Insider trading often refers to an illegal act. But senior insiders in listed companies can buy or sell their stock for a host of reasons. Our StarMine research team analyses how outside investors can leverage two StarMine models to interpret the “informativeness” of these legal insider trades and design profitable trading strategies.
- In theory, an outsider can analyse the SEC filings (or website postings) to unravel the private information they reveal about these trades. But it is not an easy task: insider transaction data is highly complex.
- Discover how to assess the informativeness of legal insider trading data and leverage it to design a profitable trading strategy, via StarMine.
- Our research shows that combining insider sentiment with price momentum signals can help you gauge the informativeness of the insider trades.
For more data-driven insights in your Inbox, subscribe to the Refinitiv Perspectives weekly newsletter.
It is widely agreed that insiders – i.e., executives – rely on private information (such as stock repurchase plans, change in dividends, stock splits, auction, a take-over bid, or public offering, etc) when they trade equity in their own firms. We explain how outside investors can leverage two StarMine models to interpret the “informativeness” of the insider trades and design profitable trading strategies.
Insider informativeness
We consider insiders as value investors, wanting to buy their company’s stock when they think it is cheap and sell it when they think it is overpriced. However, not all insider trades are informative. Executives may trade for other reasons as well – purchasing shares to meet their holding requirements and selling them to meet their liquidity needs and diversify their exposure to firm-specific risk.
Insider trading is highly regulated in the US. Since the enactment of the Sarbanes-Oxley Act (SOX) in 2002, insiders are required to report trades to the Securities and Exchange Commission (SEC) – electronically, via forms 3, 4, 5 and 144 – within two days and firms are required to post those filings on their website.
In theory, an outsider can analyse the SEC filings (or website postings) to unravel the private information they reveal about these trades. But it is not an easy task: insider transaction data is highly complex. Transactions are categorised by insiders themselves according to a scheme incorporating 20 categories – and a significant fraction of insider trades are in non-common equity (preferred shares) and derivatives (options to purchase common stock).
So how can you assess the informativeness of insider trading data and leverage it to design a profitable trading strategy?
We solve for this challenge by combining two StarMine models to allow you to unlock the power of insider trading data to inform your own trading strategy.
In practice
The StarMine Insider Filings model (Insider) is a stock-ranking model for US equities that combines the complex insider holdings and insider trading data to determine an overall sentiment of the insiders towards their company. Underlying the proprietary algorithm is the basic hypothesis that insider purchasing reflects bullish sentiment while insider selling reflects bearish sentiment. The StarMine Price Momentum model (PriceMo) is a stock-ranking model that improves upon the standard momentum anomaly implementation by combining information from multiple dimensions of price momentum, from the negative autocorrelation in short-term returns to the positive autocorrelation in intermediate to long-term returns.
Our research shows that combining insider sentiment with PriceMo signals can help us gauge the informativeness of the insider trades. Our hypothesis is that when insiders start buying a stock that has gone up in price it suggests a strong bullish insider sentiment. Conversely, when insiders begin selling a stock that has already dropped in price, it is reflective of a strong bearish insider sentiment. As stated in the original whitepaper, How Can Outsiders Benefit From The StarMine Insider Filings Model? to keep each model unique and reduce correlation we do not include any price information in the insider model.
We analyse two easy to implement approaches that can be used to construct insider informative portfolios. StarMine provides model scores as ranks between 1 and 100 with 1 representing a bearish and 100 representing a bullish score.
- Insider with PriceMo Blending: In this approach, first a blended score is calculated as:
blended_score = a x Insider_score + (1-a) x PriceMo_score,
where a is a coefficient in the range [0,1] that controls the relative contribution from the two signals. Then top (bot) N securities are selected based on the blended_score to build the portfolio. The blending scheme requires a non-NULL value for the Insider score. When PriceMo score is not available for a security, it can either be excluded from the investible universe, or assigned a score of “0” before blending.
- Insider with PriceMo Screening: In this approach, the portfolio is built in three steps:
a. Insider-based selection: Select top (bot) N1 stocks based on the Insider score.
b. PriceMo-based screening: Drop securities where PriceMo score is smaller (greater) than 50 or is not available.
c. Final Selection: Order the list obtained in the previous step by Insider and PriceMo scores simultaneously (multi-variable ordering with Insider being the first variable) and select top (bot) N stocks to build the portfolio.
N1 can be varied to build the desired size (N) portfolio.
We expect the two approaches to produce highly correlated portfolios (holdings). They are considered to test the sensitivity of the portfolio performance to minor perturbations in the holding.
Cumulative return
Figures 1 and 2 show the cumulative return of 50 stock (N=50) PriceMo blended and PriceMo screened insider informative portfolios constructed over a representative large-cap and small-cap US equity universe, respectively. The large-cap universe is constructed using the largest 1000 securities by market capitalisation and the small-cap universe is constructed by selecting smallest 2000 securities from the largest 3000 securities.
The portfolios are rebalanced monthly, and their performance is compared against an equal-weighted market portfolio over the Jan-2013 to Dec-2022 period. We chose a = 0.8 (20% PriceMo blending) to build PriceMo blended portfolios. Also, when the PriceMo score is not available, the security is dropped from the final selection in both the approaches. Performance statistics are reported in Table 1.
Table 1: Performance statistics of monthly rebalanced PriceMo blended and PriceMo screened insider informative portfolios.
Discover the benefits of insider informative portfolios
Our analysis shows that insider informative portfolios consistently outperformed the equal-weighted market portfolios over 2013-2022 period, across both large-cap and small-cap universes. For a complete analysis on how to combine the StarMine Insider Filings (Insider) with the StarMine Price Momentum (PriceMo) models, read the full report here: How Can Outsiders Benefit From The StarMine Insider Filings Model? by Sanghi, T., 2023.
