Three readings (explained below) each look at a different thing about a stock and either point up, point down, or have no view: the past year’s trend, the reaction to its latest results, and what insiders and short sellers have done lately. The outlook is whichever side more of them point to. If they split, or none has a view, the outlook is Mixed / no clear outlook.

Agreement is said as a count - “2 of the 2 readings with a view point up” - and the card also says how many had no view. It is not a probability that the stock will go up, and it isn’t a measure of how “good” the stock is. One reading with a view is 100% agreement - of one - which is the thinnest evidence there is, and the card says so.

Does a Positive outlook mean it is time to buy? Not on its own. It means more of the readings currently point the same favourable way - it says nothing about whether the company is financially sound, whether the shares are reasonably priced, or whether a trade suits your own circumstances. Treat the outlook as a starting point for research: read the sentences underneath it, look for evidence that cuts the other way, and keep in mind what the readings don’t cover at all: valuation, debt, competitive position, management quality.

Why do different parts of a stock page seem to disagree? They look at different timeframes, and each says which. Last few weeks is trending or range-bound over about three weeks; Where the chart is is a pattern on the chart today; the outlook’s trend reading is the past year. A stock can drift sideways for three weeks inside a year-long fall - both are true at once.

Read the full page on outlook and agreement →

All three readings have a view, and they agree. One reading on its own is 100% agreement on the thinnest evidence there is, so it is never labelled strong; three separate things lining up is a different fact. Mixed / no clear outlook is never labelled strong or weak.

A strong label means the outlook rests on breadth, not that it is more likely to be right - every outlook, strong or not, is one input for your own judgment, not a recommendation.

One box per reading - Trend, Results day, Insiders & short sellers - with its clock underneath (past year; 60 trading days after results; last 90 days) and which way it points today: up, down or no view. Green and red boxes are the readings with a view. It is the outlook’s arithmetic at a glance; the sentences beside it say what each reading measured.

Before 8 October 2026 this was a four-point shape (the Signal Radar) built from RSI, MACD, the moving averages and volume. A stock not yet re-read since then still shows that shape until its next refresh.

Until 8 October 2026 each card had a four-point shape for RSI (top), MACD (right), trend (bottom) and volume (left); a bigger wedge meant that indicator had more to say. Those four no longer vote on the outlook - see the three readings for why - and the shape is replaced by the three reading boxes as each stock is re-read.

Every time signals refresh, the call generated for a stock is added to a permanent log rather than overwriting the last one - "View history" plots that log as a small timeline. Time runs left to right; the lane a dot sits in shows whether that day's call was Positive outlook (top), Mixed / no clear outlook (middle), or Negative outlook (bottom); the size of the dot shows how much agreement among the readings that call had; and a thicker ring around a dot means it cleared the bar for a High agreement call (see the entry above).

This is a record of what the platform has said, not a chart of the stock's price - a string of Positive-outlook dots doesn't mean the price went up, it means the same call kept firing each time signals refreshed. Hover any dot for its exact date, its agreement, and the price recorded at that signal. A brand new holding will only show one dot until more refreshes happen - that's expected, not an error.

A long run of Positive-outlook dots isn't proof the signal works, either - it can simply mean the same underlying condition (price staying above its moving average, say) kept being true across several refreshes, not that ten dots were ten separate, independently-won calls. Judging whether a signal is actually predictive needs a separate study of what happened to the price after each call, over a defined window, including every call that didn't pan out and realistic trading costs - not a glance at how many dots lined up in a row.

Every reading and chart figure on this site is a fixed, published calculation, and each one has a record: in academic research, and in our own count on the stocks people here actually hold. Both are below, so a reason can be weighed by its record rather than its reputation. This is what the studies and the count found. It is not a recommendation about any share.

Sources: Results-day drift — Bernard & Thomas, Journal of Accounting Research 1989; Martineau, Critical Finance Review 2022 for the recent weakening. Short positions — Desai, Ramesh, Thiagarajan & Balachandran, Journal of Finance 2002; Boehmer, Jones & Zhang, Journal of Finance 2008. A stock’s own year-long trend — Moskowitz, Ooi & Pedersen, Journal of Financial Economics 2012. RSI, MACD and crossovers — Brock, Lakonishok & LeBaron, Journal of Finance 1992; Sullivan, Timmermann & White, Journal of Finance 1999. Market dimmer — Faber, Journal of Wealth Management 2007. Insiders — Lakonishok & Lee, Review of Financial Studies 2001. Year-long trend — Jegadeesh & Titman, Journal of Finance 1993; Daniel & Moskowitz, Journal of Financial Economics 2016 for the weak spot. 52-week high — George & Hwang, Journal of Finance 2004. Steadiness — Ang, Hodrick, Xing & Zhang, Journal of Finance 2006; Baker, Bradley & Wurgler, Financial Analysts Journal 2011. Our golden-cross count: a one-off measurement Trendentum ran in September 2026 on 44 stocks and six years of daily prices, counted the way the backtester counts.

