The thing that got me interested in insider data was noticing how boring it is most of the time and how loud it gets at exactly the moments you care about. A company officer files a form after they buy or sell their own stock, and on any given day those filings are a wall of noise. Options exercises, scheduled sales, tax stuff, none of it telling you much. But if you stop looking at individual filings and instead roll every insider transaction across the whole market into a single number, something useful falls out. Insiders as a group tend to get greedy near the exact spots where everyone else is throwing up their hands.
I want to walk through how that number gets built, why it reads so much better at bottoms than at tops, and where I actually pay attention to it. None of this is a magic timing bell. It is more like a slow, contrarian pulse that occasionally screams.
How the ratio actually gets built
The raw material is public. In the US it is Form 4, the filing a director, officer, or ten-percent owner has to submit shortly after they transact in their own company's shares. Other markets have their own versions on their own clocks, but the idea is the same. You have a timestamped record of the people who know the business best putting their own money in or taking it out.
To turn that into a market-wide indicator you aggregate. The two common ways to do it are by count and by dollars. A count-based ratio just tallies the number of buy transactions against the number of sell transactions over some window, usually a week or a month. A dollar-weighted version sums the value of buys against the value of sells. Both have a place. The count version treats a small buy from a regional bank director the same as a huge buy from a mega-cap CEO, which sounds like a flaw but actually cuts noise, because a handful of enormous scheduled sales can swamp a dollar-weighted number.
Whichever you pick, the output is a ratio. Something like buys divided by sells, or buys as a share of total transactions. The absolute level matters less than where it sits relative to its own history, because the baseline is not one to one. Insiders sell far more often than they buy in normal times, so a healthy market can sit at a ratio that looks bearish if you assumed balance. You have to read it against its own range, not against fifty-fifty.
Why sells lie and buys don't
Here is the asymmetry that makes the whole thing work. An insider can sell for a hundred reasons that have nothing to do with the stock. They are buying a house. They are diversifying because ninety percent of their net worth sits in one ticker. They have a scheduled selling plan set up months earlier that fires automatically regardless of price. Divorce, taxes, a kid's tuition. When you see heavy insider selling, you genuinely cannot tell whether it means the person thinks the stock is expensive or just that they wanted the cash.
Buying is different. There is basically one reason an executive takes their own already-concentrated money and puts more of it into the same stock they are already massively exposed to, and that reason is they think it is going up. Nobody buys their own company's shares in the open market to diversify. So a buy carries a signal that a sell mostly does not. This is why the ratio is a much sharper tool at bottoms, where the story is about buying, than at tops, where the story would have to be told through selling that is drowning in noise.
It also explains a failure mode I see people walk into. They watch insider selling ramp up during a strong bull run, decide the smart money is heading for the exits, and short into it. Then the market keeps going for another year. The selling was mostly people trimming a position that had grown too large, which is what you would rationally do with any asset that ballooned. Reading heavy selling as a top signal is the single most common way to misuse this data.
What the ratio looks like at real bottoms
Historically, the readings that mattered showed up when things felt awful. Deep market washouts, the moments when the financial press is asking whether the whole system is broken, tend to coincide with insiders quietly stepping up their open-market buying. You get clusters. Not one CEO here or there, but a broad wave across sectors and company sizes all buying in the same short window. The ratio spikes well above its own normal band, and it usually does so a little before or right around the price low rather than at the exact tick.
That timing texture matters. The insider crowd is early and patient, not precise. They are buying because their stock got cheap relative to what they know about the business, and they are fine being underwater for a while. So the ratio is a terrible tool for calling the bottom to the day and a pretty good tool for telling you the odds have shifted in favor of buyers over a multi-month horizon. If you need an entry this afternoon, look elsewhere. If you are trying to decide whether to be adding into weakness over the next quarter or two, it earns its place.
How I actually read it
A few rules of thumb I keep in mind, none of them precise, all of them earned by getting them wrong first.
- Read the level against its own trailing range, not against a fifty-fifty baseline. A move from the bottom of its normal band to the top is the event, not the raw number.
- Weight buys far more than sells. Treat a surge in buying as information and treat a surge in selling as mostly noise until proven otherwise.
- Look for breadth. A cluster of buying across many companies and sectors is worth more than a big number driven by one or two names.
- Give it room. This is a weeks-to-quarters signal. Pairing it with a fast trigger and expecting a same-day turn is how you get chopped up.
- Cross-check with price. When the ratio spikes into a market that has already sold off hard and is starting to stabilize, that combination has historically been the meaty part. A spike into a market still in free fall can be early.
The workflow I use is simple. When I see the aggregate ratio push toward the high end of its historical range, especially after a broad drawdown, I take it as a cue to start scaling into positions I already wanted, not as a cue to reverse everything. I let the buying breadth confirm it. And I mentally throw out almost everything the same indicator tells me during calm uptrends, because in those stretches the sell side is doing all the work and the sell side barely means anything.
The honest limitation is that this is a coarse instrument. It will occasionally spike during a decline that has more room to fall. What it does reliably is stop you from being maximally bearish at the exact moment the people who run these companies are quietly getting greedy, and over a long enough run, not making that mistake is worth a lot.