Every S&P 500 Stock, Scored Daily by the Numbers. The 20-stock multi-factor portfolio: +5.0 points a year over the S&P 500 in backtest, 1998-2026
Both sides move together, always by the same rule: the cap-weighted portfolio against the cap-weighted S&P 500, or the equal-weight portfolio against the equal-weight one. Never one against the other.
Momentum, value, quality and growth rankings for every company, recalculated three times every trading day; model portfolios that rebalance themselves every month; company profiles built from SEC filings; insider buying; and a simulator to test your own strategy. All ranked by numbers instead of opinions, free, with no sign-up and no ads, and every formula published so you can check the numbers instead of taking our word for them.
Today's coverage · 466 of the 503 index companies have all four factors
Walk-forward of 332 monthly rebalances since December 1998, recomputed from the carteras.json published on September 18, 2026. Total return: dividends are reinvested on both sides. Alpha of +4.82% with a beta of 1.03 (Jensen's, against that same benchmark with the average risk-free rate of the period). The portfolio's worst fall in the simulation was 59.9%. Gross: no curve on this site deducts commissions, spread or taxes; what they cost is measured further down. These are simulated results, not real trades. Neither past nor simulated returns guarantee future ones, and investing in stocks can mean losing part or all of the capital.
This month's portfolio
20 stocks · bought on August 21, 2026 · 6 changes from last monthEvery line starts at zero on the first day of the track record and compounds the total return —dividends reinvested— of buying the 100 highest-scoring names on that factor every month. The dashed gray line is SPY, the S&P 500 ETF: a real fund anyone can buy, not an average we compute ourselves. Click a name in the legend to switch that curve off.
Mind the size. These are the Top 100 portfolios, the widest version and the one that strays least from the index; the legend figures are 27.6-year cumulative returns, not annual ones. The portfolio above is the Top 20, far more concentrated: it returns more (13.7% a year) and falls further (−59.9% at its worst). The trade-off runs both ways, and that is the decision to make.
monthly rebalances, to . Walk-forward: each date scored only with prices up to that day and fundamentals already filed with the SEC. Dashed: SPY, the S&P 500 ETF (a real fund, dividends reinvested). Open the portfolios → · Download the data (Excel)
Open the rankings and you get the entire index sorted by score: what the engine rates highest today, what it rates lowest, and the four factor numbers behind every single name. Nothing to install, no email to hand over, no trial that expires. It is the same kind of factor screen the institutional smart-beta indices are built on — published in the open, three times a day.
None of it is an opinion. The rules are fixed before the market opens and are never overridden intraday — no analyst call, no last-minute adjustment, no favorite stock. They are written out in full below: the formula behind each of the four factors, where every input comes from, and how the backtest is run. And we don't hold your money, we don't route orders and we're not paid when you trade, so the numbers have no reason to flatter anyone. No black box.
The same rules every day
The ranking comes out of the formulas written out further down this page, applied to every company in the index in the same order. No name is moved up or down by hand.
Four factors, one score
Every company is scored on momentum, value, quality and growth, and the final score is the average of the four. Momentum comes from prices; the other three from the accounts filed with the SEC.
Recalculated three times a day
The data is recalculated three times every trading day — at the market open, at the close and overnight — so the scores reflect the latest prices and the latest accounts filed.
The result is a unified Z-Score for each stock - a statistically normalized number that tells you, at a glance, exactly how each company compares to the entire S&P 500 universe. Averaging the four factors cancels the extremes against each other, so the scale is narrower than a textbook bell curve would suggest: today the multifactor column runs from −1.11 to +1.81. A score of +1.04 is already enough to sit in the top 1% of the index, and one of −0.73 puts it in the bottom 1%. The individual factors do reach ±3, because that is where they are capped: these scores are averages of capped Z-Scores and that squashes the bottom tail. It is one number on one scale, which is what lets two companies be compared directly — and it is a starting point for your own reading, not the end of it.
Whether you're an experienced portfolio manager looking for a systematic second opinion, an independent investor who wants to see the numbers behind each stock, or a finance student who wants to see how institutional quantitative strategies actually work - the rankings, the portfolios and the simulator are free and open, with no registration required.
The Four Factors Behind Every Score
A factor is a measurable company characteristic that has historically been associated with differences in return - not a stock tip, but a shared trait. Value, for example, is not "this company is good", it is "this company is cheap relative to what it earns". Academic finance spent forty years testing hundreds of candidate traits; only a handful survived out-of-sample testing in different decades and different markets. Our engine uses four of those survivors, and it scores the whole index, though not every company has all four: today there is momentum for 497 companies out of the 503 in the index, growth for 494, value for 494 and quality for 473, and 466 have all four, every day, whether or not the result is flattering.
Why four instead of the one that works best? Because a factor is not a machine that prints money - it is a premium that appears in some periods and vanishes in others, sometimes for years. Momentum gets destroyed at sharp market turns. Value spent most of the 2010s underperforming. Quality lags badly in a speculative rally. What makes the combination useful is precisely that these droughts do not coincide: each factor tends to work when at least one of the others is struggling. Below you will find, for each one, the economic reason it works, the exact formula we use, and - the part almost nobody publishes - the conditions under which it fails.
1. Momentum
Why It Works
The Momentum factor identifies stocks with strong, persistent upward trends. Jegadeesh and Titman documented it in 1993 and it has since been confirmed across almost every market and asset class studied: on average, the winners of the last twelve months keep beating the index over the following months.
The uncomfortable part is that nobody fully agrees on why. The most accepted explanation is behavioral: markets under-react to good news. When a company reports results well above expectations, the price does not jump instantly to its new fair value - it drifts there over weeks, because investors anchor to their previous opinion and revise it slowly. On top of that sits the disposition effect (we sell winners too early to lock in a gain, which holds the price back) and the mechanics of institutional flows: a fund that decides to build a position of hundreds of millions cannot buy it in one session, so it buys in slices for weeks, sustaining the trend.
When it fails: at sharp reversals. Momentum's worst moments are not slow bear markets - they are violent rebounds off the bottom, when the most beaten-down stocks bounce hardest and the portfolio is holding precisely the opposite. In 2009 and in the spring of 2020 the factor gave back years of accumulated advantage in a matter of weeks. It is known in the literature as a momentum crash, and no filter eliminates it.
How We Calculate It
The engine ranks companies by their risk-adjusted momentum, with this formula —which is neither ours nor secret, it is the one in the literature:
where:
• Return12m-1m = 12-month cumulative return, excluding the most recent month.
• σ = Annualized daily volatility over the same 12-month window.
