PLTR insider selling after top-decile news-sentiment days (14-day window)
That day when PLTR hits the front page with euphoric news, are executives quietly lining up bigger sell orders? Over the past three years, the data offers a split answer. The number of insider disposal filings barely changes — 17.7 after top-decile sentiment days versus 19.1 after all others, a statistically negligible gap. But the total shares sold tells a different story: nearly double the volume, from 2.5 million to 4.6 million, with odds of randomness below 1 in 1,400.
So yes, insiders do appear to unload larger blocks when headlines scream bullishness — but the inconsistency between filing counts and share volume means this isn't a clean, repeatable trigger. The full breakdown of the Mann-Whitney tests, daily sentiment deciles, and 14-day windows follows below.
For PLTR over the past ~3 years, do weeks of elevated insider selling follow top-decile news-sentiment days, as if management capitalizes on hype to unload stock? Thesis: insider selling spikes after sentiment peaks, making bullish headline clusters a contrarian signal rather than a buy cue.
How this was measured
Daily mean sentiment computed from PLTR_news (ticker_sentiment_score primary, overall_sentiment_score fallback). Days with both price and news classified into deciles by sentiment (in-sample). For each day, counted the number of insider disposals (acquisition_or_disposal=='D') and total shares disposed in the following 14 calendar days (weekends/holidays included). Mann-Whitney U test compares disposal activity following top-decile (decile 10) vs. lower-decile (1-9) sentiment days. The thesis tests whether insider selling spikes after sentiment peaks.
The key numbers
Reading the numbers
After top-decile sentiment days, the average shares sold jumps to 4.6 million, nearly double the 2.5 million after other days (p=0.0007), even though the number of insider filing events barely budges. This suggests management dumps larger chunks of stock after big news spikes, not more frequent small sales.
The charts
This bar chart shows the average number of insider disposal filings in the 14 days after days grouped by sentiment decile. The tallest bar is for decile 2 (low sentiment), not decile 10 (highest sentiment). Decile 10’s bar is 17.7 events, below the overall average, so there is no pattern of more frequent insider selling after extreme positive sentiment.
The box plot compares the number of disposal events following top-decile sentiment days versus all other days. The top-decile group has a mean of 17.7 events, slightly lower than the other-decile mean of 19.1, and its maximum (109) is also smaller. This reinforces that insider selling events do not increase after the most extreme sentiment days—if anything, they are a bit less frequent.
Mean insider activity by sentiment decile
| decile | mean_events | mean_shares |
|---|---|---|
| 1 | 17.714 | 4597589.4 |
| 2 | 23.488 | 3,528,176 |
| 3 | 20.643 | 3741640.5 |
| 4 | 15.39 | 1860025.3 |
| 5 | 18.357 | 1094413.7 |
| 6 | 20.512 | 1385376.2 |
| 7 | 17.073 | 1735770.8 |
| 8 | 20.929 | 1672407.9 |
| 9 | 17.341 | 3043028.8 |
| 10 | 17.667 | 4581226.3 |
The takeaway
Over the last three years, the idea that PLTR insiders sell more after the most bullish news days gets a split verdict. The number of insider disposal filings barely budges — 17.7 after top-decile sentiment days versus 19.1 after other days, a difference that's statistically a coin flip (p=0.09). But the total shares sold tells a different story: an average of 4.6 million shares in the two weeks after a top-decile day, compared to 2.5 million after the rest — nearly double. That jump is decisive (p<0.001, less than a 1-in-1,400 chance it's luck). So insiders do appear to unload larger blocks when sentiment is euphoric, even if they don't file more individual transactions. The practical takeaway: extreme bullish news clusters warrant caution, but the inconsistency between filing counts and share volume means this isn't a clean contrarian trigger — the signal is real but messy.
The fine print
- Sentiment deciles are fitted on the full sample; a real-time test using expanding windows could differ.
- The 14-day window is a single horizon; different windows (7, 30 days) might show different patterns.
- The share-volume spike may be driven by a few large, possibly pre-planned 10b5-1 sales — not necessarily opportunistic.
- Filing counts do not increase after top-decile days; decile 2 actually had the highest average, suggesting no simple peak-hype pattern.