Two measures of the same thing - where the price has been heading over about a year - that must agree. The first: is the last close above or below the average of the last 200 closes? The second: is the price higher or lower than a year ago, leaving out the most recent month (short-term moves tend to partly reverse, so the studies leave that month out)? Both up reads up; both down reads down; one each way is no view. It changes a few times a year, not a few times a week.

What the record says. The most tested idea in finance: shares that rose over the past year have tended to keep rising for some months, across many countries and decades. Its known weak spot is a sharp market rebound after a fall. The evidence table has the sources.

How the price reacted to the company’s latest results: the move from the close before the results date to the first close after it, compared with an ordinary two-session move for that stock over the previous six months. A move at least twice the ordinary size counts, in its own direction, for 60 trading days; a smaller one has no view. Results dates come from Twelve Data’s earnings calendar; a stock with none on file has no view.

What the record says. After a large results-day reaction, prices have historically tended to keep drifting the same way for weeks. Smaller in recent years among the biggest US companies; still visible in smaller ones. The evidence table has the sources.

What people with money at stake have done lately. Directors buying shares on market in the last 90 days, by more value than they sold, reads up. For ASX stocks, a reported short position of at least 3% of the shares that rose by a full point between ASIC’s reports reads down. Both at once is no view. Directors’ sales never count: people sell for tax, a house or diversifying, and studies find those sales carry little information.

What the record says. Directors’ purchases have carried information; sales mostly have not. Large, rising short positions have tended to be followed by weaker prices. The evidence table has the sources.

RSI is a number from 0 to 100 that measures how fast and how much a stock's price has been moving up versus down recently (usually over the last 14 trading days). Think of it as a speedometer for buying and selling pressure, not for the price level itself - it says nothing about what a stock is worth.

When there's been a lot more buying than selling lately, RSI climbs toward 100. When there's been a lot more selling than buying, it falls toward 0. As a rule of thumb (not a law of physics): RSI below 30 is called "oversold" - the price has fallen quickly, and some traders watch for a possible bounce, though a genuine downtrend can also keep RSI low for a long stretch. RSI above 70 is called "overbought" - the price has risen quickly, and some traders watch for a possible pullback, though a strong uptrend can also keep RSI high for a long stretch.

What the record says. Looked good in one 1992 study of the Dow; did not hold up when re-tested on later years, and shows no consistent edge on individual shares after costs. Since 8 October 2026 it is a chart figure with no vote, and the backtester no longer counts it. The evidence table has every rule side by side.

Read the full page on RSI →

MACD compares two moving averages of the price - a faster one and a slower one - to gauge whether upward or downward momentum is building or fading. When the faster average crosses above the slower one, chart-watchers call it a "bullish crossover": the price has recently started climbing more strongly than its own longer-term trend. When the faster average crosses below the slower one, that's a "bearish crossover" - the opposite.

MACD is built entirely from past prices, so it tends to confirm a move slightly after it's already started rather than predict one in advance. It's most useful as one confirmation among several, not as a standalone early-warning system.

What the record says. Describes a move that has already started. No study finds it predicts the next month on its own after costs. Since 8 October 2026 it is a chart figure with no vote, and the backtester no longer counts it. The evidence table has every rule side by side.

Read the full page on MACD →

A moving average smooths out day-to-day noise by averaging the closing price over a set number of days - a "20-day average" is just the average of the last 20 closing prices, recalculated every day as new prices come in. Comparing a short average (e.g. 20-day) against a longer one (e.g. 50-day) shows whether the recent trend is running stronger or weaker than the bigger-picture trend.

When the short-term average crosses above the long-term one, it's often called a "golden cross" and read as momentum turning up. When it crosses below, that's a "death cross," read as momentum turning down. Both are confirmations of a move that has already been underway for a while - by definition, they can't catch the very start of a new trend.

What the record says. On single shares, a one-off count Trendentum ran in September 2026 found golden crosses followed by a higher price less often than an ordinary day (71 of 168). On the whole market, a slow average has historically cut the worst years — which is what the backdrop row in Market Conditions does. The evidence table has every rule side by side.

Read the full page on moving averages →

Volume is simply how many shares changed hands on a given day. On its own it doesn't say whether a stock went up or down, but comparing it to a recent average shows how much conviction was behind a price move. If a stock jumps on unusually heavy volume, many traders read that as more likely to stick, since a lot of market participants bought in to push the price. The same move on very light volume can mean the opposite - a handful of trades moving a thin, easily-swayed price.

This platform flags a "volume spike" whenever a day's volume is well above (by default, 1.5x or more) its own 20-day average, that day included. It is a chart figure with no vote: it shows the multiple only, not which way the price moved.

