Open the Political Alpha trade feed on a random afternoon and the live ticker will mostly be red. At capture it was showing four consecutive sells from the same senator, four different symbols, every one of them in the $1,001 to $15,000 band. Scroll to the sector heatmap and the pattern holds at the aggregate level. Eight of the ten sector cards on screen carried a Sell label, with Information Technology showing 81 buys against 106 sells over a thirty day window and Financials showing 34 buys against 54 sells.
The instinct is to read that as a cohort turning defensive. It is almost never that. Sells dominate this feed structurally, for reasons that have nothing to do with anyone's view on a company, and if you have ever felt a small jolt of alarm at seeing a legislator dump a stock you own, that jolt was probably free of information.
Buying is a decision and selling frequently is not
The asymmetry is the whole thing. To buy a stock, somebody has to choose that stock over every other stock and commit cash to it. There is exactly one common reason to do that. To sell a stock, there are a dozen reasons, and only one of them is a view about the company.
You need cash. It is December and you have losses to realise. A managed account rebalanced. A holding vested. A trust was restructured. An account moved custodian. An inherited position was liquidated. A divorce, an estate, a house. A conflict was resolved by exiting the position. Every one of those produces a disclosure that looks identical on the tape to a considered exit, and the tape carries no field distinguishing them.
This is a well established asymmetry in the broader literature on corporate insider transactions, where the durable finding is that purchases carry more information than sales for exactly this reason. The mechanism transfers cleanly to legislative disclosure. It is the same structural fact about human beings and money.
The four-question test before you read anything into a sell
This takes about two minutes per filing and it kills most of them. Run it in order, because the first question eliminates the majority.
Is it alone or is it in a batch? Pull up the filer's disclosures for that date. If the same date carries several unrelated tickers, you are looking at one administrative event that produced several rows. That is the ticker pattern in the screenshot above. It is not four signals, it is one non-signal replicated four times.
Is it the whole position or a slice? A partial trim in a name somebody keeps holding is portfolio maintenance. A complete exit from a name that had been held for a long time is at least a different kind of event. The per-member profile pages carry full trade history, so you can see whether the name persists in their filings afterwards.
What is the calendar doing? Late in the tax year, sells cluster for reasons that have nothing to do with markets. Same for the weeks around a new session or a change in committee assignments, when conflict-driven cleanup happens. If a sell falls in one of those windows, discount it heavily.
Does this person sell often? Someone with hundreds of transactions a quarter is running an active account or has someone running it for them, and any individual sell of theirs carries close to zero information. Someone with four transactions a year selling a long-held position is a genuinely different event. The leaderboard shows total trade counts per member, and the spread is enormous, running into the thousands for the busiest filers.
The one sell pattern that survives the test
What is left after all that is narrow, which is the correct outcome. The pattern I still pay attention to is a complete exit, by a low-frequency filer, in a name they had held for a long time, disclosed quickly, and not accompanied by other unrelated transactions on the same date.
That combination is rare enough that you might see a handful in a year, and rare is what gives it any weight at all. Even then, treat it as a reason to go and look at the company rather than a reason to do anything to your position. The disclosure tells you a person exited. It does not tell you why, and reconstructing why from a filing is a story you are telling yourself.
What the sell side is genuinely useful for
There are two jobs the sell data does well, and neither is a trade signal.
The first is measuring the cohort's overall posture, which the sector heatmap does at a glance. The aggregate skew towards sells is itself a baseline, and departures from the baseline are the interesting part. At capture, Communication Services and Materials were the two cards showing net buying, with 30 buys against 24 sells and 23 buys against 20 sells respectively. Those are small margins, but they are the ones that stand out against a tape that leans the other way everywhere else, and that contrast is more informative than any individual row.
The second is as a filter on your own enthusiasm. If you are about to act on a cluster of purchases in a name, it is worth checking whether the same cohort has been quietly selling it elsewhere. A name being bought by two members and sold by five is not the consensus you thought you had found.
The mistake this actually prevents
The behaviour I want you to avoid is specific. You hold a position. You see a disclosure that a legislator sold it. You feel a flicker of doubt, and over the following week that flicker becomes the reason you close a position you had good reasons to own.
That sequence is expensive and it happens constantly, because a red row with a name attached is emotionally louder than a spreadsheet. The defence is mechanical rather than emotional. Run the four questions. If the filing does not survive them, it is administrative noise, and it should have exactly no weight in a decision about your money. The average delay between the trade and its disclosure was 32.5 days at capture, so the price already did whatever it was going to do about this a month ago, and you are reacting to a photograph rather than an event.