Research that keeps score.
DeepCap is an independent equity research boutique based in Finland. We build tools that help asset managers, family offices and serious private investors rank opportunities, monitor positions and test their convictions before the market does.
Where we started
In 1998, two Stanford graduate students published a paper describing how to rank the entire web. PageRank did not try to understand what every page meant. It asked a narrower and better question: if someone wandered the web at random, following links, how likely were they to land on this page? The answer was a probability. Ordering pages by it gave Google its early edge over search engines that mostly counted keywords.
DeepCap began with the same instinct applied to stocks. Nobody reliably predicts where a share price will be next year. We asked a different question: how likely is each stock to beat the median of its universe over the next six or twelve months? We then ranked every stock by that answer. A probability can be checked against what happens. A price target mostly gets revised.
Our first ambition was to put that ranking in the hands of everyday investors, and DeepCap Personal still does. Working with professional investors, though, taught us something no model could. Picking the stock is rarely where the serious damage happens. It happens in the months after the purchase, when the original reason for owning a company erodes and nobody wrote down what “wrong” would look like.
The problem we solve
Every portfolio manager has held a position past the point where the thesis broke. The cause is rarely a lack of intelligence. The evidence tends to arrive slowly and in fragments, while the position is already on the books and the story has already been told to the committee. Behavioural finance has studied this for forty years under names like the disposition effect and escalation of commitment. Knowing the names has not made the habit rarer.
Foresight, our flagship platform, brings that discipline to concentrated equity portfolios.
Monitoring against your own rules
Every position is checked after each market close against the exit criteria you set at entry. When a thesis breaks, you get a flag rather than a debate.
A calibration record
Forecasts are written so they can be proven wrong, then resolved against what actually happened. Over time this builds a private record of how often your 70% convictions come true.
The stress and pre-mortem lab
Stress tests and pre-mortems happen before capital is committed. In a pre-mortem, the team imagines the position has already failed and works backwards to find out why. In development, November 2026: scenario tests that shock the drivers your claims depend on.
Foresight is available for pilot use. Monitoring against your own rules and the calibration record are live today. In the stress and pre-mortem lab, the pre-mortem is live, with a first release of historical stress tests and earnings reactions. In development, November 2026: scenario tests that shock the drivers your claims depend on.
Our mission
To make investment conviction accountable. Every position is held for a stated reason, every reason is tested against evidence, and every forecast is scored against the outcome.
Good research should also not stop at the S&P 500. Our models rank stocks across Europe, the US and Asia. The investors we serve, from the Nordics to Singapore, own companies well beyond Wall Street.
The standards we hold ourselves to
Probabilities, not predictions
We state the likelihood of an outcome and the horizon it applies to. We do not publish price targets.
Point-in-time data
Look-ahead bias is the most common way backtests flatter themselves. Since June 2026 our models date company accounts by filing date, estimated for European companies, although later restatements can still replace the original figures. Since October 2026 they no longer use P/E, beta or similar figures that carried later values into past dates.
Simulations labelled as simulations
Our model portfolios are simulations, and the Performance page says so, with each start date, the trading costs and what each figure excludes stated beside the charts.
Every AI claim traced to its source
When our language models extract statements from a document, each statement must match the document text, or it is discarded.
Your theses stay yours
Client theses are stored encrypted, and your calibration record is visible only to you. Your theses are never used to train models serving other clients.
Independence
DeepCap manages no money, earns no commissions and takes no payment from the companies it covers.
Who we work with
Asset managers and investment teams
Foresight, our datasets and bespoke research add discipline to concentrated portfolios.
Family offices
Foresight holds external managers and internal decisions to the same written standard.
Private investors
DeepCap Personal offers ranked research at a subscriber's price.
DeepCap is deliberately small. Questions go straight to engineering, not to a sales desk, and a request for a new market, screen or report changes the product rather than joining a queue.
DeepCap Oy, Finland · DeepCap Pte. Ltd., Singapore. DeepCap provides research and tools for informational purposes and does not provide investment advice.