What the record says. Confirms a move was widely traded; not shown to predict what comes next on its own. The evidence table has every rule side by side.

Read the full page on volume spikes →

In the US, company insiders - executives, directors, and large shareholders - are legally required to publicly disclose, via a "Form 4" filed with the SEC, whenever they buy or sell shares of their own company on the open market. Trendentum pulls these filings directly from SEC EDGAR (a free, public government database) and tallies up insider buying versus selling over roughly the last six months.

Some investors read heavy insider buying as a vote of confidence, since insiders know their own business better than any outsider - and heavy insider selling the opposite way. But insiders also sell for entirely ordinary personal reasons unrelated to the company's prospects (buying a house, diversifying savings, tax planning, a pre-scheduled trading plan), so this is one data point to weigh, not a verdict. Outside the US, directors' trades come from the director notices carried by Twelve Data for Australia, Canada, the UK and Europe, and the stock page's Insider activity panel shows them; Japan, Hong Kong and Singapore aren't covered. Only counts and values are kept - names are never collected.

What the record says. Purchases carry information; sales mostly don’t. Since 8 October 2026 only purchases count, as part of Insiders and short sellers, and filings outside the US are read too. The evidence table has every rule side by side.

Portfolio Health is a single narrative summary of how your portfolio is positioned right now, built entirely from things this platform already computes elsewhere - it introduces no new opaque score. It has four sections:

• Concentration - each holding's share of your total position value (quantity × latest price, converted to USD only when a portfolio holds more than one currency - every other page on this site still shows each holding in its own native currency). If one position stands well above the rest, the headline says so.
• Sector exposure - the same position weights grouped by sector (from the company lookup this app already makes for each holding - Twelve Data first, with other providers as fallbacks - not a new data source). A holding with no sector on file, common for ETFs, is shown separately rather than guessed into one.
• Markets and currencies - the same weights grouped by the market each holding is listed on and by the currency it trades in.
• Factor exposure - your portfolio's strongest and weakest measures from the five-axis shape described in the next entry below, read back as one sentence. This section doesn't compute anything new; the full per-holding breakdown behind it is under “How your holdings compare” on your dashboard.

The headline at the top picks whichever of these three is most notable right now (a concentrated position first, then a dominant sector, then factor exposure) and states it in one sentence - it's composed from plain template text, never an AI-generated summary, so it can never say more than the numbers underneath it actually show. Needs at least two holdings, same reason as the five-axis shape: with only one position, there's nothing to compare it against.

Read the full page on concentration →

On your dashboard this is the top half of “How your holdings compare”. It used to be its own panel called the Portfolio Signal Scorecard.

This is the factor ranking table (see the next entry below) taken one level up: instead of ranking each holding, it averages those same rankings into one score per measure for your whole portfolio, then shows exactly which holdings pulled that average up or down. It doesn't introduce a new kind of number - every score here is built the same way the ranking table already builds its ranks, just aggregated.

Five axes, each a measure this platform already computes elsewhere:

• Mom 3M / Mom 6M - 3-month and 6-month price momentum, identical to the ranking table.
• Low Vol - how calm the price has been over 60 trading days, identical to the ranking table.
• Trend - built from the Market Conditions reading (ADX trend strength): a holding currently trending upward scores well here, one trending downward scores poorly, and a choppy or still-forming trend scores neutral. This describes how each position you already own is currently behaving, the same kind of price-behavior description every signal reason on this platform gives - it's not a suggestion about what to do with it.
• Insider - net dollar value of company insiders' own open-market buying versus selling (SEC Form 4 filings, US-listed stocks only). A holding with nothing on file - including every non-US stock code (ASX, TSX/TSXV, LSE, IDX, HKEX, SGX, TSE, NSE/BSE, XETRA, NZX - none of which file with the SEC) - is left out of this axis's ranking rather than being marked down for it.

Each axis's per-holding score is a 0-100 percentile rank within your portfolio only (100 = your best-ranked holding on that measure, 0 = your worst) - same scale the ranking table uses. The axis's overall score is the plain average of those ranks across your holdings. Click an axis on the dashboard and each holding's "lifting" or "holding back" percentage is simply its own rank minus that average - by construction, the lifting and holding-back numbers for any axis always add up to zero, so together they account for the entire average with nothing left unexplained.

With one holding, or too few holdings with usable data for a given axis, there's nothing to rank against - that axis (or that holding, for that axis) scores a neutral 50 rather than a real comparison, the same convention the ranking table uses.

On your dashboard this is the bottom half of “How your holdings compare”. It used to be its own panel called Factor Rankings.

