Boosted.ai Knowledge Base

New & Improved Boosted Insights!

Feature Guides

Stock Drilldown

A complete guide on using Boosted Insights - Stock Drilldown.

The Stock Drilldown feature enables you to conduct a deep dive into a stock by providing you with comprehensive insights for a better assessment of an investment opportunity. You can access Stock Drilldown by clicking on any stock from Watchlist or Ideas.

How to Use Stock Drilldown?

Stock Drilldown comprises 5 distinct pages: Summary, Factors, Similar Stocks, Drivers, and Patterns.

Summary - An overview of the machine’s recommendation alongside its explanation for the pick and other relevant information.
Factors - A breakdown of how a stock’s human risk factors and machine-learned factors contribute to trailing and forward returns.
Similar Stock - Using a stock’s factor and price movement to find other names in the universe that are similar for hedging or mimicking purposes.
Drivers - A comparison of a stock’s top bullish and bearish drivers between 2 points in time.
Patterns - A list of the top patterns that contribute to buy/sell signals used in the machine’s decision-making process.

You can switch strategies by using the Model dropdown in the top-right corner.

Note, you can only see Stock Drilldown for strategies that are active within Watchlists. To see Stock Drilldown for other strategies you're subscribed to, you will need to navigate to Watchlists and select a different strategy through the gear icon in the top-right corner.

Page Breakdown - Summary

Component Breakdown - Recommendation

At the top of the Summary page, you get the machine’s recommendation alongside the most interesting attributes that contribute to it.

Right of Recommendation, you get the stock’s market data. Market Data shows information at closing from the previous trading day.

Component Breakdown - Historical Price Performance

Historical Price Performance lets you evaluate how accurate the machine has been in predicting a stock’s performance. The chart plots the stock’s performance against the benchmark over a 1Y, 3Y, and 5Y time range. On the Share Price line, you may see a rating: Outperform, Underperform, or Neutral.

Outperform - When a stock’s rating moves to 5 stars, it will be given a rating of Outperform on the Historical Price Performance chart. If a stock gets a rating of Outperform, then the machine is expecting the stock price to go up.
Underperform - When a stock’s rating moves to 1 stars, it will be given a rating of Underperform on the Historical Price Performance chart. If a stock gets a rating of Underperform, then the machine is expecting the stock price to go down.
Neutral - When a stock’s rating moves to 2-4 stars, it will be given a rating of Neutral on the Historical Price Performance chart. If a stock gets a rating of Neutral, then the machine is expecting the stock price to stay flat.

Component Breakdown - AI-Selected Comparables

AI-Selected Comparables is a snippet of Similar Stocks, providing you a quick snapshot of stocks that are most similar to the one selected.

Component Breakdown - Driver Moves

Driver Moves highlights the biggest variable movers and how they impact bullish and bearish signals.

Component Breakdown - Top Drivers

Top Drivers shows the top variables contributing to the bullish and bearish signals for the stock.

Component Breakdown - Sector & Industry Ratings

Sector & Industry Ratings chart shows the average ratings for stocks in those categories, so you have a better understanding of how the selected stock compares to its peers overall.

Page Breakdown - Factors

The Factors page lets you dive deep into a stock’s human risk factors and machine factors to give you a better understanding of how the stock has changed over time and the non-human factors that contribute to the stock’s volatility.

Human Risk Factors are the common factors found in Barra models. Here we show a stock’s factor values, how they compare to others in the universe, the change over time, and their impact on returns.

Value - Human risk factor values are displayed relative to each other through z-score, which are the number of standard deviations apart from the average factor value of the universe. The only exception to this are ESG factors (environment, sustainability, and governance) where they are measured on a scale of 0 to 1.
Quantile Rating - You can see how your selected stock’s factor value compares to all of the other stocks in the universe by looking at the quantile it belongs in. For example, your stock has a Growth value of +1 Z-score and that is in Quantile 1, then you know your stock is in the top 20% of stocks for growth. The star rating is the average stock rating for all of the stocks that fall into your factor’s quantile.
Timing - The chart shows how your stock’s factor value changes over time.
Return - Based on stocks with similar factor values, these are the average trailing returns and the expected forward returns.

Machine factors are relationships found in the universe that help explain volatility but can’t be attributed to single variables or patterns. For each machine factor, Boosted.ai uses natural language processing to assign characteristics to two related groups of stocks for your interpretation. Stocks move together within groups, but inversely between groups. In the example above, Group 1 include … stocks, which moves together but opposite of … stocks found in Group 2.

Value - Unlike human risk factors with measurable value, machine factor values show which group a stock is more related to. If a value is positive, then the stock moves with Group 1, whereas a negative value means it moves with Group 2. If a value is exactly 0, then the stock moves with neither Group 1 or Group 2.

Quantile Rating - A stock’s machine factor value will fall into a quantile, which tells you how its value compares to other stocks in the universe; this gives you perspective on how closely other stocks are related to specific groups. For example, if your stock’s Machine Factor is -1 and it is in Quantile 5, then you know all of the stocks in the other quantiles moves more closely with Group 1. The star rating is the average stock rating for all of the stocks that fall into the machine factor quantile.

Timing - The chart shows how your stock’s machine factor value change over time.
Return - Based on stocks with similar machine factor values, these are the average trialing returns and the expected forward returns.

Page Breakdown - Similar Stocks

Similar Stock provides other stocks in the universe that are similar to the one being evaluated, which enables hedging or replicating a stock’s performance. There are 2 lenses when looking at stock similarity: Correlation and Machine Similarity. Correlation is how looks at how The machine evaluates similarity through Price, Factor, and Overall similarity.

Price Similarity - Looking at how closely stock prices move together.
Factor Similarity - Looking at how closely human risk factors are to each other.
Overall Similarity - A blended view of Price and Factor similarity.

How to Use Similar Stocks

Page Breakdown - Drivers

Drivers lets you see how a stock’s biggest variable changes between two points in time. You can compare differences by explain weight and data percentile using the “Security Comparison” dropdown.

Explain Weight - The importance of a variable for predicting a stock’s performance.
Data Percentile - The actual value of a variable.

To interpret this chart, the solid bar represents the difference in explain score while the translucent bar represents the difference in data percentile.

In the example above, we see that Log Market Cap, Alpha, Volume to Price, and Cash and Equivalents to EV Ratio were more important during the 11/28 rebalance than the 11/21 rebalance. Although explain score changed drastically between the two rebalance periods, the actual values of those variables did not change much. This tells us that the machine’s viewing of the stock has shifted due to changing market conditions.

Page Breakdown - Patterns

The Patterns page gives you a behind-the-scenes look at the machine’s decision-making process by surfacing the most important patterns (variables and combination of variables) that contribute to buy/sell signals. Evaluating a stock’s patterns will give you a deeper understanding of how the machine makes its recommendation and how well it predicts performance in the current market.

You can see how a pattern performed during the training period and compare its performance live to see if it is still working or not. The key metrics to evaluate are buy/sell signal, excess return, hit rate, and instance %.

Buy/Sell Signal - The magnitude of a pattern’s contribution to the overall machine’s buy/sell recommendation - more positive means a stronger buy signal while more negative means a strong sell signal. A pattern’s signals are set during the training period, so it will not change until the next retraining.
Excess Return - When the selected stock falls into this pattern, how much excess return do they yield on average.
Hit Rate - A measure of when the selected stock falls into this pattern, and how often they generate a positive return on average.
Instance % - A measure of how often the selected stock falls into this pattern.

To learn more about how to use patterns, see our article on Equity Explorer.