Editor’s note: This article has been substantially revised with additional testing and official election-process information.
A new mathematical analysis of the 2026 midterms has uncovered a pattern that election officials, data analysts, and voters should not casually wave away.
At the center of it is one number: 0.85.
The X account Truth Is Freedom found that Republican candidates repeatedly finished near 85% of President Donald Trump’s 2024 vote share. The account calls it the “15% Band” and argues that the consistency raises serious questions about the reported results.
Those questions deserve answers backed by data, audits, and ballot records—not slogans from either side.
The published math does not, by itself, prove that votes were switched or that any person, party, or voting system committed fraud. The pattern still deserves daylight.
It is a reason to put the full dataset under a microscope.
The original analysis compared 59 Senate and governor races with President Trump’s 2024 performance in the same states.
Its central formula was:
Republican 2026 share ≈ 0.85 × Trump’s 2024 share.
The account also expressed the Democratic side as Harris’s 2024 share plus 15% of Trump’s share.
The headline-grabbing formula is real. But there is also a mathematical catch that must be confronted honestly: the two equations are not independent findings.
In a two-party calculation, Harris’s share is the remainder after Trump’s share. Once the Republican result is defined as 0.85 times Trump’s share, the Democratic result is automatically the complement.
The second equation therefore cannot independently establish that votes moved from one side to the other. Anyone presenting it as separate confirmation is claiming more than the algebra supports.
The fixed starting point is legitimate. The Federal Election Commission’s official 2024 results compile the certified state totals that awarded President Trump 312 electoral votes and Kamala Harris 226.
The FEC report identifies state election offices as its source and preserves each state’s popular-vote totals, electoral votes, and candidate breakdown. That makes it a sound baseline for checking what the viral analysis actually calculated.
What those records cannot show is how an individual 2024 Trump voter behaved in 2026. Presidential and midterm electorates are different, turnout changes, voters skip lower-profile contests, candidates attract crossover support, and national political conditions can move many states at once.
Using the certified 2024 denominator is necessary, but it cannot solve those comparability problems. A valid causal claim still has to distinguish changed participation and voter preference from an altered count.
That means the 85/15 relationship describes the reported results. It does not yet identify the mechanism that produced them—and that unanswered question is exactly where serious scrutiny should begin.
An independent state-by-state reconstruction published by Lead Stories found that part of the pattern is real: a cluster of heavily Republican states landed close to the 0.85 estimate.
The reconstruction used reported 2024 presidential and 2026 Senate results, added Alaska, Arkansas, and Montana, and published its detailed comparison spreadsheet. That allows readers to inspect every state-level prediction directly rather than relying on the visual impression created by the original chart.
Here is where the viral claim meets a harder test: the same reconstruction found substantial misses in both directions. Across 32 Senate states, only eight were within one percentage point of the formula.
The deviations ranged from 6.5 points below the prediction in Rhode Island to 8.1 points above it in Maine.
The presentation also matters. The original chart used a compressed 20%-to-80% vertical scale across a very wide frame, making differences of several percentage points appear smaller than they are when the races are listed state by state.
Maine is especially important. Its 2026 Republican Senate candidate reportedly exceeded Trump’s own 2024 share, something a universal 15% reduction rule cannot produce consistently.
That variation does not erase the national Republican decline or make the broader pattern irrelevant. It does destroy the easy leap from “an average near 0.85” to “one fixed algorithm operated everywhere.”
One critic summarized the core problem directly: comparing aggregate vote shares across two elections does not track the movement of identified voters or ballots.
Truth Is Freedom later expanded the analysis to 692 counties across six states where Republicans lost Senate races.
The account says a model using the 0.85 baseline, county Republican strength, and Hispanic population placed roughly 97% of those counties within five percentage points of its prediction.
If that 97% figure survives independent replication, it would demand a serious explanation. Right now, however, it cannot be independently verified from the public post alone.
The account has not released a complete reproducible package containing every input, the full county table, inclusion and exclusion rules, exact coefficients, code, revision history, and comparisons with ordinary election-swing models.
The five-point tolerance also matters. A model can classify many observations as “within range” when the permitted range is broad, especially after variables are selected with knowledge of the same data being explained.