Unlike the signals above, which look at one stock at a time, factor rankings compare every stock in your own portfolio against each other - the same basic approach professional quant funds use to decide which of the stocks they already like are the most attractive right now, relative to one another. It never compares you against the wider market, and it is not a price prediction: a top-ranked stock can still fall, and a bottom-ranked one can still rise.

That can look strange in practice. If three holdings returned −5%, −15% and −25% over the same window, the one down 5% ranks first - even though it lost money. The reverse also happens: a holding up 10% can rank last if the rest of the portfolio is up 20% and 30%. A high rank here means stronger relative readings within this specific group on the three measures below, not a statement about the stock's value, its safety, or the chance of a rise - and adding or removing a holding can shift every other holding's rank too, even when nothing about their own price behavior changed.

The composite score is built from exactly three ingredients, weighted equally - no single one counts for more than the others:

1. 3-month price momentum - the total return over the last ~63 trading days. Stocks that have been rising tend to keep rising for a while (a well-documented pattern), though momentum can reverse sharply and without warning.
2. 6-month price momentum - the same idea over a longer, ~126 trading day window, so a stock that's had a strong half-year counts even if its most recent quarter was quieter.
3. Low volatility - how calm or choppy the price has been over the last 60 trading days (annualized). A lower number is treated as better here - many investors treat a calmer ride as lower-risk for a similar return.

For each of those three, every holding gets ranked from best to worst only against the other stocks you hold and turned into a percentile score from 0% (worst of your holdings on that measure) to 100% (best of your holdings on that measure) - evenly spread out, so with 3 holdings the ranks are 100%, 50%, 0%. The composite score is simply the average of those three percentiles, then shown out of 100. In table form, for two holdings:

Stock A swept all three measures here, so it scores 100 and Stock B scores 0 - but with three or more holdings it's rarely this clean, since a stock can win on momentum while losing on volatility. Click any row in the Factor rankings table on the dashboard to see this same breakdown worked out for your actual holdings, with your actual numbers.

A holding without enough price history yet (a very recently added stock) scores a neutral 50% on whichever factor it's missing, rather than being penalized or dropped from the list - and if you only hold one stock, everything scores a neutral 50 across the board, since there's nothing to compare it against.

Take a share’s price today and compare it with a year ago, leaving out the most recent month (the last few weeks tend to bounce the other way). If it is well up, it has a strong year-long trend; well down, a weak one. That is the whole measure.

Since 1993 this has been checked more times than any other idea in investing, across dozens of countries and decades of prices: shares that rose over the past year tended, on average, to keep rising for some months. It is not a law. It fails hardest, and fastest, right when a falling market turns back up, when last year’s losers jump the most. So it is a measure with a strong record and a known weak spot, and the weak spot is the market backdrop, which Market Conditions shows.

In Trendentum. Each stock’s DNA and the rankings show 3- and 6-month momentum, the same idea over a shorter window. The twelve-month version is one of the outlook’s three readings since 8 October 2026: Trend (past year). The evidence table has the sources.

Today’s price as a fraction of the highest price in the past year. A share at 98% of its yearly high is “near”; one at 60% is far below it. The common belief is that a share near its high has run out of room. The research found the opposite: shares near their yearly high went on to do better, on average, than shares far below it, and the effect did not reverse later.

The usual explanation is a habit of mind. People anchor on the old high and are slow to pay more than it, so when good news arrives the price takes longer to reflect it, and the rise is spread over the following months. None of this says anything about a single share on a single day; it is a pattern across many shares and many years.

In Trendentum. Every stock page shows the 52-week low and high beside today’s price. The distance is shown, not voted.

Steadiness is how much a share’s price wobbles from day to day, scaled to a year. The site calls it volatility. A share that moves 1% on an ordinary day is calm; one that moves 4% is wild. Most people assume the wild one must pay better for the ride.

Over decades of data it did not. The wildest shares, as a group, did not earn more than the calm ones; often they earned less, and they fell far further on the way. Part of the reason is that wild shares are exciting and get bought for the excitement; part is that professional funds measured against an index find it awkward to hold calm shares in size. Either way, the calm ones have not been the poor relation.

In Trendentum. Each stock’s DNA shows its volatility and whether it is expanding or contracting; Portfolio Health ranks calmness among your own holdings; the rankings treat a calmer ride as the better one. Shown, not voted.

Correlation puts "do these two move together?" on a scale from +1 (lockstep) through 0 (unrelated) to -1 (opposite). It is what diversification actually rests on: holding more stocks only spreads risk if they do not all fall on the same days, and ten stocks that move together are one bet with ten names. The sector map's "Moves with your book" column is this figure, computed between each sector's index and your holdings combined at today's weights over the history they share; only sectors with an index get one. It is a description of the past year, not a promise about the next one - assets that moved independently in calm markets have a habit of falling together in a crisis.