Every piece of that formula is there for a reason. Excluding the most recent month is not a detail: at horizons under four weeks the effect inverts and prices tend to revert, so including it would systematically buy stocks about to give back part of their run. Dividing by volatility converts a raw return into a return per unit of risk, which distinguishes a company that has climbed 40% in a steady line from one that has climbed the same 40% through violent swings on rumours - the first trend tends to persist, the second does not.
The volatility is measured over the same 12 months as the return, so numerator and denominator describe the same period. A much shorter window would react to any temporary spike in nervousness and would keep reshuffling the ranking on noise; the aim is to measure the stock's structural character, not last week's mood. This is the exact formula the backtest runs - the ranking you see and the track record we publish are the same calculation, not two similar ones.
2. Value
Why It Works
The Value factor buys what is cheap relative to what the company actually produces - its earnings, its book value, its sales. It is the oldest documented effect in the literature: Fama and French built it into their 1992 model after finding that, over long periods, the cheapest quintile of the market beat the most expensive one.
There are two competing explanations and both are probably partly right. The behavioral one says the market overreacts: bad news gets extrapolated into the indefinite future, the price falls further than the deterioration in the business warrants, and when reality turns out to be merely mediocre rather than catastrophic, the price recovers. The risk-based one is less comfortable: cheap companies are cheap because they genuinely are more fragile - more debt, more cyclicality, more chance of not surviving a recession - and the extra return is simply the payment for carrying that risk. If the second explanation dominates, value is not a free lunch; it is a wage for discomfort.
When it fails: when cheap is cheap for a reason. That is the value trap - a business in structural decline whose multiple keeps falling because its earnings keep falling. And in aggregate, value can be wrong for a very long time: it underperformed growth for most of the 2010s. A decade is longer than most investors' patience, which is exactly why a rules-based system does not capitulate at the worst moment, which is where the same investor acting on their own most often goes wrong. That does not guarantee the strategy wins.
How We Calculate It
Our scoring engine avoids sector valuation biases by averaging three distinct fundamental yields:
where:
• EY (Earnings Yield) = trailing-twelve-month EPS / price.
• BY (Book Yield) = 1 / Price-to-Book (P/B).
• SY (Sales Yield) = 1 / Price-to-Sales (P/S).
We work with yields (the inverse of the multiple) rather than the multiples themselves for a practical reason: a P/E can go negative or explode toward infinity when earnings approach zero, which wrecks any average. Its inverse behaves properly across the whole range, so a company with collapsing earnings simply scores badly instead of contaminating the entire ranking.
Averaging three different yields is the defense against the classic trap. A bank looks cheap on book value, a retailer on sales, a cyclical on earnings at the top of its cycle: any single metric favors certain business models and certain sectors. Requiring cheapness on the average of the three demands that the discount show up from more than one angle - which does not eliminate value traps, but does filter out the ones that are only an accounting artefact.
3. Quality
Why It Works
The Quality factor looks for companies that are simply better businesses: they earn a lot on the capital they employ, they keep fat margins, and they do it without loading the balance sheet with debt. Novy-Marx showed in 2013 that gross profitability predicts returns with roughly the same power as classic value measures - a result that surprised the profession, because it says the market persistently underpays for boring, profitable, unexciting companies.
The economic reason is durability. A company earning 30% on its capital year after year is telling you something a single quarter cannot: that it has a barrier keeping competitors from copying it - a brand, a network, a patent, a distribution scale. Markets tend to assume that kind of advantage will erode faster than it actually does, and the companies that hold it end up beating the expectation embedded in their price. The low leverage requirement is what makes the factor defensive: a company with little debt does not depend on refinancing at the exact moment credit dries up, which is what turns a bad year into a terminal one.
When it fails: in speculative rallies. When the market is paying up for stories and promises, solid profitable businesses look boring and lag the index for quarters at a time. There is also a structural limit: quality is rarely on sale - everyone can see a good business, and it usually trades at a premium - so buying quality without looking at price means overpaying. That is exactly why this factor is one input of four and not the whole system.
How We Calculate It
Our engine evaluates five key financial pillars to assign a robust quality score:
where:
• ROE = Return on Equity | ROA = Return on Assets
• GM = Gross Margin | OM = Operating Margin
• LEV = Financial Leverage (Total Debt / Equity, inverted)
The five pillars answer complementary questions. ROE measures the return delivered to the shareholder, but it can be inflated purely by taking on debt - which is why it never appears alone. ROA is the honest counterweight: return on every asset employed, regardless of who financed it. Gross margin reveals pricing power (whether the company can raise prices without losing customers), and operating margin shows whether that power survives the cost of actually running the business. Leverage enters inverted because here less is better: a debt-heavy balance sheet is a bet that nothing will go wrong.
Requiring at least three valid metrics prevents a company with an incomplete or unusual filing from scoring on a single lucky number. Note the limitation: these figures come from what the company reports, and accounting has legitimate margins of discretion. The factor detects statistical mediocrity, not fraud.
4. Growth
Why It Works
The Growth factor looks for companies whose revenues, earnings and cash flow are expanding fast. Here we owe you a distinction that most platforms blur: unlike the previous three, "growth" is not a factor with a well-established premium in the academic literature. In fact the classic finding points the other way - on average, the fastest-growing companies have delivered lower subsequent returns than cheap ones, because the market pays in advance for growth it expects and often overpays.
So why include it? Because growth measured on its own is a bad signal, but growth measured against its price is a very useful one. Two companies growing at 25% a year are not equivalent if one trades at 15× earnings and the other at 90×: the first only needs to keep doing what it does, the second needs to keep doing it flawlessly for a decade. The classic way to capture that is the PEG ratio, and we deliberately do not use it: PEG needs analysts' forward estimates, and there is no way to reconstruct what those estimates said on a given day years ago, so no backtest could ever support it. We would rather run three components we can prove than four we cannot. Growth also acts as a corrective on Value: it is what stops the model from filling the portfolio with companies that are cheap because they are shrinking.
When it fails: when interest rates rise. Growth companies are worth what they will earn in the distant future, and a higher discount rate hits distant profits hardest - 2022 was the demonstration, with heavy losses in exactly that segment. It also fails when growth is accounting rather than real, which is why we require it to show up in operating cash flow too: revenue can be booked with aggressive credit terms, but cash either comes into the bank or it does not.
How We Calculate It
The growth composite combines three rates, and all three can be rebuilt from the accounts filed with the SEC:
where:
• Rev_Growth = Year-over-Year (YoY) revenue growth rate.
• EPS_Growth = Earnings Per Share (EPS) growth (YoY).
• CF_Growth = Operating cash flow growth rate (YoY).
