Additional Platform Predictors
This is a new agent-based model, which incorporates trend persistence, fundamental valuation, and counter-trend methods. Each of 1000 computational agents is randomly assigned a blended trading strategy and money to manage. At each time step, the agents use their assigned models to go long or short, and are rewarded or penalized based on performance. The aggregate weighted vote and the vote by the highest performing agent are made available and plotted on the screen.
The concept of support and resistance is as old as markets, yet if their existence is statistically significant, it implies that a feedback loop exists between price movement and investor decisions. If prior price levels do influence investment decisions, then markets can be said to have a collective price memory, which makes them somewhat predictable. Parallax has designed a tool to identify and plot the most likely prices at which to expect this behavior. We have found that if price stops just shy of support or resistance (0.5 standard deviations for a one-bar move with historical volatility), then for the next few bars, price has a significantly higher probability of not closing beyond the line.
Price-volume crossover patterns are an attempt to capture a pattern of crowd investment behavior prior to a significant rise or fall in prices. With Marc Chaikin, we studied 24 of these patterns and found four that were significant. They are located using this indicator.
How far does the price have to move before the trend is persistent against your trade? In other words, if I’m long, I want a stop that means something. The stop should be placed so that if hit, it means that price is exhibiting “trend persistence” against my trade. This is far preferable to just setting a line in the sand at an arbitrary fixed price or percent. By setting a persistence threshold at h = 0.7, we can use Hurst analysis to estimate a high and low price that would have to be reached today for trend persistence (up or down) to be present.
The output is a series of colored columns representing the Hurst Exponent at different scales. The shortest scales are on the bottom. The colors are as follows: red represents persistent down trends; green represents persistent up trends; gray represents random price movement; and blue represents mean-reverting price action.
Our Price Wizard valuation tool takes a snapshot of current valuation, but does not incorporate any information about how quickly that valuation might be rising or falling. Likewise, no market price information has been included in PriceWizard. Our Fundamental Rank neural net incorporates the 1-, 3-, 6- and 12-month rates of change in valuation and actual price, along with the current valuation, to give an overall attractiveness rank from 0 (unattractive) to 100 (attractive).
This tool analyzes data from trend persistent periods, as designated by SmartChannel parallel channels. A neural net has been trained on such factors as volatility, volume accumulation, trend persistence, trend-line contacts, price position within the channel, and power law price action, in order to predict future price action. The duration of forecast is approximately 20% of the channel length forward. A low number is bearish and a high number bullish.
This version of relative strength counts the number of up and down bars within the most recent N bars and performs a simple statistical test on the result. It assumes that up and down bars have equal probabilities of occurrence. A reading above or below the lines drawn on the chart indicates a statistically-significant deviation from random has occurred.
Adaptive Moving Average
Adaptive moving averages are nothing new. Perry Kaufman and others pioneered this technology. The AMA is designed to adapt to the market by flattening during a sideways period and closely following price during a trending period. Since our Hurst tools are well suited to measure the state of trend, we have redesigned the AMA to use them.