Read the full page on correlation →

Professional and institutional investors don't all use one approach - most rules-based strategies fall into a handful of well-known archetypes, each with its own logic for when to get in, when to get out, how it manages risk, and the kind of market it tends to suit. None of these is "the" right one, and this platform doesn't pick one for you or run one on your behalf - understanding them is what lets you decide which philosophy the signals above map onto, and how much weight to give each one.

• Trend-following / momentum. Enters once price shows sustained directional movement (a moving-average crossover, a breakout, strong recent momentum); exits on a trailing stop-loss or once that momentum clearly weakens or reverses. The edge some point to: markets can trend for extended stretches as information spreads gradually and participants pile in over time, rather than everyone reacting instantly. Tends to have the most to work with in strongly trending markets; tends to struggle in choppy, range-bound ones, where it can generate repeated false starts ("whipsaws"). Because a trend can end abruptly, this archetype usually leans on an explicit stop-loss and accepts many small losses in exchange for occasional large wins.
• Mean-reversion. Enters once price has moved unusually far from its typical range (deeply "oversold" or "overbought" - see the RSI entry above), on the view that it tends to snap back; exits once price returns toward its average, or after a set holding period. The edge some point to: short-term price moves can overshoot on emotion or forced selling/buying, and have historically tended to partially correct. Tends to have the most to work with in range-bound, choppy conditions; tends to struggle badly in a genuine trend or breakout, where "it should bounce back" keeps being wrong and losses can compound - sometimes called "catching a falling knife." Usually paired with a tight stop-loss, since a real trend break defeats the whole premise.
• Value / fundamentals-based. Enters based on a company's price relative to some measure of its underlying worth (earnings, assets, cash flow) rather than its recent price action; exits typically much later, once the price approaches an estimated fair value or the underlying fundamentals change. The edge some point to: markets can misprice a company for a period before that gap closes. Tends to reward patience over long horizons and can do well in out-of-favor sectors; tends to struggle for stretches when momentum-driven markets keep pushing already highly priced stocks higher, sometimes for longer than expected.
• Multi-factor / quantitative. Combines several independent signals ("factors" - momentum, value, quality, low volatility, and others) into one blended score, on the logic that different factors tend to lead in different environments, so blending several reduces how much everything rides on any one of them being right at a given time. This platform's own holding-ranking feature (above) is a simplified version of this idea, using two of the most well-documented factors: momentum and low volatility. The trade-off is that a blended approach is rarely the single best performer in any one environment - the goal is steadier results across many environments, not the top result in any one of them, and factors can still move together (lose their diversification benefit) during periods of broad market stress.

One thing nearly every serious version of these approaches has in common: the exit and the risk controls are decided before entering the trade, not improvised afterward - and how much to put into any one position (see "Risk-reward ratio & position sizing" below) is treated as being just as important as picking the right entry signal.

The risk-reward ratio on a single trade compares the potential reward to the potential risk: the distance from your entry price to a profit target, divided by the distance from your entry price to a stop-loss. Buy at $100 with a stop at $95 (risking $5) and a target at $115 (a potential $15 gain), and that's a 3:1 risk-reward ratio.

This matters because it changes how often a strategy actually needs to win. A strategy with a 3:1 ratio only needs to be right a little over 25% of the time to break even, before costs - while a strategy with a 1:1 ratio needs to win more than half the time just to do the same. Neither ratio is "better" in isolation; what matters is whether the strategy's actual win rate, in practice, clears the bar its risk-reward ratio sets (see "Win rate vs. expectancy" below for how those two combine).

Position sizing is the separate question of how much of a portfolio to commit to any single trade, so that a losing streak - which every strategy has, sooner or later - doesn't do lasting damage. One widely-discussed approach is to risk only a small, fixed share of total capital (commonly-cited examples are in the 1-2% range) on any single position, sized off the distance to its stop-loss - so a trade with a wider stop automatically gets a smaller position, and one with a tighter stop gets a larger one, rather than every position getting the same dollar amount regardless of how far away its stop sits. That's one commonly-discussed approach among several, not a rule this platform enforces, and no sizing approach removes the risk of loss.

Note: this platform doesn't currently set price targets, stop-losses, or position sizes for you - these terms are explained here so the vocabulary is there if you're building your own trading plan around the signals above.

A drawdown is the decline from a portfolio's (or a single strategy's) peak value down to its subsequent lowest point, before it recovers - usually expressed as a percentage. A portfolio that falls from $100,000 to $70,000 has a 30% drawdown, whether or not it later climbs back. Maximum drawdown is simply the single worst peak-to-trough decline over some stretch of history, and it's one of the most commonly cited risk statistics for a strategy or fund, since it captures the pain an investor would actually have lived through - not just the average return over the same period.