The three growth rates are measured year-over-year rather than against the previous quarter, which neutralizes seasonality: comparing a retailer's Christmas quarter with its spring quarter says nothing about the business. And they are three and not one because the gap between them is itself information - revenue growing while earnings shrink means margin compression; earnings growing while cash flow does not usually means the growth lives in accounting entries rather than in the bank account.
The figures behind these rates are trailing-twelve-month (TTM) and are rebuilt every quarter from the 10-Q filings, so the factor notices a stalled expansion within weeks of the report rather than waiting for the annual accounts. The only lag left is the real one: a quarter takes about six weeks to reach the SEC, and until it does the model works with the previous one - exactly like any live investor.
5. Multi-Factor (Global Score)
The Multi-Factor score integrates all four dimensions into a single, unified rating:
Because the individual factors run on cycles that do not line up. Value tends to work when Momentum stalls; Quality defends when Growth contracts. Averaging them does not produce the best factor of each year - it produces something more useful for anyone who has to live with the portfolio: a curve with fewer stretches bad enough to make you abandon the strategy. The point of combining is not to maximize the return of the good year, it is to survive the bad one still invested.
You will notice the formula is a plain average: each factor weighs 25%, always, with no optimization. That is deliberate, and it is one of the most important decisions in the whole system. We could search for the weighting that maximizes historical return - and we would find it, guaranteed, because with four parameters and decades of data you can always find the combination that would have won. That number would be spectacular and worthless: it would be describing the past, not predicting anything. An equal weight has no fitted parameters, so there is nothing in it that can be overfitted to a sample. Sometimes the honest option is also the one that survives longest.
Understanding Z-Scores
Every score on the platform is expressed as a Z-Score - a statistical measure of how many standard deviations a company sits above or below the S&P 500 mean. It exists to solve a problem that sounds trivial and is not: the four factors are measured in units that cannot be compared. Momentum comes out as a ratio of return to volatility, Value as an earnings yield in percent, Quality as a mix of margins, Growth as annual growth rates. Adding them raw would be adding kilometres to kilograms - and worse, whichever metric happened to have the widest numeric range would silently dominate the total.
Normalizing fixes that. Each metric is converted into "how far is this company from the average of the index, measured in standard deviations", which strips out the original units and leaves everything on one comparable scale. From that point on, adding them means something. Quick reference:
| Z-Score | What It Means | Measured Percentile |
|---|---|---|
| +2.00 | Exceptional - top of the ranking | 99% (95-100) |
| +1.50 | Well above average | 97% (91-99) |
| +1.00 | Above average | 93% (84-97) |
| 0.00 | Slightly above the median company | 57% (53-67) |
| −1.00 | Below average | 4% (0-15) |
| −1.50 | Well below average | 1% (0-3) |
These are measured percentiles, not textbook ones. The normal-distribution figures - 93% for +1.50, 50% for 0.00, 7% for −1.50 - are wrong for these scores: a factor score is the average of several capped Z-Scores, and that averaging squashes the bottom tail. The numbers above are what we actually measure across the four factors on five dates spanning 2022 to 2026, with the observed range in brackets. The practical consequence is the row that matters most: a score of 0.00 is not the middle of the pack. It sits above roughly 57% of the index, so "zero" already means slightly better than the median company, not average.
Winsorization: Z-Scores are capped at the [−3, +3] range so that a single extreme case cannot distort the ranking. Without that cap, one company with an anomalous figure - an accounting one-off, a metric near zero in the denominator - would stretch the standard deviation of the whole index and push everyone else toward the middle, compressing five hundred real differences into nothing. The treatment is standard in the factor index industry, and it has a cost worth stating: a genuine outlier is capped along with the artefacts. A company that really is four deviations above the rest appears as a +3, indistinguishable from one at +3.2.
What happens to a broken figure. Capping is the right treatment for a real outlier, and the wrong one for a filing error - a capped error does not get discarded, it gets promoted to best-in-index. One S&P company files a quarter whose XBRL "revenue" is $216,000 instead of $2.9 billion, which works out to a gross margin of 5,070; capped at 1.0 it would have looked like one of the most profitable businesses in America. So each metric now has two thresholds: outside what a real company can physically be, the metric is thrown away; inside it, extreme-but-real values are capped as described above. We also check that earnings per share and the share count refer to the same thing (EPS × shares should be net income) and drop all three per-share metrics when they do not, which is what a share-class mix-up or a units error looks like. The cost is coverage, and we would rather show it than hide it: . A company missing from a factor is one whose filings we could not read cleanly, not one we judged and found wanting.
Two things a Z-Score is not. It is not an absolute verdict: the score is always relative to the index today. A Momentum of +2.5 in a market where everything is falling means "falling less than almost everyone", not "rising". Read alongside the benchmark, never on its own. And it is not a probability of going up: it says the company scores in the top percentiles of a characteristic that has historically been associated with better returns - across hundreds of stocks and years, not for that specific company next month. The factor speaks in averages over a portfolio; it says nothing about the fate of any individual position.
How It Works
Data Ingestion
The engine collects prices, fundamentals and corporate actions for all ~500 companies in the index, three times every trading day (market open, market close and midnight). Prices arrive adjusted for splits, because otherwise a 10-to-1 split would look like a 90% crash. They are not adjusted for dividends: each payment is credited separately on its ex-date, which is the only way to avoid counting it twice. The fundamentals come from filings published to the SEC, and each figure is stored with the date it was actually filed - the detail that makes an honest backtest possible, and the same rule Sharadar follows, which supplies the history up to August 21, 2026.
Factor Scoring
Each company is scored on the four factors. The raw metrics are winsorized to stop an anomalous case from dragging the whole distribution, then normalized into Z-Scores so the four can be compared and averaged. The same rule is applied to all ~500 companies, with no exceptions and no manual overrides for any name.
Rankings and Portfolios
The results are sorted into full rankings and assembled into model portfolios, each with its weights, its equity curve and its risk metrics - Sharpe, Sortino and, above all, maximum drawdown. From there you can replicate a portfolio at your own broker or build your own rules in the simulator.
Why Quantitative?
Not because a machine is smarter than an analyst - it isn't - but because of something more mundane: it is consistent. Most of the gap between what a strategy returns on paper and what its followers actually earn does not come from picking the wrong stocks. It comes from abandoning the plan. Buying after the rally, selling in the panic, keeping the losing position "until it comes back", turning a 5% position into 30% because that one feels right. A model does none of that, not out of discipline, but because it structurally can't.
