Ground Truth

How we grade

Two grades on every compound, never folded into one number: how far human testing has got for the use people take it for, and how safe it is. Rubric last checked Oct 28, 2026.

The rubric

Every use is graded on its own. The headline grade on a compound page is the grade for the use people actually take it for. If a different use has stronger evidence, that shows as one plain line underneath.

Evidence, A to D, for one use

Failed trials still count

A randomized trial counts toward the letter whether it worked or not, because the letter says how far testing got. The result shows separately as a tag: worked, mixed, did not work, or not tested in humans.

Geography, lack of replication, preprint status and surrogate endpoints are limitations we state in the text. They are not extra thresholds on the letter.

Approval is its own line

Approval is its own line: Approved for (indication, agency, year), and a note that it does not carry over to the popular use. An approval raises the grade of that one use to A and nothing else.

Example: SS-31 is approved for Barth syndrome, a rare disease (FDA accelerated approval, 2025). People take it for energy, recovery and longevity, which have never been tested in healthy adults. So the headline is D for energy, with the approval on a separate line that says it does not carry over.

Regulatory news never moves a grade

Regulatory news never moves a grade. A ban, a compounding decision, a warning letter or a committee vote is shown as a status, not as evidence.

Safety, A to F

Safety is graded separately from evidence, from six inputs:

Doses

Every dose on the site sits in a table with its source on the same row: the drug label, a published trial, a named episode with a timestamp, or a named clinician. No seller, vendor, compounding pharmacy or calculator site is ever a dose source, and a clinician who sells the product does not count. When we find no eligible statement, the row says so. We never average doses or call the most repeated number correct.

How a grade gets made

First we ask who profits if you believe a claim. A conflict is not an automatic disqualification, but you are owed the context, so we flag it on the experts and sources we cite, and on ourselves.

For most of these compounds the human trials barely exist, so we also read the real-world record: athletes and bodybuilders who ran them years before any study, and clinicians who have used them with patients. That record sits in its own section on each compound page, always labeled as anecdote. It never raises a letter.

Then we ask what else is going on. Almost nobody runs a peptide alone. On testosterone, a GLP-1, a cut or a new training block, much of the result is the proven thing doing the work. Every compound page says what else could explain the result.

Last, we try to prove the grade wrong: we look for the trial that missed its endpoint, the retraction, the replication that never came, and the money behind a study. If the grade survives, it ships. Grades are dated, and they change when the evidence does. Tell us if you think one is wrong.

Is it placebo?

We do not put a probability on placebo. We say what the trials show: either no controlled trial for this use exists, so a placebo effect cannot be ruled out, or trials for this use beat placebo on a named endpoint.

Placebo is real and works through the brain's own chemistry. It moves how you feel (pain, fatigue, energy, mood) far more than anything a lab can measure, and it still works when people are told the pill is inert. So "I can feel it working" is not evidence on its own.

If your goal is a feeling and you get it, that is real to you, though you may be paying peptide prices for it. If your goal is a number (visceral fat, a lab value, a tendon that holds), placebo cannot fake it, and only measuring tells you. The way we test it ourselves:

  1. Write down what you expect to change, specifically.
  2. Change one thing at a time.
  3. Measure a baseline before you start.
  4. Track a number, not a feeling.
  5. Blind yourself where you can.
  6. Run it on, off, then on again.

Trial Mode in Ground Control is built on that structure.

Is the vial what it says?

A grade is about the molecule. What is in your vial is a separate question, and only a lab answers it. The five tests that matter:

Reading a certificate of analysis: match the lot number to your vial, check the date, confirm identity, purity, endotoxin and net peptide content (the actual milligrams, often less than the label), and make sure the lab result can be checked independently. A lab picked and paid by the seller is marketing, not verification. Independent labs such as Janoshik and Finnrick publish batch results.

The bacteriostatic water you mix with is a contamination risk too. Use pharmaceutical-grade water, keep the mixed vial cold, and replace it about every 28 days, when the preservative stops being reliable.

How little anyone knows

Your body runs thousands of peptide signals at once, most of them unmapped. When a compound page calls a mechanism assumed, that is the accurate word: we have a story for how it probably works, usually from cells or animals, and rarely know what it does in a person who is also training, sleeping badly and eating differently this month.

So when you feel a change, you are reading one result out of many moving parts and crediting the one you can name. Sometimes that is right. Often it is the training, the diet, the testosterone or the season. That is why we push measuring over guessing.

Why so few are tested

The evidence behind most peptides is thin mainly because nobody can make money running the test. A Phase 3 trial costs tens of millions of dollars and most drugs that enter human testing never reach approval. A company only takes that bet if a patent lets it earn the money back.

Most popular research peptides fail at the patent step. BPC-157 is a fragment of a protein in human gastric juice, MOTS-c is made by your own mitochondria, TB-500 is a piece of a natural protein. Long-published natural sequences are close to impossible to patent as new compounds, so no sponsor funds the trial. The peptides that did get large trials and approval, tesamorelin and the GLP-1 drugs among them, are owned, branded products. The evidence tracks the patent, not the biology.

Meanwhile compounding pharmacies, telehealth platforms, cash-pay clinics, gray-market sellers and affiliates all profit from these compounds in their untested state, and none has a reason to fund a trial that might say they do nothing.

This is a blank, not a cover-up. There is no buried positive trial; "no good human data" means we do not know. The same economics block safety trials and adverse-event tracking, so "few reported harms" is weak reassurance in the same way "few reported benefits" is weak condemnation. FDA has named real concerns (immunogenicity, impurities, thin safety data, compounded lots that differ from commercial ones), and several of those are manufacturing problems as much as molecule problems.

Sources for this section include the Tufts Center for the Study of Drug Development cost work, the HHS ASPE report on clinical trial costs, Wong and colleagues on trial success rates (Biostatistics, 2019), and FDA Category 2 and advisory committee materials.

Not medical advice. The grades describe evidence, not what you should do. Talk to a licensed clinician before you act on anything here.