The reason it's often weighed as heavily as (or more heavily than) average returns: recovering from a loss is mathematically asymmetric. A 20% drawdown needs roughly a 25% gain just to get back to even; a 50% drawdown needs a 100% gain. The deeper a drawdown goes, the harder it becomes to fully recover from - so two strategies with the same average annual return can look very different once you account for how rough the ride to get there actually was.

There's a behavioral side too: deep or prolonged drawdowns are commonly - and understandably - where investors abandon a strategy or a plan, sometimes right before or during the stretch that would have recovered it. Knowing roughly what kind of drawdown a strategy has historically produced, before committing to it, is part of being able to actually stick with it.

Read the full page on drawdown →

Backtesting means running a strategy's rules against historical price data to see how it would have performed, without risking real money - a standard step before committing capital to any rules-based approach. It's a useful exercise, but a strong-looking backtest is easy to produce by accident even when the underlying strategy has no real edge, mainly through three common traps:

Overfitting - tweaking a strategy's rules repeatedly until they look great on one specific stretch of historical data. The risk is that the rules end up fit to the noise and coincidences of that particular period rather than capturing something that reliably repeats, so a historically great-looking backtest can perform far worse - or lose money outright - once it meets new, unseen data.

Survivorship bias - testing a strategy only against companies that still exist today. A dataset built that way silently leaves out every company that got delisted, went bankrupt, or was acquired along the way, which inflates the apparent past performance of almost any strategy, since the biggest failures are invisible to the test.

Look-ahead bias - accidentally letting the backtest use information the strategy wouldn't actually have had on that date. For example, testing a rule like "buy after an earnings beat" using the final, fully-confirmed reported numbers, when in reality those numbers weren't finalized or public on the exact day being simulated.

None of this means backtesting is worthless - it means a backtest, however strong, is not proof of anything about the future. A strategy that performed a certain way historically, backtested or not, is no promise of performing that way again.

Read the full page on backtesting →

Win rate is simply the percentage of trades that were profitable. On its own, it's a misleading way to judge a strategy, because it says nothing about the size of the wins versus the size of the losses.

Expectancy fixes that by combining how often a strategy wins or loses with how large those wins and losses tend to be: (win rate × average win) minus (loss rate × average loss). Consider a strategy that wins 80% of its trades, but the average win is small ($50) and the average loss is large ($400): expectancy works out to (0.8 × $50) − (0.2 × $400) = $40 − $80 = −$40 per trade - a strategy that wins most of the time while losing money on average. Now consider the opposite: a strategy that wins only 35% of the time, but its average win ($300) is much larger than its average loss ($100): (0.35 × $300) − (0.65 × $100) = $105 − $65 = +$40 per trade - profitable despite losing most individual trades.

This is the same idea as the risk-reward ratio above, from a different angle: a favorable risk-reward ratio can make a strategy with a modest win rate net profitable, and a poor risk-reward ratio can make even a high-win-rate strategy a net loser over time. Expectancy is what folds both pieces into a single number.

A symbol's current regime describes the overall character of its recent price behavior, separate from which direction it's headed: is it trending strongly, or is it range-bound / choppy with no sustained direction? Is its volatility (how sharply the price is swinging day to day) currently elevated or unusually calm compared to its own normal? These are different questions from "will this go up or down" - a regime reading describes the current texture of price movement, it doesn't predict what happens next.

Trend and volatility can pull in different directions on the same stock. Two stocks can both go from $100 to $110 - one climbing in a fairly straight line, the other diving to $80 and spiking to $120 before landing at $110. Same start-to-finish return, very different ride along the way, so an upward trend and high volatility can perfectly well coexist. And low volatility here only ever describes a calm recent price - it says nothing about excessive debt, a high valuation, business problems, or a sudden unexpected event. "Calm lately" and "safe" aren't the same claim.

The "Market conditions" section on this dashboard reads trend using ADX (Average Directional Index), a widely-used technical indicator that measures how strong a directional trend currently is, independent of which way it's pointed (a companion pair of lines then says which direction). An ADX reading below roughly 20 is the standard textbook threshold for "no trend worth calling one" - read as range-bound or choppy; a reading at or above roughly 25 is the standard threshold for a trend with real strength behind it. Volatility is read by comparing a symbol's own recent (20-day) realized volatility against its own longer-term (60-day) average - never against the wider market or any other stock - so "elevated" means choppier than usual for that specific symbol, not necessarily dramatic in any absolute sense.

This connects back to the strategy archetypes described above: trend-following approaches have, historically, tended to have more to work with in trending conditions, and mean-reversion approaches have, historically, tended to have more to work with in range-bound ones. That's a general historical tendency, not a rule - conditions can shift from one regime to the other at any time, without warning.

As with everything else on this platform, this section only ever describes what the data currently shows - it doesn't tell you which strategy to use right now, or what to avoid. That call, like every other one here, is left to you.