It also solves a problem of sheer scale. Following 500 companies means reviewing 500 sets of accounts, 500 price histories and 500 valuations - and doing it again next month. No human does that without cutting corners: in practice we all end up watching fifteen familiar names and calling the rest "the market". The model has no favorites and no fatigue: the five hundredth company gets exactly the same treatment as the first.
| Discretionary Investor | Quant500 Factor Model | |
|---|---|---|
| Discipline in a drawdown | Sells in the panic, buys back after the rally | Same rule however deep the fall |
| Coverage of the index | Follows ten or twenty familiar names | Scores all ~500 companies, three times a day |
| The rules themselves | Criteria drift with the news cycle | Four factors, 25% each, never re-tuned |
| Source of the data | Headlines, tips and last quarter's story | SEC filings, kept with the date they were filed |
| Position sizing | The favorite quietly grows into 30% | Weights set by the chosen method, not by conviction |
| Repeatability | Hard to reconstruct a decision months later | Every score recomputable from the published formula |
| Context and judgement | Sees the acquisition, the regulation, the fraud | Takes the filed accounts at face value |
That last row is not a courtesy. There are things a human still does far better. The model reads numbers, not context: it does not know that a company is being acquired, that a regulation is about to wipe out its main line of business, or that its growth comes from a business model that will not exist in five years. It cannot detect fraud - it takes the filed accounts at face value - and it has never seen a market regime that isn't in its data. What a factor model does well is narrow five hundred companies down to a short list on objective criteria, without getting tired or falling in love. Deciding what to do with that list is still your job, and that is why we explain the reasoning instead of just handing over the ranking.
What the Platform Includes
Everything below is open and free, with no account required. There is no premium tier holding back the good numbers: the ranking you see is the ranking, complete, from the top of the index to the bottom.
Complete S&P 500 Rankings
The full ordering of every company on each of the four factors and on the combined score, updated daily. You can sort by any factor to see who leads on Momentum but sits at the bottom on Value - the disagreements between factors are usually more informative than the overall ranking. The bottom of the list is published too: knowing what the model rates worst is as useful as knowing what it rates best.
Model Portfolios
Ready-made portfolios (Top 10, 20, 50, 100) with each position's weight, the historical equity curve and the risk metrics: Sharpe, Sortino and maximum drawdown. The number of positions is not a detail - a Top 10 concentrates the factor signal and swings much harder, while a Top 100 barely moves away from the index. Compare the drawdowns before choosing.
Custom Simulator
Build your own rule: pick the factor, the number of positions and the weighting method (equal, by score, risk parity or market cap), and see the resulting equity curve against the benchmark. Worth using in the opposite direction to the obvious one - not to hunt for the combination with the highest return, but to see how much the curve changes when you alter a single choice.
Deep Company Analysis
The full profile of any company in the index: valuation multiples, financial statements, institutional ownership and its options data (expected move, max pain, put/call ratio and implied volatility, as a sentiment gauge and never part of any portfolio's score). This is where you go to answer the question the ranking cannot - why a company scores what it scores, and whether the reason is something you are willing to own.
Insider buying and selling
The Form 4 filings that officers and large shareholders send to the SEC, read the same day they are published and linked to the original document. Both sides are published: sales are constant and usually just stock compensation, which is why the ones worth watching separately are the purchases, because they cost the buyer money. It is a data feed, not a model signal: it does not enter any factor score.
My Portfolio
You enter what you actually hold and the site measures it with the same metrics it uses on its own portfolios: total return, risk, and the comparison against RSP, the equal-weight S&P 500 ETF. Its real use is the uncomfortable one: seeing whether what you hold beats buying the index and sitting still.
Backed by Decades of Financial Research
We did not invent any of this. The four factors we use come from papers published between 1992 and 2013, peer-reviewed, replicated by researchers with no stake in the result and tested on decades of data across different markets. Eugene Fama, who co-authored the foundational work, won the Nobel Prize in Economics in 2013. The same models underpin the smart-beta indices that manage hundreds of billions in institutional assets.
That is deliberate on our part, and it is worth being explicit about why. A proprietary model that nobody can inspect asks you for one thing: trust. A published model asks for none - you can go and read the original paper, check whether it has replicated out-of-sample, and find the critiques written by people who think the premium has since been arbitraged away. Our contribution is not a secret formula, it is the engineering: keeping the data clean, applying the same rules to ~500 companies every day, and doing the backtest without cheating. Anyone can verify what we implement, because we did not invent it.
How We Keep the Backtests Honest
Anyone can print a bigger number than ours; almost nobody shows you what went wrong. The easiest way to make a strategy look brilliant is to pick the winners using today's data and then measure them against past prices. It produces spectacular numbers and exactly zero predictive value, and it is the single most common flaw behind the returns advertised in retail quant marketing.
Here it is done the other way round, and that has a price we pay in public. Between August 31 and September 1, 2026 we measured five candidate signals for improving the engine — low volatility, post-earnings announcement drift (PEAD), FINRA short interest, net share issuance, and eleven alternative definitions of the four factors — and all five were left out. One of them lifted the curve by +4.35 points of CAGR and was rejected anyway, because that signal on its own predicts in the wrong direction: with no mechanism, a better curve is a coincidence that happens to look good. And on September 1, 2026, once 49 price series of companies that no longer trade were folded in, every curve on this site went down: the multi-factor 20 fell from 17.29% to 16.82% a year, and the new figure was published the same day. Publishing what we reject and what we revise downward is not modesty — it is the only reason you can believe what we do publish. Here is how our engine avoids fooling itself — and, just as importantly, what it still cannot fix.
What the engine does
- Walk-forward, always. To build the portfolio for a date in the past, the engine only sees data that existed on that date. It then moves forward a month and repeats. Nothing from the future ever enters the calculation.
- Point-in-time universe. At each past date, only the companies that were actually in the S&P 500 that day can enter the portfolio: up to August 21, 2026, according to Sharadar's historical index membership, back to 1998 and including the companies that later disappeared; since then, according to the Wikipedia list (CC BY-SA), cross-checked against a second source. Without it you buy, in 2015, companies the index did not admit until 2022 - picked precisely because they did well afterwards. Measured on our flagship portfolio, that shortcut handed out 13.33 points of annual return for free (measured on Sep 1, 2026 over a 4.94-year window —59 rebalances, August 2021 to July 2026—: 25.93% on the real universe against 39.26% on today’s).
- Point-in-time quarterly fundamentals. Revenue, earnings, EPS and cash flow are trailing-twelve-month figures: up to August 21, 2026, Sharadar's, dated on the day they were filed with the SEC; since then, rebuilt quarter by quarter from the SEC filings - including the fourth quarter, which no company reports on its own and which we derive from the annual accounts minus the nine-month cumulative. Each figure enters the model on the day it was actually filed, not the day the period closed. The market couldn't read those numbers before they were published, so neither can the model.
