Addaptron Software announced the release of Stock Market Forecast Tools SMFT-2. It is a new integrated system, the next generation of the software that intended to replace older SMFT-1 version. As a result of a few recent development projects, the new version is based on new advanced methods, provides more accurate prediction, and is easier to use.
Prediction modules is built with back-test calculation to estimate the accuracy of forecast within the recent performance periods. Additionally, the back-testing computations is important if more than one method is used. It allows estimating the weight of each method in a composed result; the weights that are proportional to the ability of the methods to predict the price.
SMFT-2 currently includes five major modules:
- TA Predictor – prediction for day or week period based on technical analysis, pattern recognition and Neural Networks (generates composite result). Back-analysis models optimization and batch calculation for comparative analysis included.
- Waves – Elliott Wave model: back-test optimization, up to 10 waves forecast.
- Cycles – prediction based on cycle analysis.
- Week day – search for maximum performance using price behavior depending on week day. It allows discovering the best entry/exit days of week; batch calculation included.
- Month day – search for maximum performance using price behavior depending on month day. It allows discovering the best entry/exit days of month; batch calculation included.
The implemented methods are statistically proven and widely used. All modules share the same EOD (end-of-day) input data. The software is provided with a free Downloader that allows downloading EOD historical quotes files from the Internet for free. A fully-functional software SMFT-2 is free during initial 30-day period. The software and associated documentation are delivered via download links over the Internet. For technical requirements, installation instruction, and download link, visit SMFT-2 download page.
The WordPress.com stats helper monkeys prepared a 2014 annual report for this blog.
Here’s an excerpt:
The concert hall at the Sydney Opera House holds 2,700 people. This blog was viewed about 9,500 times in 2014. If it were a concert at Sydney Opera House, it would take about 4 sold-out performances for that many people to see it.
Addaptron Software provides prices prediction of major ETFs for the next day. This article describes one of the simplest algorithms to use prediction data. The algorithm consists of four simple rules:
- Buy Limit order with predicted Low price is placed before market open.
- The order is canceled if it is not completed during the first half of the market day.
- Once order is completed, Sell Limit order is placed with predicted High price.
- If Sell order is not completed until almost closing market, it is canceled and Sell Market order is executed just before closing stock market.
The following charts and tables demonstrate one of the practical examples of described above approach.
Predicted and actual prices chart (dark green and red filled candles – predicted, light green and red line candles – actual):
Intraday performance chart (actual prices) for the same period:
The result of using the described algorithm. The example is based on the following assumptions: 100 shares are used for trading, transaction fee is $10 so that 2 transactions (buy and sell) within day cost $20:
Another example (symbol BOIL) can be find here
Addaptron Software released a new version of Stock Market Forecast Tools SMFT-1. The software includes several improvements: models optimization, more different input file formats, and optional free Downloader. SMFT-1 is an integrated system that includes three major software modules: FTA-2 – a modified version of InvAn-4 that is a comprehensive tool used by serious investors for many years, the most popular software program SMAP-3 for stock market cycles analysis and forecast, and Neural Network Stock Trend Predictor NNSTP-2.
FTA-2 itself consists of six major modules:
- “Technical analysis” – more than 50 popular technical indicators; chart analysis; indicators (each separately or all) used as input for Neural Network (NN) to build 10-day price forecast. The forecasts from all indicators result into a single forecast – each forecast added with the weight proportionally to the current ability of the indicator to predict prices.
- “Waves” – Elliott Wave NN forecast.
- “Candles” – candlestick pattern NN forecast model.
- “Pattern recognition” – pattern-recognition filter and predictor.
- “Correlation” – correlation analysis tool to perform analysis and evaluate the future trend using a mutually-correlated pair (or in opposite correlation) with time shift.
- “Comprehensive 3-month fundamental-technical ratings model” – analyzing-predicting model that is based on key fundamental ratios and technical parameters reflecting a company-stock state and dynamics.
SMAP-3 is able not only to extract basic cycles of the stock market (indexes, sectors, or well-traded shares) but also to predict an optimal timing to buy or sell stocks. Its calculation mainly based on extracting basic cyclical functions with different periods, amplitudes, and phases from historical quote curve. Additionally, SMAP-3 enables finding optimal timing to buy/sell by analyzing months of year, days of month, and days of week (the calculation is based on statistical analysis).
NNSTP-2 is to help stock traders in predicting stock prices for short terms. It predicts future share prices or their percentage changes (can be chosen in settings menu) using Fuzzy Neural Network (FNN). It operates automatically when creating the FNN, training it, and mapping to classify a new input vector.
Many discuss if Stock Market Technical Analysis (SMTA) can be used to predict prices. Normally, there are two camps – the side who believes in SMTA predictability and other who does not. In many particular cases, both sides have a solid evidence to support their point of view.
From a statistical point of view it looks reasonable to search for a good predictability in the areas where random or multi-news driven fluctuations are self-compensating each other. One of such areas is stock market indexes, for example, S&P-500. This index consists of 500 biggest companies which make difficult to easily disrupt an equilibrium that is formed during relatively long time frames.
