Every label, screen and published list on KeepWinning comes from a fixed rule applied to daily closing prices. No company is picked, promoted or removed by hand at any step, and nothing on the site is an opinion about what a price will do next.

How the trend label is computed

Two simple moving averages of the daily close are used: the 50-day (MA50) and the 200-day (MA200). The MA200 is called rising when it sits higher than it did 20 trading days earlier, and falling when it sits lower. The five labels follow from the last close, the two averages and that slope.

• Strong uptrend — the last close is above the MA50, the MA50 is above the MA200, and the MA200 is rising.
• Uptrend — the last close is above both averages, without meeting the strong-uptrend conditions.
• Neutral — the close sits between the two averages, or the averages have crossed.
• Downtrend — the last close is below both averages, without meeting the strong-downtrend conditions.
• Strong downtrend — the last close is below the MA50, the MA50 is below the MA200, and the MA200 is falling.

A company with fewer than 52 daily closes gets no label rather than a guessed one. Where there is enough history for the MA50 but not yet the MA200, the label falls back to the close against the MA50 alone.

The market regime

The same idea applied to an index rather than a company: bull trend when the index is above its MA200 and that MA200 is rising, neutral when it is above but not rising, and bear otherwise. The regime is shown first, on purpose — it is the top rung of the ladder the whole product follows: market, then stock trend, then fundamentals.

The screens

Each market page publishes three screens. They are pure filters over that market’s tracked companies: every company meeting the condition is included, in rank order, and none is added or removed for any other reason.

• Above MA200 — the last close is above the 200-day average.
• Uptrend — the trend label is Uptrend or Strong uptrend.
• Near 52-week high — the last close is within 5% of the highest close of the past 52 weeks.

Data and update cadence

Price history and reported quarterly figures come from third-party providers and are rebuilt nightly after the last covered market closes, with intraday quote refreshes during trading hours. Every generated page carries the date of the data it was built from. Quarterly revenue and earnings change only when a company reports.

Deep quarterly earnings history is derived rather than read straight off a single field: where a provider gives earnings per share and a share count but no absolute earnings line, the quarter is reconstructed from those two, and the result is dropped when it fails a sanity check against the reported figures. That is why a company can show a full revenue history and a shorter earnings one.

What this screener gets wrong

A mechanical rule is only useful if you know where it fails. These are the failure modes of this one, stated plainly rather than buried in a disclaimer.

• Moving averages lag. A trend label describes what a price has already done. By the time a stock is labelled Strong uptrend, a large part of the move that earned the label has happened, and a label can still read Uptrend on the day a top forms.
• Choppy markets produce whipsaws. When price oscillates around the MA50 or the two averages sit close together, the label flips between Neutral and Uptrend on small moves. Those flips are noise, not signal, and the rule has no way to tell the difference.
• The rule ignores everything that is not price. Earnings surprises, takeovers, dilution, fraud, delisting risk and sector rotation do not enter the label at all. A company can be in a textbook uptrend on the day the news turns.
• Fundamentals coverage is uneven and thin for small caps. Deep quarterly history exists for well under half of the companies tracked; many UK, French, Benelux and Swiss issuers report semi-annually and so have no quarterly series at all. A card with no fundamentals block means the data is missing, never that the numbers are bad.
• The data can be late or wrong. Prices are delayed, corporate actions are sometimes adjusted late, and a trend label computed from a late-adjusted series stays wrong until the next nightly rebuild corrects it.
• No valuation, no forecast, no advice. There is no fair-value model, price target or earnings estimate anywhere in the product. A screen result is a list of companies that met a condition, and nothing more.

More: Data sources and update cadence, Reading MA50/MA200 trend labels, Trend distribution by market