- Total return. Dividends are credited at their ex-date and prices are split-adjusted, applied identically to the strategy and to the benchmark.
- Same rules for every period. Monthly rebalancing, equal weighting, Z-Scores winsorized at ±3σ. No parameter is re-tuned year by year to make the curve look better.
What it still can't fix
- Before 1998 there is no quality data within our reach. Since September 2026 the history is computed with Sharadar, a paid database with every company that has been in the index —including the ones that disappeared: in January 2011 we had prices for 68.6% of the index, and now for all of it— and with every accounting figure dated on the day it was filed with the SEC. Its prices start on December 31, 1997 and the momentum factor needs twelve months of history, so the first purchase is in December 1998. Further back, company-by-company accounts only exist in Compustat, beyond our reach. The backtest runs up to August 21, 2026; from then on, each month is computed by the daily cycle with the usual data, so after that date the curve is lived, not rebuilt.
- No source is perfect. From 2010 to 2026 we crossed Sharadar with our usual data company by company: 95% of days match to within 0.1%, and what does not match is explained, almost always because Yahoo skipped a spin-off that Sharadar does carry. Three operations Sharadar does not record —the Talen spin-off from PPL (2015), Iron Mountain's stock dividend (2014) and the Waters shares handed to BDX shareholders (2026)— are corrected by hand and declared in the code. Before 2010 there is no second source to check against: if Sharadar misses any other in those years, we do not know.
- The curves are gross, and the cost is measured separately. No curve on this site deducts commissions, spread or taxes: what you see in your account will be less. How much less is measured over the 330 rebalances of 1999-2026: on €100,000 at €0.30 per trade it is 31 basis points a year assuming a 3 bp spread, and the net comes out at 11.03%, above the index (10.18% for the equal-weight index after the 0.20% a year charged by RSP, the ETF that tracks it). But with the spread our estimators infer from daily highs and lows —an upper bound— the net drops to 8.32-8.94%, below the index: the gross edge is small and costs can eat all of it. The breakdown and the sensitivity to the spread are above; the Model Portfolios page works it out with your own commission and capital. Taxes are still not included: that depends on where you live.
- It is one market. The walk-forward starts in December 1998: 27.6 years and 332 rebalances, with the dot-com bust, the 2008 crisis, the COVID crash and the 2022 rate rise inside. (Three similar numbers live on this page and each counts something different: rebalances are the buy decisions; the months that can be measured are one fewer, because the last one has not closed yet; and the signal test uses one fewer still, because it needs the following window to know whether it was right.) That is many cycles, but of a single market —the 500 largest US companies— and what worked here need not work elsewhere.
- The edge is measured, but not proven. Across the 332 rebalances, the multifactor top 20 beats the equal-weight S&P 500 by +0.10% a month - which compounds into the 1.0 percentage points a year of the equal-weight comparison - and wins in 187 of 331 months, 56%. The direction is consistent, but the margin is not large enough against its own variability: if the true edge were zero, there would still be roughly a 33% chance of seeing 27.6 years this good or better purely by luck. Put as a range, which is more honest than a single number: with a block bootstrap (10,000 resamples, six-month average blocks), the Top 20 portfolio's annual edge over the equal-weight index lies between −4.2 and +6.3 points at 90% confidence and between −5.2 and +7.3 at 95%: zero is inside both, and in 37% of the resamples the edge comes out negative. Measured factor by factor, none of the four on its own clears that bar. We are not saying it does not work: we are saying that not even 27.6 years and five hundred companies are enough to separate it from noise in the equal-weight version, and anyone investing should allow for that. We also measured it with the standard test of factor research (the Fama-MacBeth IC: does the score rank the ~500 companies each month?). On the long history —331 months from December 1998 to August 2026, with every company that was in the index and total return— it comes out at t = 1.41 (mean IC +0.012;
medir_senal.py, Sep 11, 2026). On the previous history (185 months, 2011 to 2026) it came out at t≈0.7, and measured only on today's survivors it would have come out at t=2.1, which would already "prove" the edge: we do not publish that number, because it is not the signal, it is survivorship bias. At the measured effect size it would take about fifty-five years of data to separate it from noise: twice as many as exist. And year by year there is no stable edge, there are streaks: the multifactor IC is positive in 17 of 28 years (1999-2026), with a +0.08 in 2015 followed by a −0.11 in 2016, and by five-year stretches it goes from +0.016 (1999-2003) and +0.027 (2004-08) to practically zero between 2009 and 2023, and to +0.040 since 2024. The good stretch is precisely the recent one, the one a newcomer sees today: it is not the norm. Anyone following this should treat whole negative years as part of the deal, not as a sign that something broke.
How it did in each kind of market
The whole-window figure hides where it comes from: a strategy can win overall and have made all of it in a single kind of market. Here the 20-stock multi-factor history is split at the peaks and troughs of the S&P 500, fixed before looking at any result (datos/PREREGISTRO_BACKTEST_LARGO.md), in both of its versions.
Cap-weighted, against SPY
It wins in 5 of 8 periods and loses in 3. Annual return (CAGR) within each period.
| Period | Dates | Portfolio | SPY | Difference | Worst drop | Months behind |
|---|---|---|---|---|---|---|
| Dot-com bubble | Dec 1998 – Mar 2000 | 35.4% | 11.5% | +23.9 | −16.0% | 8 |
| Dot-com bust | Mar 2000 – Oct 2002 | −22.5% | −18.3% | −4.2 | −52.5% | 31 |
| Bull market 2003-2007 | Oct 2002 – Oct 2007 | 27.7% | 16.6% | +11.0 | −11.7% | 17 |
| Financial crisis | Oct 2007 – Feb 2009 | −38.0% | −34.0% | −4.0 | −48.4% | 13 |
| Growth decade | Feb 2009 – Feb 2020 | 17.5% | 15.4% | +2.1 | −18.5% | 66 |
| COVID and rebound | Feb 2020 – Dec 2021 | 9.5% | 21.2% | −11.7 | −26.5% | 14 |
| Rate hikes | Dec 2021 – Sep 2022 | 4.7% | −18.8% | +23.5 | −13.2% | 3 |
| Mega-cap concentration | Sep 2022 – Aug 2026 | 39.0% | 20.0% | +19.1 | −25.9% | 14 |
| Whole window | Dec 1998 – Aug 2026 | 13.7% | 8.7% | +5.0 | −57.2% | 66 |
Equal-weight, against the equal-weight S&P 500
It wins in 4 of 8 periods and loses in 4. Annual return (CAGR) within each period.