Read the full page on market conditions →

The price has gone sideways in a narrow band for weeks, with no trend either way. Chart-readers call this a base or a range.

Our rule: ADX(14) below 20, and the last 15 or more closes inside a band no wider than 3x ATR(14) as a share of the price (between 4% and 12%).

What it does not tell you: Which way it will leave the range, or when. The stock page counts what followed ranges like this one breaking, with and without heavy volume.

The price has just closed above the top of a sideways range it had been in. Chart-readers call it a breakout, and look at how much trading there was that day.

Our rule: Within the last 5 days, a close above the top of a range that was Consolidating the day before. Volume at or above 1.5x its 20-day average is noted and counted separately.

What it does not tell you: Whether the move holds. Plenty of breaks fall back into the range; the record on the stock page counts both kinds.

The price has just closed below the bottom of a sideways range it had been in.

Our rule: Within the last 5 days, a close below the bottom of a range that was Consolidating the day before.

What it does not tell you: Whether it keeps falling. The record on the stock page counts what followed breaks like it before.

A stock in an uptrend has dipped back to its 50-day average after running well above it.

Our rule: ADX(14) 25 or more, the 50-day average higher than 20 trading days ago, the close not below the 200-day average, and the close within 2% of the 50-day average having been 8% or more above it in the last 30 trading days.

What it does not tell you: Whether the uptrend carries on. On the day, a pause in an uptrend and the start of a downtrend can look the same; the record shows how often each followed before.

The price is unusually far above its 200-day average - further than it has been on nine days in ten in its own history.

Our rule: The close above the 200-day average by more than this stock's own 90th percentile of that gap over every earlier day with a 200-day average (at least 120 of them).

What it does not tell you: That it will fall. Strong stocks can stay extended for months. The record shows how often a stock this stretched was higher or lower a month and three months later.

The price has been below its 200-day average for two months or more, in a downtrend, with each month's high lower than the one before.

Our rule: ADX(14) 25 or more with -DI above +DI, the close below the 200-day average for 40 closes in a row or more, and each of the last 3 blocks of 21 trading days topping out lower than the one before.

What it does not tell you: That it won't recover. It describes the slide so far; the record shows what followed slides like it.

Far more trading than usual, or a far bigger move than usual, on a day with no announcement this app reads.

Our rule: Volume 3x its 20-day average or more, or a close-to-close move of 2.5 standard deviations or more of the last 60 days' moves.

What it does not tell you: Why it happened. Heavy volume says many people traded, not who or why. It is not evidence of manipulation - the company's own announcements are the place to look.

None of the seven rules applies today; the chart is between states.

Our rule: None of the seven rules above fires on the latest close.

What it does not tell you: That nothing is happening. The trend, the averages and the outlook elsewhere on the page still describe it.

This entry replaced a green “live data” badge that used to sit in the header of every page. The badge was a claim rather than information, it was wrong in three separate ways, and because it was always green it said nothing on the one day it would have mattered.

These are not live prices. Everything on this platform is built from daily bars - one open, high, low, close and volume figure per trading day, per stock. The readings use finished trading days only: while a market is open, today's unfinished bar is left out until that market has closed and the figures have settled. So a reading changes at most once per trading day in each market.

That is the right shape of data for what this platform does. Every measure here reads a stretch of history rather than this second's price. The outlook's three readings look at the past year, the 60 trading days after results, and the last 90 days of filings. The chart figures, which do not vote, look at shorter windows: RSI at 14 days, MACD at 12 and 26, the moving averages at 20 and 50. None of them would give a different answer if they were fed a price one minute fresher.

When the figures update:

• Whenever you open a page, if what's stored is more than 15 minutes old. This is the one that matters most - if you are looking at the dashboard, you are looking at something fetched within the last few minutes.
• On a schedule in the background, so the record stays continuous even on days nobody opens the app. That is what makes the history behind each stock worth reading.
• Never automatically while a page sits open. Leave the dashboard open for hours and the figures stay as they were when it loaded - the “Updated N ago” note at the top of Stock Outlooks is the honest reading, so trust it over the numbers looking recent.

Prices can be up to an hour old. A price fetched for one stock is reused for up to an hour before it is fetched again, so that opening several pages doesn't hammer the data provider. For daily-bar analysis that is immaterial; if you are about to actually trade, check the live price with your broker, not here.

Where it comes from: daily prices from Twelve Data first, then EODHD, with Alpaca and Yahoo Finance as fallbacks; SEC EDGAR for US directors' filings; ASIC for ASX short positions; and Twelve Data for results dates and director notices outside the US. These are not the real-time exchange feeds a professional desk pays for, and they can be briefly unavailable. When a price lookup fails, this platform now tells you it failed rather than showing you a number - see the next paragraph.