Indeed, as experiments show, one of the most consistent SMTA prediction ratios can be observed for S&P-500 index. Normally, the ratio of its successful/unsuccessful predictions is within the range of 60-68%. Average root-mean-square deviation of predicted-actual values during one week can fluctuate within 2-9 absolute values of index prices (Open, High, Low, Close prices).
Recently, Addaptron Software introduced a new service – Next Day S&P 500 Index Forecast The forecast is calculated using a multi-model SMTA forecast system. User can order the index forecast for the next day, i.e., candlestick values – Open, High, Low, Close prices. The forecast will be sent to your email address by automated tool.
A multi-model forecast provides a significant improvement over the best individual forecast. It can be explained by existence of many different independent factors contributing to the error in each forecast which distributed around actual value.
Today, airplanes are equipped with a few different altimeters – barometric altimeter, radar altimeter, and GPS. Not to mention, pilots also use a visual estimation of altitude. Why do we need to use so many measurement tools? The first reason is that any of them can fail. There is another reason why multiple measurements are used everywhere – it is accuracy. For example, many different methods are used in weather forecasting to improve the accuracy of forecast.
The phenomenon of multiple forecast improvement can be compared with Expert Method. This method can be illustrated by the following. As example, an experimentalist shows a pen and asks a group of several people to write down their estimate of the length. Then he collects notes and calculates the average number – normally it is almost a precise result. Why it works? Because everyone makes errors in different directions so that averaging self-compensates erroneous deviations.
Concerning the stock market forecast, as experiments show, a multi-model forecast provides a significant (10-25%) improvement over the best individual forecast. Also tests show the advantage of using information even from multiple forecasts of different quality. That is because there are many different independent factors contributing to the error in each forecast and the results from different models are normally distributed around actual value.
Evidently, the methods should be different by their nature. Traditionally, fundamental factors and technical analysis are the major stock market tools. Within technical analysis, there are several different models: technical indicators, chart pattern analysis, Elliott Wave theory, cycle analysis, candlesticks model, trend lines analysis, regression models, etc. Most of these methods are statistically proven and widely used that often create self-fulfilling results.
As a rule, learning and also correctly using many of technical analysis methods may require a lot of time, especially, in a modern dynamic trading environment. Fortunately, the forecast methods combined with computer power have become a good solution to make the works less time-consuming and more effective. Except different linear and non-linear solvers and analytical methods, these days, Neural Network (NN) can help to automate a lot of computational tasks. A properly trained NN may enable predictions to the highest accuracy.
However, implementing NN application can be difficult for non-experts. Besides, one of the big obstacles of implementing NN predicting system is a formalization of inputs. Luckily, there are some software tools that already well-developed and do not require a deep technical understanding. These tools are optimized for each method and users might not notice even all computational power behind the buttons, windows, and charts.
A new 2012.02 version of Fundamental-Technical Analyzer FTA-2 has been released. Now it has a new module which enables using Neural Network to process past candlestick patterns and predict the future candlestick, i.e., its Open, High, Low, and Close prices. A candlestick can consists of one or many trading days. The module calculates result composed from different historical periods that allows making the forecast more accurate. Also it can perform comparative forecast analysis for many symbols.
More details can be found on FTA-2 software webpage, as well as, in User’s Manual that is available from the main menu after the installation of FTA-2. To try the software for free, go to download page and follow the instruction.
About Candlestick Forecast Model
The idea of using the chart with candlesticks (or candles) for predicting market prices is very old. Two centuries ago, Japanese rice trader found that the candlesticks pattern chart could be used as a tool to predict future prices in a free market with a natural demand-supply balance. The method was improved later by others and today it is successfully used by many traders and investors in the stock market.
A candlestick is presented using high, low, opening, and closing prices during a certain trading period, for example, trading day. A regular candlestick figure consists of Real Body, Upper Shadow, and Lower Shadow. The number of candlesticks that is normally used for predicting can range within 1..12. Evidently, the number of different combinations of several candlesticks in a row can be big. Some believe that there are only 12 major candlestick patterns, others consider this number is 70 or even more. Anyway, in case of chart analysis, it is necessary to remember at least major patterns and process many charts in order to make forecast successful.
Apparently, statistical methods combined with computer power can be a good solution to make the candlestick patterns recognition work less time-consuming and more effective. For example, Neural Network (NN) can help to automate a candlestick patterns recognition task. NN should be properly trained in order to be able to recognize and predict further movements. One of the obvious problems of implementing a candlestick pattern NN predicting system is a formalization of inputs, i.e., the way how to express each candlestick shape and relative position of all candlesticks in numerical values.
Parabolic SAR can be improved and successfully used for predicting stock market prices, especially, in trending markets
Traditionally, Parabolic SAR* is considered as a trend following indicator. Probably, few traders would think about using it for prediction. But after testing, I started to believe that it can be successfully used for prediction in conjunction with other indicators, especially, in a trending market. The explanation why it works can be the following. When a trend reverses, the probability of its continuation is more than 50% in average. The software with Neural Network (NN) is able to catch it statistically and show the result in extrapolated curves. The function of SAR in this method is to provide NN with reversal point signals.