| Period | Dates | Portfolio | Equal-weight | Difference | Worst drop | Months behind |
|---|---|---|---|---|---|---|
| Dot-com bubble | Dec 1998 – Mar 2000 | 22.4% | 0.3% | +22.0 | −9.2% | 9 |
| Dot-com bust | Mar 2000 – Oct 2002 | −15.5% | −5.8% | −9.7 | −37.9% | 31 |
| Bull market 2003-2007 | Oct 2002 – Oct 2007 | 26.4% | 22.6% | +3.8 | −12.6% | 21 |
| Financial crisis | Oct 2007 – Feb 2009 | −42.4% | −37.2% | −5.2 | −55.4% | 7 |
| Growth decade | Feb 2009 – Feb 2020 | 14.7% | 16.9% | −2.2 | −20.2% | 131 |
| COVID and rebound | Feb 2020 – Dec 2021 | 2.1% | 19.9% | −17.8 | −27.3% | 19 |
| Rate hikes | Dec 2021 – Sep 2022 | 10.8% | −12.0% | +22.9 | −14.5% | 3 |
| Mega-cap concentration | Sep 2022 – Aug 2026 | 32.9% | 14.3% | +18.6 | −16.7% | 19 |
| Whole window | Dec 1998 – Aug 2026 | 11.4% | 10.4% | +1.0 | −61.1% | 314 |
The worst drop is measured on the monthly points of the curve, so it falls somewhat short of the daily one. "Months behind" is the longest run of consecutive months in which the portfolio does not get back to its best moment against the benchmark: what someone following it lives through. The last period is still open and grows every month.
What to expect if you start today
This is the fine print, and we run it in large type. Shown before you put in a euro, it lets you decide; hidden until the first bad year, you have already decided without knowing it.
The figures above cover 27.6 full years. Nobody invests that many years at once: you start on one day and check after one, three or five. This is what the same curve would have given anyone starting in any month since December 1998, in the equal-weight version of the portfolio and always against the equal-weight S&P 500 with dividends:
- Over one year (320 windows): half the time the portfolio beats the equal-weight index by between −6.0 and +12.0 points; the median is +2.7. It loses to the index in 42% of the years. Worst: −59.7 points (May 2020 to May 2021); best, +57.9. In absolute terms, one year of this portfolio has ranged from −26.6% to +46.5% (5th and 95th percentiles) and ended in a loss 21% of the time.
- Over three years (296 windows): median annual edge of +0.4 points, between −5.5 and +6.2 for the middle half. Behind the index in 49% of three-year periods; the worst ran −17.5 points a year behind for three years (from June 2018).
- Over five years (272 windows): median annual edge of −0.5 points, between −4.6 and +2.0 for the middle half. Behind the index in 55% of five-year periods: five years do not guarantee beating the index. Worst, −9.4 points a year from January 2008; best, +20.1 from June 2021, which is the stretch anyone starting now is looking at.
- How long you sit behind the index. The portfolio/equal-weight index curve has spent 99% of months below its previous high. After each high it takes a median of 6 months to beat it, but the longest stretch lasted 314 months (March 2000 to May 2026, reaching −53% against the index at the bottom). If you cannot watch the index stay ahead for 26.2 years, do not follow this portfolio.
- The luck of your start day. The published curve rebalances on specific dates, and that date moves the number. Rebuilt on all 21 possible schedules (rebalancing shifted by 0 to 20 sessions, same rules and data, measured from January 2011 to August 2026), the CAGR ranges from 14.3% to 17.4% (median 16.0%) and the annual edge over RSP from +2.6 to +5.5 points; the maximum drawdown from 33% to 40%. The schedule this site uses comes in at number 3 of 21, +0.7 points from the median: the number we publish sits in the upper part of what the same rule produces. Rebalancing every two months instead of every one, the CAGR lands between 13.1% and 17.2% (42 possible schedules).
medir_suerte_calendario.pymeasures it with the same data and the same rules as the published curve: none of these numbers is written by hand. - What it costs to follow. 6 of the 20 positions change at the median rebalance and the portfolio is equal-weighted again every month: 311 trades a year, measured over the 330 rebalances of 1999-2026 by
medir_costes.py. On €100,000 at €0.30 per trade, commissions come to 9 basis points a year and market impact 10. The only piece we do not measure is the spread: assuming 3 basis points — the order of magnitude for a large S&P 500 stock — the total cost is 31 bp a year and the net CAGR 11.03% against 10.18% for the equal-weight index after RSP's fee. At this turnover, every basis point of spread costs 3.8 bp a year, so you can plug in your own: 1 bp gives 23 and 5 bp gives 39. Our two estimators infer the spread from daily highs and lows and return 52-66 bp, an upper bound for stocks this liquid; with those the cost would be 219-274 bp and the net 8.32-8.94%, below the index. On €10,000 instead of €100,000 the fixed commission weighs ten times more and the total cost rises to 296 bp. And paying currency conversion on every trade adds 269 bp. Taxes are not included: the portfolio realizes gains every month while an index fund defers them.
None of this is a forecast: it is the spread of what already happened, before costs, in a market that rose in most of those years and went through three bear markets. Over those 27.6 years, 13 of the 20 portfolios beat the equal-weight index, and all of them spent whole years behind it. Both halves are true, and anyone starting today has to count on both: that is why both are here.
If you are going to follow this, do it with an account
The thing almost nobody catches in time: the big insider buys — executives and major shareholders putting in their own money — the same day they report them to the SEC, with the Form 4 linked. It takes three steps and none of them costs anything: create the account, confirm the email we send you, then switch the alert on in your profile (the icon at the top right). It starts off, it goes off again with the same click, and nothing else lands in your inbox. Along the way all three are saved: your simulator portfolios, your personal portfolio and your Academy progress, synced across devices.
A backtest is a stress test of a rule, not a forecast of your account. We publish the limitations because a platform that only shows you the flattering half of its own methodology is selling you something. We are not: a broker, a fund, a signal service, or a licensed advisor. Nothing here is a personal recommendation - the same numbers are shown to everyone, and the decision is entirely yours.
See What the Numbers Say Today
The full S&P 500 ranking, the model portfolios and the simulator are open right now - no account, no card, no email.
Two more things, free and with no account
Both come from the filings companies send the SEC. Every row links to the original document.
Frequently Asked Questions (FAQ)
How the model works, what it does not do, and what the published figures mean. Anything not answered here is set out in full on the methodology page.
