Swedish

Screenshots

Portfolio Scan Factors

Trend: %/Day: (Regression Slope)

Trend is a measure of price growth. There are sites and software where you can sort stocks or funds after the performance of a specific time period. Optimal Trader is much more powerful:

1. You can use a combination of three different trend factors with different time ranges: one short term, one medium term and one long term trend
2. You can add different weights (importance) to each time period
3. Noise in price data is efficiently smoothed by an adaptive moving average (OptAMA) without introducing lag
4. Trend values become even less sensitive to noise if it's value is based on regression analysis. All price values of the time period will be included in the calculation of the trend, not just the first and the last value of the time period.

To consider only one time range is too simple and can be dangerous. There are several explanations for that:

1. To scan for stocks with high returns for the latest two month period gives you no information of when this growth has taken place. If the largest part of the growth has taken place the first two weeks for a specific stock it may not be a suitable investment object.
2. The price may have fallen substantially before the start of the two month period so the sharp rise may only have been a recoil.
3. How has the price developed under the two month period? The price may have oscillated within an interval which can make the return look much stronger than it has actually been. This is because the price today may be on the top of the interval while it was in the bottom two months ago.

By combining the results of several time ranges the problems described above can be strongly reduced.

In Optimal Trader you can set three different time ranges and add different weights to each of them in the scan.

If you use trend factors it is advisable to always control the charts of the stocks which have performed best. Recoils can always occur when returns have been exaggeratedly strong.

 

Expected Returns/Day (Statistical Classification)

Expected returns are estimated with the help of the statistical classification indicator. The current price pattern of the stock is analyzed and classified to belong to one of 3888 pattern-groups. The classification is based on several properties such as volatility, trend strengh and fractal dimension. A large database with 500000 results is used to give an estimation of future returns based on the current pattern group. The values are based on average results for stocks in the database which exhibited the same pattern.

You can select if this factor should reflect expected returns one day, one week or one month ahead, by changing the horizon. No matter which horizon you choose, the expected returns will be shown per day. That means that if you have selected 1 month as the horizon and the expected returns are 0.1%, the expected returns are 0.1% per day on average the next month. Select your horizon based on your investment style (more about trading horizon).

Notice that expected return values only are valid if factor normalization is not selected. Otherwise values for all equities are scaled by the same amount.

 

Artificial Neural Network

The neural network analyzes stock behaviour and includes other technical indicators to forecast future price trends. You can include results from the neural network when scanning your portfolio.

Notice that analyzing can take a long time to complete if you include this factor and have many objects in your portfolio.

You should use at least 300 days for neural network modeling. You can use fewer days, but this is not recommended because the results may be uncertain.

When you have analyzed your portfolio with the neural network Optimal Trader does not need to re-analyze the portfolio if you are only changing the weights.

Values for the neural network are between 0% and 100% (if no factor normalization is selected). The larger the value, the stronger the buy signal.

 

Volatility

By including volatility estimations you can adjust the risk of your portfolio. You can either maximize volatility, that is value stocks higher if their price movements vary strongly, or minimize volatility, that is value stocks higher if their prices vary less. You can set the time range used to determine volatility.

Volatility is a measure of the average percentual change per day for the stock. Strictly meaning, volatility is actually the average deviation from the average daily price change. If the average return under a period is 0.5% and volatility is 0.2%, probability is around 70% that daily returns lie in the range 0.3% and 0.7%.

Notice that the values only are valid if factor normalization is not selected. Otherwise values for all equities are scaled by the same amount.

 

Short-Term Volatility Factor

This factor expresses the current short term volatility relative to the volatility in a longer perspective (normal volatility). If the Short-Term Volatility Factor is 15%, it it means that volatility is 15% higher now than the long-term volatiliy. If the volatility factor is -20%, it means that the price is moving calmer now than it usually does. The time period used to calculate the long term volatility is the one set for the previous factor (the normal volatility) and the time range of the short term volatility is the one set by this factor.

Some investors use this factor to detect that a change has occurred with this factor. For instance, this factor will get a high value if the price of a stock which has slumbered for some time (low volatility) suddenly begins to move more. This may indicate future price advances.

Notice that the values only are valid if factor normalization is not selected. Otherwise values for all equities are scaled by the same amount.

 

Beta Coefficient

A measure of the movement of each individual equity in relation to the rest of the portfolio. It measures the part of the equity's statistical variance that cannot be reduced by diversification, because it is correlated with the returns of the other equities in the portfolio.

The correlated risk, measured by the Beta coefficient, is what creates almost all of the risk in a large portfolio. The Beta Coefficient is thus the key measure for creating a diversified portfolio and minimizing risk. For example, if every equity in a portfolio was uncorrelated with every other equity, then every stock would have a Beta of zero, and it would be possible to create a portfolio that was nearly risk free, simply by diversifying it sufficiently so that the variations in the individual equities prices averaged out.

In reality, equities within a portfolio tend to be correlated and it is difficult to create a diversified portfolio (roughly 70% of a normal stock's movements are correlated to the rest of the market).

If the beta values within a portfolio are large this is an indication that the equities are correlated with each other, resulting in higher risk than a portfolio where the betas are smaller.

The Beta Coefficient can also be used to find interesting equities in a sector. If you set up a portfolio with stocks from a specific sector, you can sort the stocks by their betas and find stocks that do not behave like the rest of the sector.

When No Normalization is selected the following applies: By definition, the market itself has an underlying beta of 100%. A Beta value close to 100% means that the equity is behaving like the rest of the portfolio. A Beta value larger than 100% means that the equity is moving like the rest of the portfolio, but stronger. A Beta value of 0% means that the equity is not correlated to the rest of the portfolio. A negative Beta value means that the equity moves opposite to the rest of the portfolio.

When Absolute Normalization or Relative Normalization is selected you cannot read the absolute beta values in the result table. You can only read the relative difference between them.

 

Hurst Exponent

An estimate of the predictability of a time series.

When No Normalization is selected the following applies: The values of the Hurst Exponent range between 0% and 100%. A Hurst Exponent value H close to 50% indicates a random walk. A value between 0% and 50% exists for time series where an increase will tend to be followed by a decrease (or a decrease will be followed by an increase). Future values will have a tendency to return to a longer term mean value. The strength of this increases as H approaches 0%. A value between 50% and 100% indicates that the time series is trending. The larger the H value is, the stronger the trend. Series of this type are easier to predict than series falling in the other two categories.

When Absolute Normalization or Relative Normalization is selected you cannot read the absolute hurst values in the result table. You can only read the relative difference between them.