2024-2026, healthcare and technology, where insider flow became more selective
The later part of the period is less about one giant sector call and more about discrimination within sectors. That is healthy. Real signals become subtler as markets normalise.
Healthcare, fertile ground, but only in the right subsectors
Healthcare often attracts insider buying after biotech drawdowns, reimbursement scares, or device-cycle disappointments. The problem is that healthcare is not one sector in any economically coherent sense. Large pharma, managed care, medtech, tools, and pre-revenue biotech belong in the same GICS bucket but not in the same valuation universe.
That means sector-level net buying intensity can be informative only when the buying is broad enough, or when the article is honest enough to admit it is really an industry-group signal wearing a sector badge.
From 2024 onward, healthcare looked more like a selective hunting ground than a blanket rotation. Insider buying in profitable medtech or cash-rich therapeutics firms after deratings is different from serial insider support in speculative biotech. The former can lead recoveries. The latter can simply lead to more filings.
Technology, the signal weakens when compensation muddies the water
Technology presents the opposite problem. It is a large, liquid sector with substantial insider activity, but much of that activity is contaminated by stock-based compensation, scheduled sales, and founder control structures. Open-market buying by senior executives in large-cap technology can still matter, but it is rarer and often drowned out by mechanical selling.
This makes technology a poor candidate for naive sector-level insider models. If you want a useful signal, you usually need to isolate semiconductors, software, or hardware separately, exclude founders with super-voting control, and focus only on discretionary purchases.
The dry version is that the sector is noisy. The less dry version is that a chief executive receiving another grant of stock options does not count as "bullish conviction", however enthusiastically investor relations phrases it.
Building the indicator properly, what to measure and how to test it
If the goal is to use insider flow as a leading indicator for sector ETF performance, the methodology needs to survive a referee report and a bad month.
A workable definition of net buying intensity
At minimum, define sector net buying intensity as:
[
\text{NBI}{s,t} = \frac{\sum \text{Open-market buy value}{s,t} - \sum \text{Open-market sell value}{s,t}}{\text{Sector market cap or ADV}{s,t}}
]
Then standardise this within each sector over a trailing history to create z-scores. A one-standard-deviation burst in utilities is not the same thing as one in technology.
Better still, build variants:
- Issuer breadth: share of sector constituents with at least one qualifying purchase.
- Insider breadth: number of unique officers/directors buying.
- Role-weighted intensity: heavier weights for CEOs, CFOs, and chairs.
- Cluster score: purchases occurring within a short window after a drawdown.
The cluster score is especially useful. Insider buying after a 25 percent sector drawdown is not equivalent to insider buying after a 5 percent dip.
Match the signal to investable vehicles
For implementation, pair each GICS sector with a liquid sector ETF or futures proxy. In the US context, the SPDR sector ETFs are the obvious baseline. Then test forward returns over 1-, 3-, and 6-month horizons after signal formation.
The usual pattern in the literature is that insider buying has predictive power over intermediate horizons rather than the next afternoon. That makes sense. Information diffuses, narratives catch up, and valuation gaps close gradually.
Use relative returns, not only absolute returns
A sector signal should be judged against alternatives. If energy rises 8 percent after a positive insider signal while the market rises 12 percent, the signal did not help a rotation investor much. Evaluate excess returns versus the broad market and versus sector peers.
This is also where the "leading indicator" claim becomes testable. If insider flow spikes before analyst revisions and before relative performance inflects, you have a plausible lead. If it spikes after the ETF has already rerated, you have discovered a lagging indicator with excellent public relations.
What the 2020-2026 cases suggest, and what they do not
The case studies point in one direction, with caveats attached like legal footnotes.
Where the signal appears strongest
The best sector-level insider signals tend to appear in sectors with four traits:
- recent severe drawdown,
- cyclical or balance-sheet-sensitive fundamentals,
- common macro driver across firms,
- relatively clean open-market purchase data.
That favours energy, financials, industrials, and selected healthcare groups after stress events. It is less favourable for sectors dominated by compensation-driven selling, conglomerate-like composition, or founder control.
Where investors overstate the edge
There are three common exaggerations.
First, that insider flow predicts every turn. It does not. It is most useful near extremes.
Second, that more buying is always better. Not necessarily. Repeated buying into a value trap can reflect genuine conviction and genuine error at the same time. Executives are informed, not infallible.
Third, that sector aggregation eliminates all noise. It reduces noise. It does not abolish it. A sector can show strong buying because a handful of distressed firms are averaging down with board support while the healthier names do nothing.
The missing proprietary numbers
Because no fresh data pull was provided from our 162k filing archive, I cannot tell you that energy net buying intensity hit x standard deviations in y month, or that it led XLE by z basis points over the next quarter. That would be the point where many articles become imaginative. We will remain unfashionably sober and write n/a.
Still, the absence of a live extract does not erase the framework. It simply means the next step is empirical rather than literary.
A practical research agenda for allocators
The sensible way to use this idea is not as a stand-alone trading oracle, but as a ranking input in a sector allocation process.
A simple implementation sketch
Each month:
- collect qualifying insider purchases and sales by issuer,
- map issuers to GICS sectors,
- compute sector net buying intensity and breadth,
- rank sectors by standardised score,
- combine with valuation and price trend filters,
- allocate to the top-ranked sectors, market-neutral or long-only.
The valuation filter matters because insider buying in expensive sectors often has weaker subsequent payoff. The trend filter matters because catching falling knives is a hobby, not a process.
What to test before using real money
At minimum, test:
- 30-, 60-, and 90-day formation windows,
- 1-, 3-, and 6-month holding periods,
- equal-weighted versus role-weighted purchases,
- crisis versus non-crisis subsamples,
- US-only versus cross-market signals under MAR and SEC regimes.
The cross-market angle is underexplored and attractive. European PDMR disclosures under MAR are not as tidy as US Form 4 data, but they are timely enough to support a sector model. France, in particular, offers a useful test bed because AMF-regulated disclosures sit within a harmonised EU framework while still requiring careful issuer mapping.
