Calibrated crypto market-structure research, built on public exchange data — turning leverage and funding noise into one honest crowding rank, shown with its uncertainty.
Why this exists
Crypto is drowning in tools that sell confidence: AI that “predicts” price, social-hype trackers, scanners that score five thousand random tokens. Almost none can show that their signal survives an honest out-of-sample test. The gap isn't data — it's discipline.
Ioxer is built the opposite way. It starts from one signal that earned its place — funding crowding in liquid perpetual markets — read contrarian: when the crowd piles onto one side and pays to stay there, forward returns tend to disappoint. Every candidate factor must beat out-of-sample validation, multiple-testing control, survivorship checks, and a tradability gate before it joins the score.
The thesis: a calibrated read on market structure — is this move clean, crowded, or vulnerable to reversal? — is more useful, and more honest, than another prediction. We show the probability, its uncertainty, and the factors that failed. You decide.
Who built it
Ioxer is the work of a single independent engineer - Illya Kuznetsov - self-funded through the calibration phase. Built solo Feb–May 2026: every factor scorer implemented from the primary paper, every cron, every page, every test. The moat is the engineering, not pedigree.
No VC capital has been accepted. This is deliberate - the only way to maintain the honest pricing and honest timeline described on the pricing page is to not have external runway pressure.
Independent does not mean unreviewed. Every commit runs through type-checking and automated CI gates, every factor that ships is tested out-of-sample and documented on the methodology page, and external math review is planned before the paid Pro tier launches. Until then we publish what works and what doesn't - including the live calibration progress and the module-by-module status page.
Timeline
Early 2026
Built as a quant research engine
The calibration + uncertainty stack — out-of-sample validation, multiple-testing (FDR) control, conformal prediction intervals, survivorship-aware testing — built from the primary literature.
May 2026
Rebuilt crypto-native
Repointed at crypto market structure: perpetual funding, prices, and the liquid-coin universe (~50 coins). One factor survived the honest gauntlet — funding crowding; others (momentum, low-vol, on-chain growth, DeFi TVL) were tested and dropped.
●29 May 2026
Live forward tracking opened
Every daily IOX score is now recorded point-in-time, before the outcome is known. 1d, 7d and 30d horizons live — the real proof accumulates from here.
30 Jun 2026
30-day horizon matured
First 30-day forward outcomes recorded — one more cohort completes every day, so calibration and hit-rate by IOX bucket become measurable on the medium horizon.
Projected
~27 Aug 2026
90-day horizon matures
The slowest, most meaningful horizon completes — the first real 90-day forward outcomes. Early signal, not yet statistical proof (a small edge takes far longer to prove).
Late 2026
Pro tier + bulk/Enterprise API
Alerts, historical depth, and more factors as they pass the gauntlet — plus higher-rate bulk / webhook API access for funds and quant teams. (The public read-only IOX feed is already live now at /docs.)
Principles
Honest before impressive
If a number isn't measured, we say so. The interval width is published alongside the score. The calibration state is displayed on every coin. We'd rather under-promise on track record and deliver than the opposite.
Composition, not opacity
Every factor that ships is tested out-of-sample and documented on the methodology page — including the ones that failed. The moat is the discipline and how the layers compose, not secret ingredients.
Free as a weight on the industry
Good quantitative research gated behind institutional paywalls is a market inefficiency. Making it free - and well-engineered enough to actually use - is the bet.
Small team by design
One engineer until revenue justifies more. Building with one hand tied makes you ruthless about what doesn't matter. The whole stack delivers serious research without a big budget - and it works.