The close falls outside the 7,468–7,636 range the model published that morning: the S&P 500 rises 1.14%
The session is over and here is what it did. In the morning we publish a range for the close; at night we say whether it landed inside, hit or miss, and the running score sits a few paragraphs below. Below that: who enters and who leaves the ranking, what executives bought with their own money, and who reports next trading day.
The close fell outside the range
In the morning the model published a range of 7,467.55 to 7,636.07 for this session's close, on data through Sep 16, 2026. It closed at 7,637.76, a 1.14% move on the previous close. On whether the day would move more than usual (the threshold was 0.449%), the model said "yes" and the answer was "yes": hit.
01The call for the next session, September 18, 2026
Computed from today's close. It is not this session's call, whose result is at the top: this one is checked when that session closes.
The range is the close ± 2.31 times the expected move, that has contained the close eight times out of ten over the last 1,079 out-of-sample sessions. It is neither a ceiling nor a floor: two times out of ten the close has landed outside it.
Historical frequency, not a forecast. We measured direction with the same model and the same protocol over 1,079 out-of-sample sessions: it is right 53.48% of the time and saying "up" every day is right 53.66%. There is no signal; the detail is in section 08.
Validated over 1079 out-of-sample sessions, walk-forward: each day is predicted training only on earlier ones, retraining every 21 sessions. It is right 59.87% of the time against 51.16% for the majority class (always calling the more frequent outcome). Brier 0.23433 against 0.24987 with no model.
Of the 139 variables computed (the index, the 500 stocks one by one, VIX, rates, gold, copper and the calendar) the model keeps the 120 most correlated with the target every 21 sessions, chosen using past data only. Today's are listed one by one in prediccion.json.
How to read a 71.9%
A 71.9% does not mean "no". It means that in seven of every ten days like today, the market moves more than usual. If tomorrow turns out turbulent, the model did not fail: the 71.9% happened. That is why the running score above is what counts, and not the promise behind any single figure.
02What changed in the ranking
From the session of Sep 16, 2026 to that of Sep 17, 2026, across the 466 scored companies.
The ranking is recomputed every session, and this is what moved from one to the next. Next to each move is the factor that produced it: momentum changes every day, value whenever the price does, and quality and growth only when new accounts come in. Why the price moved is not in this data, so it is not told.
| Company | Top 20 | What moved |
|---|---|---|
| RDDT in | #21 → #20 | Momentum 3.6 → 3.5 |
| CIEN out | #19 → #23 | Momentum 8.4 → 8.0 |
The biggest moves in the whole index: TRGP climbs 54 places, from #141 to #87 (Momentum 7.4 → 8.2) and FERG drops 40 places, from #298 to #338 (Momentum 4.9 → 4.5).
Source: rankings_diarios.json.gz · rank by multi-factor (the average of the four), across the 466 companies with all four computed · scores on the 0-10 scale used on company pages, where 5 is the index average. See all 500 scored →
03Who filed, and what it did to them
The 1 companies that filed with the SEC on September 11, 2026, and how their score moved once the new numbers went in.
The section above shows the ranking moving; almost always it is the price moving it. This is the other thing: a quarterly or annual report rewrites value, quality and growth at once, and those are audited accounts, not a quote. Here is the before and after, on the 0-10 scale used on company pages, where 5 is the index average.
| Company | Rank | Multi-factor | What moved |
|---|---|---|---|
| ORCL | #434 → #389 (+45) | 4.3 → 4.5 | Momentum 3.3 → 3.0, Value 4.1 → 4.3, Quality 4.3 → 4.2, Growth 5.5 → 6.6 |
Filing date on EDGAR, not the quarter end. Scores come from the rankings calendar: the same ones each company page publishes. No opinion on the results, only the measured effect.
04What the executives did
SEC Form 4, session of September 16, 2026. 1 sale, 1.7 million dollars.