1. Basic Concepts for Beginners
What exactly is the S&P 500 and why is it the yardstick?
The S&P 500 is the index of the 500 largest listed companies in the United States: Apple, Microsoft, Amazon, Nvidia and four hundred-odd others. Buying it means owning a slice of all of them at once instead of picking, which is why it is the usual yardstick — any strategy that selects stocks has to beat it for the trouble to be worth anything. What it is not is a guarantee. It falls when the market falls, it has had long stretches with no gain at all, and this site publishes the drops as prominently as the returns.
What does "quantitative analysis" mean in simple terms?
It means deciding with measurable numbers instead of impressions. Instead of reading the news on fifteen well-known companies and picking whichever sounds best, the same formula is applied to all ~500 companies in the index and the result is sorted. Nothing has favorites, nothing gets tired at company number four hundred and nothing makes an exception because it happens to like a company.
Two nearby things are worth not confusing it with. It is not trading: nothing is bought or sold here, a list is published and you place the orders at your own broker. And it is not real time: the figures are recalculated three times every trading day, which is more than enough for a portfolio reviewed once a month.
Do I need deep mathematical or financial knowledge to use Quant500?
No. The four factors have been published in academic journals for decades (Jegadeesh and Titman, Fama and French, Novy-Marx) and what this site does is apply them to the ~500 companies in the index every day and show the result sorted. The model does the heavy computation; your job is to read a table. That said, we'd rather you understood what you're looking at than trusted it blindly: every formula on this page is written out, and the Quant Academy explains each factor from zero. Using the numbers takes five minutes; understanding why they say what they say is what will actually keep you invested when the curve goes down.
How is Quant500 different from buying stocks on my own?
Buying stocks because you "have a good feeling", because you read a piece of news, or because you like the brand, is basically playing the lottery at a casino. Quant500 does not guess or play by chance. Our system analyzes fundamentals filed with the SEC (earnings, revenue, margins, debt, cash flow) and price trends and ranks all ~500 companies on the same criteria, every day, without deciding it likes one of them. It won't tell you which stock will rise - nothing can. What it removes is the part you're worst at: the story you tell yourself about why this one is different.
Is this the same as using ChatGPT or other artificial intelligence to invest?
No. Generative AI (like chatbots) is designed to "guess" the next word in a sentence based on text patterns; it knows nothing about finance or real risk. In contrast, Quant500's models are deterministic statistical and quantitative models. They are based on strict mathematical rules, covariance formulas, and time-series analysis, with decades of published academic research behind them. It is not a chat with a bot: it is the same arithmetic applied to every company, and anyone can redo it.
2. The model and how it decides
What exactly does the model look at to pick companies?
Four factors, the same ones for all ~500 companies: Momentum (how the price has behaved over the last twelve months, skipping the most recent one), Value (what the company costs relative to what it earns, what it is worth on the books and what it sells), Quality (return on equity and assets, margins and debt) and Growth (how much revenue, earnings per share and cash flow have grown in a year).
And here is a misunderstanding worth clearing up: it is not a sieve. There are no filters to pass one after another. Each factor becomes a comparable score (a Z-Score) and the final ranking is the arithmetic mean of the four. That means a company can make it into the portfolio with one clearly poor factor if the other three make up for it, and more than one does. The portfolio is built by taking the top N on that mean, highest first, with no further screening.
Why does a rule beat intuition over a full cycle?
The investor's biggest enemy is their own brain. Human beings are evolutionarily programmed to feel paralyzing panic when the market crashes, and irrational greed when the market rises non-stop. This leads us to chronically buy high and sell low. The model, by contrast, is a rule: it does the same thing whatever is going on. If the data changes, the weights change - at the next rebalance, by the same rule, with no hesitation and no need to be right about anything. The model is not smarter than a good analyst. It is simply incapable of panicking, and over a full market cycle that turns out to matter more.
What is the "Z-Score" that appears in the tables?
It is the standard way of putting things measured on different scales onto the same one. It says how many standard deviations a company sits away from the index average on that particular metric: a +1.5 in Quality means one and a half deviations above the S&P 500 average on quality, and that number can then be averaged with the Value score, which in raw form would live on a completely different scale. Without that step there would be no way to combine the four factors at all.
It is not secret and it is not ours: it is first-year statistics, the formula is written out further up this same page, and anyone can redo the calculation from the SEC filings. What is worth knowing is its limit: a high Z-Score describes the recent past, it does not promise the future, and we clip it to the range [-3, +3] so a single anomalous figure cannot flatten everyone else (that too is explained, with its cost, in the methodology).
How often is the portfolio recalculated?
The data is refreshed three times every trading day, but the portfolio is not rebuilt at that pace: the backtest calendar is monthly (every 21 sessions, the only cadence we have measured). Monthly is the only cadence we have measured, and we do not claim it is the best one. What rebalancing costs — commissions, spread and turnover — is measured over the real rebalances and published in the methodology, not estimated.
What is "Max Pain", and does it feed the model?
The derivatives (options) market is where funds and market makers take positions that do not show up in the share price. "Max Pain" is the exact price of a stock at which the overwhelming majority of option buyers would lose their money, benefiting the market makers. There is a documented tendency for prices to gravitate toward that level as expiry approaches, though the effect is statistical, contested in the academic literature, and far from reliable on any single day. We surface it as one more piece of context - useful for understanding where the derivatives market has its money parked, never as a prediction of tomorrow's price.
3. Risk, drawdowns and what the model does not protect you from
What happens to my money if the market suffers a sudden crash or collapse?
Straight answer: a portfolio of S&P 500 stocks falls when the S&P 500 falls, and ours is no exception. Any platform telling you otherwise is selling you insurance it cannot underwrite. What the model does do is stay disciplined exactly when a human wouldn't: it keeps ranking on the same rules, and at each rebalance it rotates out of names whose fundamentals and momentum have broken down - without waiting for you to find the courage. The Quality factor in particular tilts toward low-debt, high-margin businesses, which have historically fallen less than the index in drawdowns. That is a cushion, not a shield. This is why we publish the Max Drawdown of every model portfolio next to its return: look at it before you look at the CAGR, and size your position for the worst number you see there.
What is the "Max Drawdown" and why should it be your main obsession?
The "Max Drawdown" measures the greatest historical abyss: the maximum percentage drop a portfolio has suffered from its highest peak to its deepest valley. It is the figure that decides whether you would stay with a strategy, which is why we publish it next to each portfolio return instead of hiding it. Look at it without decoration: the twenty model portfolios, equal-weighted, have gone through drops of between 50% and 82%, and the equal-weight 20-stock multifactor fell 63.5% at its worst. A 50% fall needs a 100% gain just to get back to where you started. Size your position for the worst number you see there, not for the annual return.
