About Portfolio Modeling
Quantly's underlying technology is based on the concept of value investing — a philosophy developed by Benjamin Graham and used by Warren Buffett. It is based on the principle that a stock or ETF should only be bought below its fundamental value in order to deliver consistent returns.
Quantly's engine uses data science to identify undervalued stocks and ETFs, and to rank them based on probability of movement. It also incorporates options data and technical analysis to identify entry points, to identify higher probability options strategies that may be used in place of equities when rebalancing (optional), and to maximize profit potential from market volatility.
The Forecast Model
Underlying Quantly's engine is the Forecast Model System — a logistic regression model that trains itself to predict the probability that a given stock or ETF will rise in value after 14, 30 and 60 days (or other timeframes). The system considers multiple variables including long-run earnings per share growth, stock volatility, beta, the correlation between the S&P 500 Index (SPX) and the stock or ETF, and other factors.
Quantly provides a Score Range for each opportunity identified. This value/momentum score, based on a scale of 1 to 10 with "10" being the most favorable expected risk-and-return outlook. The Score Range conveys the long-term outlook (on a 1- to 5-year time horizon) Quantly has assigned based on its proprietary algorithm, technical and fundamental analysis.
With the investment criteria you provide, Quantly uses the Score Range to identify stocks and ETFs with the highest statistical likelihood of providing a positive return on investment.
1. Determine what data provides the best features for forecasting stock or ETF movementQuantly uses logistic regression because it is an effective machine learning technique that also measures bias and variance to assess whether or not the model is a good predictor. Quantly collected different types of data and fed it through an optimizer to "train" the model and determine the best weights to put on each data type to improve predictive accuracy.
2. Model validationQuantly trained the model on 60% of the data collected (an industry standard) and used the remaining data for validation/calibration. Validation data is used to see how the model performs on data outside of the training sample. Using this data, Quantly constructed learning curves, which contrast the accuracy of predictions using data inside and outside of the training set to determine if the model suffers from high bias (i.e. it's not a good predictor) or high variance (i.e. it's curve-fit to the data). Based on these results, Quantly determined how to improve the algorithm by adding more training data or by trying new combinations of data features. This is a continual process of validation and recalibration to improve predictive accuracy.
3. Performance analysisQuantly evaluated the best performance the model was able to achieve on the test data and what data features and weights were used. Based on results indicating predictive accuracy, the model was converted to a neural network, a powerful but more involved machine learning technique.
Modern Portfolio Theory holds that it is possible to optimally allocate investments between assets to maximize expected return, based on a given level of market risk. Yet, as market shifts happen by the second, as soon as an investment portfolio has been constructed to be optimized against specific objectives, it's almost instantly sub-optimal.
Quantly's tools monitor the performance of your investment portfolio and, when rebalancing is enabled, continually rebalances allocations in an effort to be positioned to outperform market benchmarks. The Quantly platform can suggest a more optimal portfolio, and, when allowed, use options to achieve the desired balance.
The rebalancing algorithm constructs these efficient portfolios by examining asset correlation and using the optimization scheme you specify, including:
- Buy and hold: Maintains a static asset allocation; relies on manual rebalancing.
- I-VEL™: Seeks to keep the portfolio near the user-defined Desired Return with minimal deviation (and minimal possible risk).
- Equal weights: Seeks to maintain a static allocation of weights and will rebalance (based on user-defined frequency) as asset prices fluctuate.
- Markowitz (w/o shorts): Seeks to reach a specified return with minimal variance using Markowitz mean-variance optimization; relies on user-defined rebalance frequency. Same as the standard Markowitz, but this algorithm will not use short selling. To minimize risk, it does not use any leverage.
- Markowitz: Seeks to reach a specified return with minimal variance using Markowitz mean-variance optimization; user-defined rebalance frequency.