Trending Factors for AI-powered stock analysis (TAOTS 5th Month of 2024)
Trending Factors for AI-powered stock analysis (TAOTS 5th Month of 2024)
Greeting investors! Welcome to our Newsletter, in which we look at the Trending Factors from the fifth month of 2024. Factors are the building blocks for AI-powered stock market analysis, stock picking, and the creation of robust investment strategies.
Introduction
We are focusing on the performance of six key factors including Momentum, Quality, Size, Trend, Value, and Volatility that influence the stock market. Our goal is to provide a comprehensive guide that will allow you to make informed investment decisions. Read more about our 6-Factor (Groups) Investment Model.
Trending factors from last month
The two significant factors for May 2024 were Volume Rate of Change (ROCR100_10) under the Momentum category, and Variance_5 under the Volatility factor group.
In May 2024, we saw (just like in previous months) a significant decrease in momentum represented by a -37.83% change, however, the ROCR100_10 factor performed a significant gain of 280.83%. For the volatility factor group, we sighted a decline of -4.87%. Nonetheless, the importance of Variance_5 has increased by 162.38%. These metrics are significant players in their respective factors, and by understanding them, we are unraveling the cryptic workings of the stock market, one piece at a time.
Description of selected factors
Volume Rate of Change (ROCR100_10) (Momentum group of factors)
The Volume Rate of Change shows whether or not a volume trend is developing in either an up or down direction. A positive Volume ROCR100 10 points toward an increasing volume trend which could indicate the strength behind a price move. High trading volumes suggest strong investor interest and a high level of liquidity. ROC is a technical indicator used to measure the percentage change between the most recent price and the price ”n” periods in the past. In our case, we use 100 periods as a base (ROCR100 10).
Variance 5 (Volatility group of factors)
Variance 5 is a measure of volatility and captures the variability of a stock’s returns. A rise in volatility often presages a change in market sentiment which could lead to moves in one direction or other.
Is it worth selecting stocks for your portfolio based on these factors? We will find out in our analysis.
Factor Statistics 01.05.2024 – 31.05.2024
| Top mover factor within the factor group | Mover value | Factor group | Group total value |
|---|---|---|---|
| Volume_ROCR100_10 | +2.8083 | momentum | -0.3783 |
| CapExToMarketCap_Position_125 | +0.8062 | quality | +0.0164 |
| CurrencyVolumeRelChange_5vs44 | -1.6597 | size | -0.5038 |
| Volume_WMA_2 | +0.2842 | trend | +0.0066 |
| DividendYield_Position_66 | +0.9100 | value | -0.0876 |
| Variance_5 | +1.6238 | volatility | -0.0487 |
The figure shows the Top Movers across all the groups of factors we use in our analytics.
Let’s take a look at how traders who believed in the two most trending factors have fared in recent years
Statistics for factors
Factor Statistics 1990 – 2024
| Top mover factor within the factor group | T-Stat | P-Value | Direction |
|---|---|---|---|
| Volume_ROCR100_10 | +0.1859 | +0.4263 | +1.0 |
| CapExToMarketCap_Position_125 | +0.2308 | +0.4088 | +1.0 |
| CurrencyVolumeRelChange_5vs44 | +1.3777 | +0.0848 | +1.0 |
| Volume_WMA_2 | +1.4363 | +0.0761 | +1.0 |
| DividendYield_Position_66 | +1.2203 | +0.1117 | +1.0 |
| Variance_5 | +0.5513 | +0.2909 | +1.0 |
In the table, we can see the T-Stat and P-Value for the entire period of history for which we evaluate the factors, i.e. 1990 – 2024. The Direction value shows the direction in which the factor affects the movement of the share price, +1 expresses the positive impact of the increasing factor on the share price, and -1 expresses the negative impact of the increasing value of the factor on the share price. In simple terms of theory, we can say that a T-statistic that is greater than 2.0 or less than -2.0 is statistically significant; and if the P-value is less than 0.05, we reject the null hypothesis and say that we found a statistically significant effect.
To understand the performance of these factors, we delve into their long-only portfolios based on quantiles, and long-short portfolio returns. The five quantiles (Q1-Q5) divide the data into equal parts according to the respective factor value, creating a picture of distribution across the portfolios. Investors usually target Q1 or Q5 portfolios of a factor, depending on whether the higher or lower value of the factor is expected to outperform.
The charts below show Factor Statistics 1990 – 2024 for the 2 selected factors from the previous month.
While the next one shows the behavior of the factors in the last 4 years.
Factor Statistics 2020 – 2024
| Top mover factor within the factor group | T-Stat | P-Value | Direction |
|---|---|---|---|
| Volume_ROCR100_10 | -1.2563 | +0.8926 | -1.0 |
| CapExToMarketCap_Position_125 | -0.5747 | +0.7160 | -1.0 |
| CurrencyVolumeRelChange_5vs44 | +0.9476 | +0.1738 | +1.0 |
| Volume_WMA_2 | +1.1162 | +0.1347 | +1.0 |
| DividendYield_Position_66 | -0.2576 | +0.6011 | -1.0 |
| Variance_5 | +0.4818 | +0.3159 | +1.0 |
The charts below show Factor Statistics 2020 – 2024 for the 2 selected factors from the previous month.
Factor Statistics Insights
The Variance 5 factor shows the same performance of the Long-short portfolio in the short term as on the long term time series. In other words, the price of stocks from the Q5 quintile grows faster than stocks from the Q1 quintile (See the same ‘Direction’ value). However, the values of the statistical variables T-Stat and P-Value do not confirm the statistical significance of the factor, so portfolio managers should be cautious when using this factor for managing investment strategies.
For the “ROCR100 10” factor, it is evident that the stock performance of the sub-quantiles is significantly different in both periods under review, which is confirmed by the values of the statistical indicators.
Do you want to invest in statistically significant factors that work over the long term? Try our Factor Investing app.
Invest in the forces that move stock prices
Leverage the power of factor investing with revolutionary software
Using factors in AI Stock Analysis
Our AI-powered StockPicking Lab is built on the factor investing approach combined with machine learning. We say that the most effective way to use AI to find the most undervalued stocks is to look for factors that influence stock price movements before matching appropriate stocks to them.
- In the first step, we focus on understanding the relationship between the high/low value of the factor under study for the stocks under consideration and the price movement of these stocks.
- Subsequently, we evaluate statistical significance using the P-value and T-statistic to select only the significant factors that we use to build the stock valuation model. There is no AI involved so far. However, this step already eliminates the basic problem of analysts evaluating stocks based on statistically insignificant factors and indicators.
These two points are the focus of our regular TAOTS. - In the next step, we can stack (ideally uncorrelated) factors into our model. This is where AI-based stock analysis comes in, as machine learning and its state-of-the-art methods should be used to select the best-performing uncorrelated factors and build robust stock strategies that work in most market situations. These investment strategies are what the StockPicking Lab provides. Read more in the article Stock Analysis with The Power of AI.