If anything is ever not real, you will see a red banner at the top of the page naming the affected stocks. In normal operation there is no badge and no banner at all: silence means the figures are genuine, and the only status worth interrupting you for is the bad one.

A stock code is the short code a market gives a listed company: BHP, AAPL, VOD. You will also see it called a ticker or ticker symbol - it is the same thing, and every broker uses one of the three names.

Trendentum follows the same exchange-suffix convention Yahoo Finance uses to tell markets apart: a plain symbol with no suffix (e.g. AAPL) is assumed to be US-listed. Using the right suffix matters because it's how the platform knows which exchange - and which data source and regulator - to look the stock up under. Or just start typing in the stock code field and pick from the suggestions; the right suffix comes along automatically.

• .AX - Australian Securities Exchange (ASX), e.g. BHP.AX Coming soon - the exchange’s data licence is being arranged
• .TO - Toronto Stock Exchange (TSX), e.g. RY.TO
• .V - TSX Venture Exchange (TSXV), e.g. GOLD.V
• .L - London Stock Exchange (LSE), UK, e.g. VOD.L. Quoted in pence, not pounds - a price shown as 131.20p is £1.3120. Enter your average cost in pence too, or your gain figures will be out by a factor of 100.
• .JK - Indonesia Stock Exchange (IDX), e.g. BBCA.JK
• .HK - Hong Kong Stock Exchange (HKEX), e.g. 0700.HK
• .SI - Singapore Exchange (SGX), e.g. D05.SI
• .T - Tokyo Stock Exchange (TSE), Japan, e.g. 7203.T
• .NS - National Stock Exchange of India (NSE), e.g. RELIANCE.NS
• .BO - Bombay Stock Exchange (BSE), India, e.g. RELIANCE.BO
• .DE - Deutsche Börse (XETRA), Germany, e.g. SAP.DE
• .PA - Euronext Paris, France, e.g. MC.PA · .AS - Euronext Amsterdam, e.g. ASML.AS · .MI - Borsa Italiana, e.g. ENEL.MI · .MC - Bolsa de Madrid, e.g. SAN.MC · .BR - Euronext Brussels, e.g. ABI.BR · .HE - Nasdaq Helsinki, e.g. NDA-FI.HE (all quoted in euros)
• .NZ - New Zealand Exchange (NZX), e.g. AIR.NZ

One thing that doesn't change with the market: SEC Form 4 insider-trading data and the AI deep-dive's SEC-filed financials are US-only regardless of suffix, since only US-listed companies file with the SEC. Director notices for Australia, Canada, the UK and Europe come from Twelve Data instead; Japan, Hong Kong and Singapore have none. Every market above still gets the price-based readings, holding rankings, and Market Conditions.

Three questions worth asking as you read a deep dive, rather than just trusting the summary:

1. What's the reported fact, versus the interpretation? Separate a published number - a change in revenue, say - from what the summary suggests it might mean.
2. What actually supports the explanation? The source links exist so you can check the date, reporting period and context yourself; they don't prove every sentence of an AI summary is correct.
3. What's left unanswered? A summary can leave things out - ask what other evidence would strengthen or weaken its read.

One phrase worth reading precisely: "nothing material has changed since your last look" describes the inputs Trendentum compares, not the company itself - it doesn't mean nothing happened, and it doesn't mean every possible development was checked.

The "Deep dive" button on a signal card asks an AI model (Claude, from Anthropic) to read a set of factual inputs about that company and write a plain-language summary - it does not go looking for information on its own, and it's told explicitly to work only from what it's given, not from anything it might already "know" about the company.

What it's given: up to three years of annual revenue, net income, cash, and debt as reported directly to the SEC (for US-listed stocks only - see the section on market suffixes below for why other exchanges aren't covered here yet); a summary of recent SEC-reported insider buying/selling; recent news headlines, when a news source is configured for this deployment; and the same current trend/volatility reading shown in Market Conditions. Nothing is invented - if a section of that input says data isn't available, the summary is expected to say so plainly rather than guess.

Same rule as every other reasoning text on this platform: it's told to describe, never to recommend. It's instructed never to say "buy," "sell," "hold," or "invest" as a directive, and never "now is a good time to..." or "you should...". A best-effort automatic check also scans every generated summary for that kind of language - if anything slips through anyway, you'll see a visible caveat above the summary rather than it being silently trusted. AI-generated text can still occasionally drift in ways a fixed set of rules can't fully catch, which is exactly why that caveat exists - treat a flagged summary with extra skepticism, and treat every summary, flagged or not, as a starting point for your own reading, not a conclusion.

The "since your last look" note at the top compares today's facts against whatever they looked like the last time you personally opened that symbol's deep dive - it's about what's factually changed (a new revenue figure, new insider activity, new headlines), never a judgment about whether the change is good or bad.