The rare thing on a Form 4 is a purchase. The largest on September 15, 2026: Murdoch Lachlan K, Executive Chair, CEO at Fox, bought 149,934 shares at $68.53 (10.27 million dollars), code P, meaning on the open market and with their own money. see the Form 4 →
A buying campaign, not a trade. Cascade Investment, L.L.C., Gates William H Iii, 10% Owner at Republic Services, has bought on 24 consecutive trading sessions, Aug 10, 2026 to Sep 11, 2026: 6,254,462 shares for 1,380.3 million dollars, and after the last one declares 117,058,444 shares held. see the latest Form 4 → It is the largest by money of the 11 runs of three or more consecutive sessions in the Form 4 filings downloaded between Aug 25, 2025 and Sep 16, 2026. That someone buys says nothing about where the price goes: what is checkable is that they filed it with the SEC session after session.
3 purchases were filed that day across 3 companies. Over the 31 days the file covers, Aug 17, 2026 to Sep 16, 2026, there are 186 purchases worth 1,470 million dollars across 62 companies, against 1962 sales: which is why a purchase gets told and a sale gets tabulated. The largest of the whole window was RSG, 90.70 million on September 8, 2026.
The biggest sale of the day tells you nothing. Hershey Trust Co Trustee In Trust For Milton Hershey School, of Hershey, sold 1.7 million dollars of stock and still holds 825,749 shares: that is 1.20% of the position, and the sale was scheduled months in advance under a 10b5-1 plan. That is tax plumbing, not a view on the company.
The only sale filed that day was scheduled.
| Company | Role | % of holding | Value | Left |
|---|---|---|---|---|
| Hershey Hershey Trust Co Trustee In Trust For Milton Hershey School · largest of the day by dollar value | 10% Owner | 1.2% | 1.71 M$ | 825,749 sh. |
Source: Form 4 filings with the SEC · alertas_insiders.json, session of Sep 16, 2026. Every name links to its filing on EDGAR: the claim and the proof on the same line.
The context over those same 31 days: 1962 sales worth 17,542 million dollars (3 with no declared price, outside that total), of which 772 were discretionary, and of those 210 exceeded 25% of the seller's position. That is where to look.
Before drawing conclusions
Executives sell for divorces, mortgages, taxes and school fees, not only because they see clouds. The SEC does not require a reason, and we do not know it.
What we do know is that these people chose the timing. It is a map of where to look. It is not a sell signal.
05The five portfolios that can actually be audited
Rebuilt month by month using only what was known on the day.
| Portfolio | Cumulative | CAGR | Sharpe | Max DD | This year | Last month |
|---|---|---|---|---|---|---|
| Value 100Earnings, book and sales yield | +1,848.1% | 11.34% | 0.48 | -74.70% | +19.58% | -2.58% |
| Momentum 10012m return ex-latest month / volatility | +1,673.7% | 10.97% | 0.52 | -55.85% | +21.08% | -1.41% |
| Quality 100Balance-sheet and earnings quality | +1,713.9% | 11.06% | 0.52 | -50.74% | +9.55% | -2.55% |
| Growth 100Revenue, EPS and cash-flow growth | +915.2% | 8.75% | 0.40 | -63.37% | +17.19% | -2.76% |
| Multi-factor 100All four factors combined | +1,582.0% | 10.75% | 0.50 | -59.72% | +21.18% | -2.25% |
| equal-weight S&P 500 computed by the engine, dividends reinvested · this year and last month from RSP | +1,428.0% | 10.37% | 0.49 | -60.41% | +12.20% | -2.95% |
Growth 100 falls short of the benchmark over 27 years: +915.2% against +1,428.0%. The weakest is Growth 100, with an alpha of -1.42%: negative, meaning what it earns does not pay for the market risk it takes to earn it. Publishing the figure that comes out worst is the only reason the others deserve any credit.
That benchmark is the whole index, measured like the portfolios
The +1,428.0% in the row above is the equal-weight S&P 500: every company in the index each month, weighted the same, dividends reinvested, computed from the same data and with the same method as the portfolios, including the companies that later disappeared. A fund that tracks it, RSP, has existed for the last 23 years, and over that time the two differ by 0.57 points a year: the fund's fee and trading costs. Neither this benchmark nor the portfolios deduct costs: the comparison is like for like, but neither is what you would see in your account.
