Stock-price outcomes for public insider transaction disclosures

1. Introduction

Publicly disclosed insider transactions — where executives, board members, and other key personnel change their holdings in their own company — are often studied as a possible signal of future stock performance. People close to a company may know more about its prospects, but their transactions can also reflect many ordinary personal or financial reasons.

The Swedish Financial Supervisory Authority (Finansinspektionen) mandates that all insider transactions be reported and published publicly. This transparency creates a rich dataset for academic and quantitative study.

Prior work in this project found a broadly balanced 24-week outcome distribution. This post takes one extra step: separate increased holdings from decreased holdings and see whether the two patterns look different.

Research Questions

• What is the overall distribution of 24-week stock-price outcomes?
• Do insider buys and sells produce different outcome distributions?
• Does the insider’s position (e.g. CEO, board member) affect the signal?

2. Data & Methodology

2.1 Data Sources

• Insider transactions: Finansinspektionen’s public register — reportable transactions by persons discharging managerial responsibilities (PDMRs) and their closely associated persons.
• Stock prices: Avanza weekly closing prices, used to compute forward returns.
• Company data: GLEIF identifier data supports company matching.

The records used here run through January 2026.

2.2 Transaction Classification

The project groups 27 observed transaction labels into Buy (Öka — increased holdings) or Sell (Minska — decreased holdings). These are convenient project-defined directions that include more than straightforward purchases and sales.

2.3 Outcome Binning

For each record, we compare the weekly closing price on the reported transaction date with the 24th later weekly observation. We use a five-percentage-point buffer around zero:

• Negative: at or below −5%
• Neutral: above −5% through +5%
• Positive: >+5%

This is simply a readable way to separate larger moves from a relatively flat middle band; it is not the only possible threshold.

2.4 Position Hierarchy

Finansinspektionen records the insider’s role, and a filing can contain more than one position (for example, a CEO who also sits on the board).

To assign each transaction a single, unambiguous role for filtering, we define a seniority hierarchy and resolve compound positions to the highest-ranked role:

1. CEO
2. Board Chair
3. Deputy CEO
4. CFO
5. Board Member
6. Board Deputy
7. Other Senior Officer
8. Other Admin Member
9. Employee Representative

For example, a filing under both CEO and Board Member is resolved to CEO. This keeps each record in one position filter.

3. Results

3.1 Sample Overview

The final dataset contains 16,557 public disclosure records with sufficient price history for this comparison, spanning 2023-01-01 to 2026-01-23.

3.2 Aggregate Distribution (±5%)

Figure 1 shows the overall outcome distribution. The negative and positive shares are close, so the aggregate view has no obvious one-sided pattern.

Figure 1

Aggregate 24-week outcome distribution

3.3 Buy vs. Sell Breakdown

Figure 2 disaggregates the distribution by transaction direction at the ±5% threshold. Insider buys show 39.4% positive outcomes versus 38.7% negative, while insider sells show 40.9% positive versus 47.9% negative. Only 19.6% of transactions fall in the neutral band, meaning the vast majority are followed by moves of at least 5% in either direction.

Use the filter below to explore all nine detailed positions or compare the four broader role categories. The displayed sample size changes with the filter.

3.4 Position-Level Analysis

The table asks a simple follow-up: within each role, do Buy and Sell outcomes look different beyond the overall comparison?

Note: Buy Pos. is the positive-outcome share after increased holdings; Sell Neg. is the negative-outcome share after decreased holdings. The exploratory chi-square tests compare each role's Buy/Sell × Negative/Neutral/Positive distribution at ±5%. Positions with fewer than 5 observations in either direction are omitted. The p-values are uncorrected across roles; Sig. uses α=0.05 and n.s. means not significant.
Source: Finansinspektionen's public register and Avanza weekly closing prices.

The exploratory tests detect differences for CEO, CFO, Board Member, and Other Admin Member. For the other displayed roles, this analysis does not detect a clear difference between Buy and Sell outcomes.

4. Discussion

4.1 Interpretation

The interesting tension is that the overall result is balanced while the direction-specific views are not. One possible explanation is that public information is already reflected in prices by the time readers see it, but this post cannot establish why the two directions differ.

Sales can happen for diversification, liquidity, or tax reasons, which may add noise rather than explain the larger negative share after decreased holdings. Buying may be easier to interpret as confidence, but that remains a possible story rather than a causal finding.

4.2 Position-Level Signal

The position filters show that the pattern is not identical across roles. Where the exploratory test is significant, this sample contains a detectable Buy/Sell distribution difference. Where it is not significant, the post simply did not detect a clear difference; smaller samples or smaller effects are both possible.

4.3 Limitations

• Publication Delay: Transactions are published with a regulatory delay (typically 1–3 business days), reducing the information advantage for real-time trading strategies.
• Confounding Factors: Stock movements are driven by many variables beyond insider activity — earnings, macro events, and sector rotation. Returns are not adjusted against a market or sector benchmark.
• Survivorship Bias: Companies that delist during the 24-week window may be underrepresented, biasing results.
• Transaction Costs: Brokerage fees, bid-ask spreads, and market impact are not accounted for.
• Sample Composition: The Buy/Sell ratio in the dataset may not be 50/50, which affects the relative weight of each group in the aggregate view.

5. Conclusion

This exploration goes one step beyond a single aggregate distribution by separating transaction directions and reported roles. Three patterns stand out:

1. The aggregate 24-week outcome distribution at ±5% is relatively balanced, with no obvious one-sided pattern.
2. Insider buys produce a distribution skewed toward positive outcomes, while insider sells have a larger negative share.
3. The observed Buy/Sell differences vary by reported role.

Future work should combine this directional signal with features such as transaction size, sector effects, timing, and a market or sector benchmark. Those extensions would test whether the associations here remain after broader market conditions are taken into account.