The initial hypothesis was the following. Since SAR is able to give a strong signal when a price trend is reversing, this signal can be used for predicting if a new trend is prone to last. To compare predictive ability of SAR with other indicators, it has been implemented into the technical analysis module of Fundamental-Technical Analyzer FTA-2. SAR calculations have been used to collect statistics based on the forecast simulations for major indexes and ETFs during August-October 2011 period. As a result, SAR’s position was mostly in “top ten” indicators’ list.
The research and presented chart are made by Fundamental-Technical Analyzer FTA-2, one of the software modules that enables composing Neural Network forecasts of many indicators with weights accordingly to each indicator’s predictive ability.
Omitting logical rules for acceleration factor and reversal conditions, a recurring core formula for Parabolic SAR is the following:
SAR (current point) = AF * [EP - SAR (previous point)] + SAR (previous point)
AF – Acceleration Factor (normally starts from 0.02 and increases by 0.02 if each next point reaches a new extreme, saturates until 0.2);
EP – Extreme Point (lowest low or highest high).
Conclusions. SAR is especially effective in a trending market. To make it more effective in a sideways market, it is a good idea to use it in conjunction with other indicators. Parabolic SAR can be enriched and successfully used for predicting stock market prices. Combing it with Neural Network allows extracting more statistically stable patterns and, therefore, providing a better accuracy in the forecast. As simulations showed, improved results can be achieved if SAR is transformed into more sensitive indicator by subtracting it from close price (it indicates the degree of SAR and price convergence).
*) SAR stands for Stop-And-Reverse. It has been used by many traders for decades. Its major application is in trading systems to define a trailing stop, i.e., to protect profit when a price trend changes. The term “parabolic” appeared to characterize the indicator parabola shape that is due to using an accelerating factor in the formula.
Article first published as Elliott Wave: Extracting Extremes and Predicting next One by Neural Network on Technorati.
Improved Elliott Wave model can be successfully used as an additional input for making investing decisions in modern market conditions
The Elliott Wave idea is to use in stock market forecast. It is based on a crowd psychology that changes between optimistic and pessimistic trends creating patterns that can be fitted to reoccurred sequences. To use waves for prediction the assumption is made that waves are developing in the sequence of Fibonacci, harmonic, or fractal ratios. So that each wave has a programmed position and characterized by a particular direction and duration with extreme as a reversal point.
In fact, the model has been used by stock market analysts for almost a century. Although it looks very attractive due to its strict formalism and deterministic outcome, its predictive power is weak because of a few reasons. Firstly, the predictive values are dependent on waves that were counted – determining where first and next wave start can be subjective. Secondly, according to Efficient Market Hypothesis, using an exact Elliott Wave model by many traders could lead to the disappearance of the patterns they anticipate. And finally, nowadays one’s trading success based on predictions is rather a chance game in a modern market with its irrational behavior.
The purpose of this research was to explore if Elliott Wave principle can be used these days in stock market forecast. To eliminate the subjectivity in counting waves, Neural Network (NN) was used to analyze and predict waves. Also instead of assuming that waves obey only the sequence of Fibonacci, harmonic, or fractal ratios, a more general approach used – the software processes all extracted waves. Besides, employing NN enables identifying both the price and date of extremes. The first experiment has been done using an artificial data set. The data consist of two sinusoidal functions with different periods. The second group of experiments has been done on real market data.
The main conclusion is that the Elliott Wave idea can be used in predicting stock market. Although it does not generate always accurate and consistent forecasts, its result can be successfully used as an additional input for making a trading or investing decision in modern market conditions.
Charts and calculation results by SMFT-1 (TA-1 sub-system Waves module interface)
A new decision support software system for stock market traders has been release. It is called Trading Forecast Weekly TFW-1. It provides weekly EOD forecasts for 20 selected ETFs by cloud computing. TFW-1 allows choosing best trading opportunities by comparative analysis, assessing different strategies using back-test simulation, finding the best configuration of algorithms among 64 possible combinations and optimizing triggering parameters, as well as, monitoring results by stand-alone computing.
The simulations and statistical analysis show that systems based on predicted entry-exit signals generate a better profit than random-entry trading systems. TFW-1 enables forecasts to increase a trading profit. The calculations are performed on server-side by a robust artificial intelligence system. Since sometimes predictions can fail, to preserve capital in a volatile market, the software enables simulating different risk management approaches. TFW-1 simulation allows testing generated predictions combined with buy-sell signals.
It is known that depending on the character of particular trading shares and current stock market condition, some ideas can work better than others. To optimize strategy for each particular case, the software enables testing a given idea and finding automatically the best set of algorithms. It is especially important for exit points to minimize losses and ultimately to maximize an overall profit. Some algorithms have customizable parameters that can also be optimized. All optimizations can be done either automatically by scanning 64 possible logical combinations and adjusting numerical parameters or manually.
TFW-1 allows monitoring the simulated or actual completed transactions, reflecting total trading activity, and evaluating the success of trading in overall by stand-alone computing. It enables working with many separate data files that is convenient in case of managing multiple assets and keeping the archives of older activities. You are welcome to download and use for free (during the first 30-day period) a fully-functional version of TFW-1.