​​​

CHERRIES’s main Goal is the creation of optimized* stocks portfolios. 

​

CHERRIES works on the basis of combining AI ​​models with an embedded mathematical engine. 

 

The output of the process detailed below is a weighted stocks portfolio. 

 

CHERRIES enables two separate portfolio building processes: 

1. DEFAULT - user selects the CHERRIES portfolio, as constructed and backtested by our own research. 

The CHERRIES portfolio is built and updated based on 15 years of historical market data. 

Those include basics such as standard deviation and trading volume, as well as fundamentals data  - both reported and calculated), e.g,  EBIDTA, Profit margin, Z-score etc. 

The CHERRIES portfolio has regularly outperformed the selected Benchmarks - the SP500 index as well as leading Israeli mutual funds. 

1. PERSONALIZED - user creates their own portfolio/investment strategy based on unique preferences and constraints. 

​

CHERRIES enables extensive adaptation features, including dozens of trading and fundamental data, and a sophisticated constraints array. 

The personalized route enables maximal compliance but does not benefit from the overperformance of the researched model. 

 

The process below pertains to both modes. 

 

CHERRIES Data Handling

RANKING 

​

CHERRIES works on the basis of EOD prices. 

 

At the end of each trading day, CHERRIES analyzes the entire database for changes in prices, dividend and split data and other market events and fundamental data. 

 

It then creates an internal ranking of all stocks in the DB (currently - all major traded US stocks.)

 

This ranking is based on a proprietary formula that rates each stock by robustness. 

1. In route 1 - DEFAULT - The CHERRIES portfolio is built out of the top 1500 stocks based on this internal ranking. 

2. In route 2 - the user controls the number of stocks to be analyzed. 

 

Covariance matrix

 

The mathematical engine creates a daily matrix of all stocks in the database, examining the covariance and correlation between each pair of stocks, thereby gaining a complete snapshot of optimal relations. ​​​
 

​

​

​

​

​

​​

​

​

​

​

​​

​

Efficient Frontier

All portfolios that are constructed optimally and all optimal portfolios are connected by an imaginary curve called: "Efficient Frontier" which delimits between the possible – below the curve, the optimal – on the curve, the impossible above the curve.

​

​

​

​

​

​

​

​

​

​

A. The structure of the curve – rises from the left, from the beginning of the scale, towards the right continuously and then flattens out until it is almost parallel to the horizontal axis – the risk rate axis, which indicates that beyond a very high level of risk, it is no longer possible to achieve a higher return.

​

B. Uniformity of directions – this curve cannot rise and fall like curves that describe changes such as stock market prices and if this happens – it is a malfunction and its causes must be examined.

​

C. The beginning of the scale – the scale is built from a horizontal axis on which all the risk levels existing in the portfolios that can be derived from the set of requirements selected by the user, and a vertical axis on which all the yield levels of all 156 portfolios are displayed, the intersection of both should be as usual (0,0)

​

D. Transformation of the axes – in practice, the axes will intersect on the vertical axis – from the point with the lowest yield of the most solid portfolio, and on the horizontal axis – from the existing lowest risk level and to the right to the higher levels.

E. All of the above features are done automatically by CHERRIES.

 

Out of all optimal possibilities, CHERRIES selects the relevant portfolio (opt), which corresponds to the risk level of the relevant benchmark. in the DEFAULT route - the SP500, in the PERSONALIZED route - the relevant BM is determined by the user

 

CHERRIES model combination

The tested CHERRIES model (DEFAULT)  is based on the combination of several different analyses from two different technologies - AI models and a mathematical engine. 

 

The mathematical engine creates the above described matrix, while the AI deploys propriety logic, both aimed at creating a maximally diverse portfolio, which maintains a low SD and DD along with ROR which out performs the benchmarks. 

 

As mentioned the output received by the user is one comprehensive list of stocks - The Combined Model -, complete with extensive data (both reported and calculated) for each stock.
 

Benchmarks and performance measurement:

Cherries portfolio is measured against the SP500 as well as an average of leading Israeli mutual funds: Top Alpha menayot hul, Yalin Laoidot menayot hul, Kesem active hul, Sigm HUL (See table below).

 

All tests take into account a buy and sell margin of 0.1% and a buy and sell commission of 0.1%. Tax withholding is not taken into account.

​