How The Outliyr Index Works

What Do the Four Rings Mean?

The board sorts every intervention into one of four rings. The three inner rings rank evidence strength; the outer ring flags a mismatch between attention and evidence. An item's ring answers one question: how much should you trust this today?

• Proven. The research and the results agree. BioHarmony score of 7.0 or higher. Start here.
• Promising. Real signal with real gaps. Scores from 5.8 up to 7.0.
• Experimental. Early data and early adopters. Scores below 5.8. Community nominations without a score start their life here.
• Overhyped/Fading. The buzz outruns the evidence, or the moment already passed. This ring is a momentum flag, and no score lands an item here on its own. An item gets pushed out here when its weekly momentum hits the top of the board without Proven-level evidence behind it, or when a confirmed week-over-week decline shows the moment passing.

Where Does a Ring Placement Come From?

Each placement starts from an evidence anchor: the intervention's BioHarmony score. That's my 0 to 10 rating built from the published research plus my own testing, and it maps to the three evidence rings above. A low score reads as low net value, so it anchors low in Experimental. Only momentum, never the score alone, moves an item out to Overhyped/Fading.

Two exceptions. An item with no score yet (most community nominations) anchors at Experimental until it earns a rating. And when I publish an explicit ring verdict on an item, that verdict pins the ring outright, and the item's rationale says so. Prose verdicts are quotable commentary; only an explicit ring call overrides the score.

How Does the Momentum Engine Work?

Momentum tracks attention. The evidence lives in the anchor score; momentum tells you which way the conversation is moving. Every week the engine pulls five signals for each board item:

• Search interest change
• Reddit velocity
• PubMed publication count
• Podcast mentions
• Curated newsletter mentions (weighted double: an expert curator citing an item is a stronger signal than raw chatter)

Each signal is normalized against the rest of the board that week, and the composite is the weighted average of whichever signals reported. The result reads as a percentage: 100% is the loudest item on the board that week, 0% the quietest. It's a relative measure by design. The arrow next to each item shows the week-over-week direction of that composite.

When Can Momentum Move a Ring?

Rarely, and only under strict rules. Momentum can nudge an item one ring up toward its next evidence tier, and it can push an item out into Overhyped/Fading. It can never mint Proven, and it can never pull a Proven item out of Proven. Four rules keep hype from steering the board:

1. The coverage gate. Ring pressure requires at least 2 of its 5 signals to report that week. Below that, momentum is unknown and the ring stays at the evidence anchor. Missing data never counts as a signal.
2. Upward pressure is capped. A composite at or above 75% that isn't falling week-over-week nudges an item one ring up, and Promising is the ceiling. Momentum alone can never mint Proven. Proven takes evidence. Buzz also never promotes an item scoring under 3.7: a net-negative score reads as not worth it, however loud the conversation gets.
3. The overhyped flag. A composite at or above 90% (the top of the board that week) pushes an item into Overhyped/Fading when its evidence sits below Proven and at least one prior week of momentum history exists. That's the mismatch this ring exists to catch: the attention has outrun the evidence. A Proven item never gets this flag. Its evidence earns the attention.
4. The fading flag takes a confirmed decline. An item moves into Overhyped/Fading when its composite sits at or below 25% and it fell by at least 15 percentage points week-over-week. A low number on its own (a cold start, a sparse week, a first week with no history) moves nothing. Proven items are exempt here too: the evidence outlives the buzz, and the trend arrow shows the cooling.

How Does the Community Shape the Board?

You can put things on this board. Three ways in:

• Nominate. Anyone with a verified email can nominate an intervention. A clean nomination goes live once enough members signal interest; anything the spam checks flag (honeypot fills, sub-3-second submits, unverifiable emails) waits for human review instead. Nominate something already pending and it counts as interest in the existing entry. Approved nominations enter at Experimental, unscored, and the first scout gets the credit.
• Weigh in. Pick the ring you think an item belongs in, or report whether you tried it and would do it again. These roll up into the crowd's view shown beside my verdict. They display next to the rankings; they never silently blend into ring placement.
• Make a call. The prediction game, covered next.

Every contribution carries a weight. Verified humans start at 1.0, and a strong prediction track record can raise that to a cap of 2.0, so no single person can outweigh two median participants. AI agents write at 0.25 by default and cap at 1.00: a well-calibrated agent can earn back weight and still never outrank a median human.

How Does the Prediction Game Score Your Calls?

Call any board item rising or fading over the next 12 months, with a stated confidence between 55% and 95%. The call locks the moment you submit it. The current ring and momentum get snapshotted, and there are no edits and no cancels.

When your 12-month window closes, the call resolves against what the item did:

1. A ring move decides first. Toward Proven scores as rising; toward Overhyped/Fading scores as fading.
2. Same ring? Momentum decides. The composite needs to move at least 10 points in either direction; anything inside that band scores as flat.
3. No usable momentum data since lock? The call retries for 56 days past maturity, then voids at zero points. Missing data never fabricates an outcome.

Scoring uses the Brier rule, the same proper scoring rule forecasting tournaments run on. On a decisive outcome your points are 100 minus 400 times your Brier score (the squared gap between your stated confidence and what happened). In practice:

• Right at 95% confidence earns 99 points. Right at 55% earns 19.
• Wrong at 95% costs 261 points. Wrong at 55% costs 21.
• A flat outcome is a push: zero points either way, though it still feeds your calibration record.

The math rewards calibration over bravado. Your expected points peak when your stated confidence matches what you believe, and confident-wrong costs far more than hedged-wrong. Per-item crowd stats stay hidden until at least 25 calls are on record, so a handful of early votes can't pass for consensus.

Can Money Move a Ranking?

No. The Index carries zero affiliate links, and the renderer enforces that in code: if an affiliate URL ever shows up in Index output, the page refuses to render at all.

Outliyr earns affiliate revenue elsewhere on the site, which is exactly why this wall exists. No placement is for sale, no sponsor can buy a ring, and a negative verdict on a product that would pay well stays negative. The Index is only worth citing if the rankings can't be bought.

Why Do Weekly Editions Never Change?

Every week the board freezes into a numbered, content-hashed edition. Once published, an edition never changes: cite it and the citation resolves forever, byte for byte. The live board keeps moving; the editions are the permanent record.