Over this same window and measured the same way, SPY, the plain cap-weighted S&P 500, returned +898.8% (8.68% a year). SPY weights each company by its size; a portfolio that weights everything the same is really measured against the equal-weight index, and SPY sits alongside as what everyone calls the index.
How much does trading take? It is measured, though not for these three: costes_estimados.json computes it on multi-factor 20, and in the cheapest scenario (a 100,000 EUR account held in dollars, 0.50 EUR a trade, no currency conversion) it comes to between 202 and 253 basis points a year, which turns its 11.37% gross into 8.56-9.11% net. It is an upper bound: the spread is not measured but estimated, and both estimators run high on liquid names. Measured on Sep 11, 2026 with data through Aug 21, 2026.
Source: carteras.json · index based at 100 on Dec 31, 1998. Walk-forward with point-in-time data: on each date scoring uses only prices up to that day and fundamentals already filed with the SEC. 331 rebalances through Aug 21, 2026.
06The rotation between factors
How far value is ahead today, and what happened the other times the gap opened this wide. Six-month window ending August 21, 2026.
Over the past month the value portfolio lost -2.58% while momentum lost -1.41%. Across that six-month window the benchmark rose 9.1%, so this is not the whole market moving: it is a rotation inside one.
The extreme of the series is March 2, 2000, 27 years ago, at +48.0 points. And what happened after each extreme? Counted, not assumed: of the 5 extremes in momentum's favor with enough data after them, in 2 the lead had flipped sign six months later; of the 5 extremes in value's favor, in 3. That is a count over 10 episodes, not a rule.
Why this is not a forecast
325 overlapping windows are not 325 independent observations: each shares five of its six rebalances with the next, so any series looks mean-reverting whether it is or not. All that can be claimed is descriptive: the gap between the two factors is -0.8 points and last month belonged to Momentum 100. What happens next month is not in this data.
Computed from carteras.json · 331 monthly rebalances between Feb 2, 1999 and Aug 21, 2026 (27 years) · 325 overlapping windows of 6 rebalances · recent stretch: the last 24 rebalances, the shortest run that still holds 4 non-overlapping windows.
07What is coming: who reports next
1 S&P 500 company on September 22, 2026, out of 17 across the whole market.
With the rank and the score they carry before reporting. Section 03 will publish the after once the document reaches EDGAR, which is zero to three days later: the press release does not move our scores, the filed accounts do.
| Company | When | Rank | Multi-factor |
|---|---|---|---|
| AZO AutoZone, Inc. | before the open | #417 | 4.4 |
Source: calendario_resultados.json · Nasdaq public earnings calendar. Earnings date announced by each company, not the SEC filing date, which is not published in advance. Scores from the Sep 17, 2026 ranking.
08The limits, and what they teach
A fixed section. Knowing where a model stops is worth as much as knowing what it gets right.
- What insiders did yesterday. Form 4 can be filed up to two business days after the trade, so the freshest data is always the day before last.
- Whether momentum keeps winning. 27 years and 331 rebalances are not enough to claim a factor mean-reverts. We can describe what happened, not what comes.
- What any company is worth. This report computes no price targets. No Monte Carlo, no DCF, no fair value. When you see one elsewhere, ask which cost of capital discounted it and who chose that number.
- Whether the S&P rises tomorrow. We measured it with the same model (139 variables) and the same walk-forward protocol, over 1,079 out-of-sample sessions: the model calls direction right 53.48% of the time and saying "up" every single day is right 53.66%. The gap is noise: there is no signal, so we don't sell one. Both figures are recomputed every edition.
If you are going to follow this, do it with an account
Free: save your simulator portfolios, your personal portfolio and your Academy progress. And 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. One click in your profile turns it on; one click turns it off.
This is not investment advice. Quant500 publishes data and its historical reconstruction. No figure here is a suggestion to buy or sell, nor does it account for your personal situation.
Closing levels of ^GSPC, ^VIX and ^TNX (Yahoo Finance) · session of Sep 17, 2026.
Report generated automatically on September 17, 2026 from the server's data files. Each section states its own cut-off date, which is not always today.
Historical Daily Market Reports Archive
Browse all previous opening and closing market report editions.