A proper test needs a model fixed in advance and evaluated against data it was not built to fit. Otherwise, critics can reasonably argue that the model was tuned around the very results it purports to explain.
Truth Is Freedom has stood by the analysis and argues that critics are attacking the detector instead of explaining why it registered a signal.
Truth Is Freedom is right about one thing: sneering at the detector is not an answer.
Release the data. Test the signal.
Show where it holds, where it fails, and why.
As of November 18, several cited 2026 totals were still unofficial. The Ohio Secretary of State says bipartisan county boards must review outstanding ballots, resolve discrepancies, conduct the official canvass, and certify no later than 21 days after Election Day.
Ohio also requires post-election audits that hand-count selected paper ballots and compare them with electronically tabulated totals. Those records can move this debate beyond internet charts and into evidence that can actually confirm or rule out counting problems.
The state’s published process separates election-night reporting from the official canvass. Eligible provisional ballots and qualifying military and overseas ballots are added before certification, while county records are reviewed for discrepancies.
Ohio also sets an automatic statewide recount threshold of one-quarter of one percent. Together, the canvass, paper-ballot audit, and recount rules create specific evidence that can either reveal a tally problem or show that a viral statistical suspicion did not correspond to an actual counting error.
The Minnesota Secretary of State scheduled its state canvass for November 19. The state identified several contests within publicly funded recount margins, including a U.S. House race separated by 323 votes and a state Senate race separated by four.
Minnesota’s state canvassing board certifies state and federal results reported by local officials. The recount-eligible margins show why a preliminary election-night snapshot cannot be treated as the final dataset for a national mathematical claim.
The four-vote state Senate margin was just 0.04%, while the 323-vote U.S. House margin was approximately 0.1%. Those are concrete, contest-specific reasons for scrutiny, and Minnesota’s public process identifies the races and the legal mechanism for checking them.
This is where officials should welcome scrutiny instead of treating every question as an accusation. A recount can compare actual ballots with a reported result; a cross-election regression cannot do that by itself.
The Oregon Secretary of State says its results remain unofficial until December 10. Valid postmarked ballots could arrive through November 10, voters have through November 24 to resolve ballot challenges, and every county must complete a post-election audit before results become final.
Oregon publishes continuing ballot-return and unofficial-result updates while counties process eligible ballots. Its official page explicitly warns that those figures remain fluid until certification.
That warning is directly relevant to the 85/15 debate. A model calculated from an early snapshot can move as valid ballots are added, challenges are resolved, and county audits are completed.
Researchers therefore need to retain the exact timestamp and source file for every calculation, then rerun the model on Oregon’s certified results. Without that versioned trail, later claims that a prediction “fit” cannot be reproduced or fairly evaluated.
Those processes can change margins and model residuals. Any honest analysis must identify the precise data snapshot it used and rerun the calculations against certified totals.
The strongest conclusion the evidence supports today is that Republicans suffered a broad, measurable decline from Trump’s 2024 performance, and an average near 85% describes part of it.
Why did that pattern appear so broadly? How much came from turnout, candidate quality, national swing, demographics, or something else?
Those are major questions. The available evidence stops short of establishing a universal algorithm or proving that votes were transferred.
That evidentiary limit cannot become an excuse to shut down the inquiry.
Publish the full dataset. Publish the code.
Predeclare the model before testing it.
Run it against certified results and genuine holdout data. Force it to compete with ordinary national-swing, turnout, candidate-quality, and demographic models.
Then examine the records that could actually demonstrate manipulation: ballot reconciliations, voter files, chain-of-custody documents, machine logs, hand-count audits, and specific discrepancies tied to a county and contest.
Election integrity is too important for “trust us.” It is also too important to turn a statistical correlation into a solved crime before ballots, audits, and records establish one.
Demand the evidence. Demand the underlying data.
Demand that election officials answer legitimate questions with verifiable records.
Then follow the facts wherever they lead.
The post The 85/15 Pattern: New Mathematical Analysis Raises Major Questions About the 2026 Election Results. appeared first on 100PercentFedUp.com.
Continue reading...
[H/T 100PercentFedUp]