Crossing 65 Was Not a Buy Signal Either

The Blitz Index's own success line is 65. When a company climbs across it, the business has passed the bar — so the stock should follow. Seven crossings, one pre-registered test, and a median of −5.9% over the following year. The +3y number looks tempting. We refused to let it overturn the verdict, and this article explains why.

5 min read · 7 crossing events · pre-registered hypothesis H1 · Not investment advice

Why the hypothesis was attractive

Sixty-five is the number that organizes BlitzLab. It is the published success threshold of the v1.0 model: S and A sit above 90 and 80, B spans 60 to 79, and the line at 65 is where we say a blitz is working. Twenty of the forty-seven scored companies clear it; twenty-seven do not. A threshold that separates the board into winners and the rest invites the most natural trading question imaginable: what happens if you buy the moment a company crosses it?

The logic is seductive because it is clean. A score crossing 65 means our scholar's assessment says the business has, on the evidence, entered the zone where blitzscaling works. If the index measures what it claims to measure, the crossing should be news — and news about improving fundamentals should show up in the price. That is hypothesis H1, registered before any data was touched, in the same pre-registration document that produced the profitability-inflection rejection and the score-slope rejection: a crossing from below 65 to at-or-above 65 should deliver better forward returns than doing nothing, and better than crossings at other lines (50, 80). The falsification rule was mechanical: if the median forward return of crossings is at or below the unconditional median of the panel, reject.

Two companion analyses were registered alongside: crossings at 50 and at 80, for comparison. The 50-line analysis died before it began — more on that below — and the 80-line analysis produced two observations, both from the same company. This article is, therefore, mostly about seven events at one line. Honesty about that scarcity is part of the method, not a footnote to it.

The method, in brief

A crossing is defined strictly: the previous year's score below the line, the current year's score at or above it, on the annual estimated-score panel of fourteen companies. Entry is set at the first trading day after that year's 10-K (or 40-F) filing, pulled from EDGAR submissions, originals only — so there is no look-ahead: nobody could have known the score before the filing. Returns are split-and-dividend-adjusted closes over the following one and three years; windows that ran past October 2026 were excluded rather than extrapolated. The benchmark is the unconditional median: the same entry rule applied to every company-year in the panel, forty-five observations, the number against which "better" must be measured.

The methodological caveat is the same one that now accompanies every study in this series, and it cuts in the hypothesis's favor: the annual estimated scores were constructed with knowledge of outcomes. A researcher who knows how the story ended will tend to place the crossing exactly where the story turned. If a claim fails even with that tailwind — with the crossings arguably optimized to look like signals — the failure is more robust, not less.

Seven crossings, one median

The median one-year return after a 65-line crossing was −5.9%. The unconditional median of the panel was +35.9%. The pre-registered rule therefore says REJECT: crossings did not beat doing nothing; they lost to it by a wide margin. Three of the seven beat the S&P 500 over the year (MELI 2019, PLTR, ANET). The average tells the familiar misleading story — +66.1%, dragged upward by Palantir's +345.7% — which is why this series reports medians.

The two companion lines add almost nothing, and reporting that is itself the honest result. The 50-line analysis found zero crossings: all fourteen panel companies were at or above 50 for the entire period — a selection bias built into a panel of prominent companies, and a reason the next study (Track 2, 150+ observations) needs obscure failures, not just famous survivors. The 80-line analysis found two events, both NVIDIA (2020: +69.8% in one year; 2023: +71.2%). Two observations from one company is anecdote, not evidence. The quarterly panel added two more 65-crossings (MDB in 2026Q2, Palantir in 2025Q3), but with no forward history to measure, they are recorded only as counts.

The temptation of +3y — and why we withheld it

Here is where a less disciplined study would flip its conclusion. Over three years, the crossings' median was +107.6% against +56.4% unconditionally — the direction the hypothesis predicted. Four events, a big spread, the headline writes itself. We are not writing that headline, because every one of those four entries is tangled with the market cycle in a way that makes the crossing signal inseparable from the entry multiple. Two of the four (MELI 2019, NOW 2021) entered into the teeth of the COVID collapse and the 2022 bear market — periods when almost everything was cheap, and being cheap is what did the work. The other two entered a 2018 drawdown. When your "signal" fires exactly where the whole market was on sale, you cannot tell whether the signal or the sale made the money.

This is precisely the pattern the sister studies found: the profitability-inflection study's winners were priced at the trough; the slope study's negative-slope winners were 2022–23 trough entries. The +3y crossing data is therefore not a contradiction of the rejection — it is the same finding wearing different clothes. We have recorded the +3y numbers as an open item for retesting on the larger Track 2 panel, where a real test can separate the two explanations. The pre-registered verdict stands: rejected.\n

One sentence for the whole study: crossing 65 is a signal that the business has cleared the bar, not a signal that the stock is now cheap — and in the following year, the market's multiple at the crossing decided more than the crossing itself.

The trilogy closes: three honest no's

This is the third entry in a pre-registered series. Buying the first profitable year: rejected (median −8.3% vs buy-and-hold −1.2% over three years). Score momentum: rejected (ρ = −0.091). Threshold crossing: rejected (−5.9% vs +35.9% over one year). All three asked, in different ways, whether the Blitz Index can time entries. All three answered no, and the no's agree on the mechanism: the index measures business quality, and what decided subsequent returns in every study was the price paid — the multiple at entry, the cycle at entry, the expectations already in the price at entry.

None of this revises what the index does claim. The same body of research that rejects entry timing keeps confirming failure detection: the D-zone caught the deep collapses, WeWork scored 29/D in a 2019 blind test, the ARK panels' D-grades avoided the blowups, and this week's gate reviews refused to score Joby and Oklo for lack of quantified contracted demand. That asymmetry is now the position: the Blitz Index is an instrument for measuring businesses, not for timing them.

The next article closes the trilogy with the full synthesis — and then the program moves to H4, the one signal that survived: the valuation-zone finding, awaiting retest on the 150-observation panel. The honest negatives stay published regardless. That is the contract this research section makes with its readers.

BlitzLab scores businesses against a scholar's standard — not investment advice. This article reports descriptive statistics of past patterns under a pre-registered methodology (H1, registered 2026-10-13, verdict: rejected). Past patterns are not predictions. No price targets, no buy/sell recommendations. Figures: GAAP, SEC-filed; estimates labeled as estimates.