conduct tests, and draw conclusions is that it tends to minimize if not completely eliminate human emotions and biases from the process. Humans often overlook data available to them, or they may underestimate the importance of some data. Some of the output of our system is generated by proprietary Probability ALgorithms. For the sake of brevity and convenience, we therefore sometimes refer to our model/system as PAL. It is PAL that has been calculating “Intraday Key Levels,” resistance levels, Group Pressure Gradients, and probabilities. No human or system can predict precisely what the markets will do the following day. That is why PAL uses probability algorithms to draw some of its conclusions. PAL gathers the basic data it requires each day, then applies numerous algorithms to analyze and interpret the data. Finally, it generates a detailed report for our use.
It is important to remember that even when market conditions seem to have clarity and definition, sentiment can change in a moment because of a news event. The result of sudden shifts in sentiment cannot be predicted, and they may result in market behavior that is completely outside the existing probability envelopes. News events, by definition, cannot be predicted. Thus, sudden trend reversals and other unexpected market behavior can occur at any moment of any day.
We have set PAL to estimate behavior patterns at probability levels that are practical for the behavior being measured. For example, if we were to use a 90% probability level, the envelope of probable excursion would be wide (less focused). If we were to use a 70% probability level, the envelope of probable excursion would be more narrow (more focused), but then the probability of market behavior actually being contained within that excursion envelope would be less. We have programmed PAL to use probability levels that give a reasonably good chance that the action being analyzed will actually fall within the predicted range, without the range being so broad that it is of little practical use. To say, for example, that the average age in a room of 10,000 people has a 100% probability of being between 0 and 150 years is a highly accurate statement, but of little use. To say that the average age in a room of 10,000 people has an 80% probability of being between 35 and 50 years is to provide much more useful information, even though the probability is less certain.
Let’s look at an example. PAL may conclude that probabilities favor an early advance tomorrow, barring negative news events. PAL’s probability algorithm for tomorrow morning is based on very short-term data available as of the close of market today. However, PAL cannot include in its computations the probability that investors will sell in order to lock in profits. For example, assume that the Dow has risen for five consecutive days and that today’s action is exceptionally positive. PAL will likely estimate a high probability that the positive action will continue if it cannot detect any contrary indications. It cannot determine that investors were astonished at the recent performance and have become so nervous that trhey want to sell in order to lock in profits. Even the investors themselves may not know this at the close of market. They may conclude it is time to sell only after they have had time to think about it long enough to become nervous. So, one of PAL’s algorithms estimates the probability of an advance or decline tomorrow on the basis the most recent market activity as of today’s close. In addition, it has other algorithms that look at tomorrow’s early action to estimate whether the market is likely to close higher or lower tomorrow. At the close, PAL may project that the probability of an advance early in tomorrow’s activity. However, things can change quickly overnight because of a news event. Hence, PAL’s other algorithm will make additional projections based on what happens in the first hour of trading on the following day. If something has happened to change the probabilities overnight, the second algorithm will likely reveal it.
PAL computes resistances and supports. Some of these are based on a probability analysis and recent volatility patterns. For example, PAL can determine that there is a 98% probability that shares will retreat when a certain level is reached. As with all lines of resistance, sellers may begin to unload their shares a little before that level is reached in order to avoid having to sell when others are driving prices lower. The same thing occurs in reverse when shares approach a level of support. The probability patterns tell PAL that sellers or buyers are likely to enter the market as a certain level is approached. Other levels of support and resistance are based on past patterns of accumulation and distribution. For example, PAL can spot levels where support or resistance is likely to be encountered based on price patterns and the way share behavior tends to respond to those patterns. Technicians know that there is likely to be support or resistance at a trendline or moving average, and that whether a trendline or moving average acts as support or resistance depends on whether the current price is above or below that line or average. Why do we compute supports and resistances? Knowing the location of support and resistance can be extremely helpful to tactical positioning, regardless of what the market does. Knowing where supports and resistances are located can be the most important and useful information a trader/investor can use in buying or selling. For example, smart investors often wait for share prices to approach or reach support before buying. Then, they place the stop loss just below that support to reduce potential loss. If there is sufficient selling to overwhelm the buyers at that level, there is good reason to sell. Since the position was taken just above support and the stop loss is just below support, the potential loss is minimal. That reduction in risk can enable a person to take a position when it would not be advisable otherwise. A stop order to buy placed just above resistance can enable a person to take a position early on a breakout with minimal risk and without a need for constant monitoring. Also, knowledge about the strength and location of support or resistance can be a help in estimating the probabilities associated with the market’s next move, or how far it is likely to be able to continue in a particular direction. Some supports and resistances are static. That is, they remain at the same level day after day. Others are dynamic. They vary from day to day. For example, a long-term declining trendline will result in resistance (selling) at a lower price level each day until there is a breakout above the declining line of overhead resistance. A rising 50-day moving average will provide a higher level of support each day until there is ultimately a break to the downside through that support.