Let's be honest, does the model guarantee profits every single month?
If someone in the finance world guarantees you profits every month, run fast because you are being scammed. No system in the universe can defy global markets every day without stumbling. What we do guarantee is that the same rules get applied to every company, every day, and that the numbers we publish are the ones the rules actually produced. What those numbers say, over our 27.6-year walk-forward (332 rebalances, 1998-2026): 13 of the 20 beat the equal-weighted S&P 500 (10.4% a year), which is the fair comparison because our portfolios also weight every holding the same. In that simulation, our 20-stock multifactor returned 13.7% a year cap-weighted, which over those 27.6 years would have turned 10,000 into 348,495 against 99,881 for the S&P 500. And the outperformance is not free: our 20-stock multifactor fell 59.9% at its worst moment, and, in their equal-weight versions, all twenty go through drops of between 50% and 82%. Against the S&P 500 bought through a fund, and again equal-weighted, 17 of the 20 return more, but 15 of the 20 fall deeper. (These figures are recomputed from the published files every time the walk-forward is extended.) Anyone telling you that a strategy beats the market without suffering is selling you something: in the simulation, this one beat it while taking the same drops or worse, and neither past nor simulated returns guarantee future ones.
What does diversification protect me from, and what not?
Betting 50% of your money on a single stock hoping it skyrockets is playing financial Russian roulette. The simulator lets you choose how weight is distributed, and each method is a different answer to the same question: equal weight (every position identical, no single bet dominates), by model score (more capital to the highest-ranked names), risk parity (weight inversely proportional to each stock's volatility, so the turbulent names don't drive the whole portfolio), or by market cap (mirroring how the index itself is built). Diversification doesn't cancel risk - in a market-wide crash correlations converge toward one and everything falls together - but it does remove the risk of a single company ruining you.
Why is the "Sharpe Ratio" the metric everyone talks about?
The Sharpe Ratio measures how much extra return you get for every unit of risk you endure: the return above the risk-free asset, divided by volatility. It is useful for comparing two strategies that return the same but do not feel the same along the way; 30% a year with violent swings and 30% a year steadily are not the same investment, even though the headline is identical. That said, it is not an absolute grade: Sharpe penalizes upside volatility exactly like downside volatility, it is computed on the past, and it moves a lot depending on the window you measure. That is why we also publish Sortino, Calmar and, above all, the maximum drawdown.
4. Platform Usage and Next Steps
How should I interpret the "Historical Wealth Evolution" chart?
When the simulation finishes the chart shows two lines. The faint dotted blue one is the benchmark. It is not the traditional S&P 500 index, which is weighted by company size: it is the equal-weight S&P 500: the equally weighted average of every company that was in the index on each date, with its dividends. We compare against that because our portfolios are equal-weighted too, and comparing an equal-weighted strategy against a size-weighted index would credit the strategy for a difference that is just the weighting. The engine computes it with the same data as the portfolio, and it is not RSP —the equal-weight ETF, which really trades— because RSP does not exist before May 2003 and the history starts in 1998. For a few weeks we used RSP precisely because an average of our own could only average the companies whose prices we held, and it inherited the survivorship bias it was supposed to judge. With Sharadar data we hold the price of every one of them, including those that disappeared, and where RSP exists the two move almost alike: ours beats it by a little over half a point a year, which is what the fund charges and pays to trade. On the other hand, the upper line (green or red) is your model portfolio. It shows how the rules you chose would have behaved in the past - sometimes above the benchmark, sometimes below it, and the gap between the two lines is the only thing worth studying. Read the drawdown as carefully as the final figure: a curve that ends higher after a 45% collapse is a strategy most people would have abandoned halfway. And remember what a backtest is not: it excludes commissions, spread and taxes. It tells you whether a rule was sound, not what your account will do.
Can I take this mathematical portfolio and replicate it in my real broker?
That is what this site is for. We are not a broker that custodies your funds, and we don't tell you what to do with your money: we publish what the rules say, and the decision is yours. When the model finishes its calculations, it shows the list of companies with their weight. The twenty model portfolios are equal-weight: in the 20-stock one each position is 5.00%, in the 10-stock one 10.00%. There is no weight optimization and no different percentage per company; the only place weights vary is the Simulator, if you pick another allocation yourself. From there you place the orders yourself, at your own broker — which also means the commissions, the spread and the tax are yours, and none of the three are inside the published curve.
What level of fundamental company data will you provide me?
In each company's profile you get what its filings say: profit margins, free cash flow growth, debt ratios and institutional ownership (what share of the company is held by funds). What we do not put here is where Wall Street analysts say the price is going: that is an opinion, and this site publishes what it can measure.
Are dividends accounted for in the performance simulations?
Yes. Every curve is computed as total return: each dividend paid is reinvested on the day it is paid, in the portfolio and in the benchmark it is measured against. Counting them on one side and not the other is the easiest way to make a strategy look better than it was.
What do the options figures on a company profile tell me?
The stock market is the present, but the options market (derivatives) is where large institutional investors take positions on the future. The options tab of each company profile publishes what that market is pricing in for the stock: the expected move to expiry, max pain, the put/call ratio (by volume and by open interest) and the implied volatility of the near-the-money contracts. When puts trade considerably more expensively than calls, there is heavy demand for protection against a fall; that is a fear gauge, and it is real information that is hard to come by.
It is published as context, never as a prediction. We have not measured that this bias anticipates falls — and on this site a signal is not published as predictive until it passes the same test as the four factors — so it does not enter the score of any portfolio. It is good for one specific thing: knowing how much movement is already in the price before you assume a piece of news is a surprise.
What information is detailed in the "Valuation" section of the analysis panel?
The Valuation tab gathers the company's market and accounting figures, from the last data downloaded:
- Market Metrics: Current price vs. 52-week highs/lows, Market Cap, and Enterprise Value (EV).
- 52-Week Range Bar: Visual indicator placing the current price within recent history to detect overbought or oversold zones.
- Financial Multiples: Analysis of P/E (Current and Forward), PEG, P/B, and Enterprise Value to EBITDA ratios to gauge the stock's "cheapness".
- Where the P/E falls: A bar placing the current P/E on a 0x to 60x scale, with the cut-offs written out (15x, 25x, 40x). It does not say whether the company is expensive or cheap: that depends on what happens next, and nobody knows. The real comparison is further down the same profile: the Value Z-Score and the rank, measured against the whole index.
- Dividend Health: Yield, payout ratio, and key ex-dividend dates.