Epoch Labs Logo
Epoch LabsRobinhood Mainnet
Epoch Labs has been at this desk since day one, reading every Robinhood Chain token that clears $10K. The hourglass only turns when the evidence is provably good enough.
$EPCabout · quantitative research & methodology

What Epoch Labs Is Not

  • Epoch Labs does not predict price. It estimates the conditional probability that a token which already reached $10K peak market cap will go on to reach $30K.
  • Epoch Labs does not have an edge that guarantees financial returns. Four feature families cannot forecast an entire market.
  • The hourglass is not decorative. If the model does not statistically beat random noise under strict capacity bounds, the hourglass stays empty and the system explicitly reports the failing gate.

1. The Concept of Epochs

In machine learning and computational mathematics, an epoch defines a complete cycle through an evolving universe of data. Epoch Labs was founded on the principle that decentralized markets produce continuous streams of behavioral data, but are rarely analyzed with empirical rigor. Rather than relying on intuition or speculative narratives, Epoch Labs evaluates token lifecycles systematically across every market cycle.

2. Why Robinhood Chain

Robinhood Chain operates with continuous liquidity creation and rapid token velocity. Hundreds of micro-assets are minted and traded around the clock, creating a rich distribution of market experiments. While the vast majority stall quickly, a small fraction achieve sustained liquidity. Epoch Labs acts as an autonomous observer, recording every launch, tracking peak market capitalizations, and discovering empirical patterns behind token survival.

3. What Epoch Labs Actually Does

Every token launched on Robinhood Chain that crosses $10,000 peak market cap enters the study population. Epoch Labs observes whether it successfully progresses to $30,000 peak market cap over a 48-hour observation window. Using four orthogonal feature families (launch hour circular encoding, day of the week, 48h holder distribution, and NLP lore semantic embeddings), an ensemble of gradient-boosted decision trees (LightGBM) estimates the survival probability.

4. Why the Hourglass Runs Slowly

In small sample regimes, an ML model can appear artificially predictive purely due to variance and overfitting. To counter this, Epoch Labs enforces a mathematically proven performance floor (proven_floor = min(floor_vc, floor_boot)) combining Vapnik–Chervonenkis (VC) generalization bounds with 2,000-fold Bootstrap resampling. The hourglass level reflects this conservative proven floor rather than raw cross-validation AUC. Epoch Labs will not trigger its own launch until this floor rigorously clears ROC-AUC 0.60 across all validation gates.

5. Open Science & Transparent Research

Anyone can independently inspect the methodology, download the full labeled training dataset, reproduce the AUC score, and verify that the mathematical gates are evaluated honestly.

The math on this page

From empirical observation to a proven statistical floor

All equations→
Survival label(1.1)
yi  =  1{ Pˉi≥30k },Pˉi≥10ky_i \;=\; \mathbf{1}\big\{\, \bar P_i \ge 30\text{k} \,\big\}, \qquad \bar P_i \ge 10\text{k}

P̄ᵢ is the peak market cap. The only question asked; price is never the target.

Capacity penalty(3.1)
ε  =  d(log⁡2nd+1)+log⁡4δn,δ=0.05\varepsilon \;=\; \sqrt{\frac{d\big(\log\frac{2n}{d}+1\big) + \log\frac{4}{\delta}}{n}}, \qquad \delta = 0.05

Vapnik–Chervonenkis: how far luck can lift the score of a model with capacity d.

Proven floor(3.3)
AUC‾  =  min⁡{ AUC‾−ε,  AUC‾ boot}\underline{\mathrm{AUC}} \;=\; \min\big\{\,\overline{\mathrm{AUC}} - \varepsilon,\ \ \underline{\mathrm{AUC}}_{\,\mathrm{boot}}\big\}

The more pessimistic bound is the only number the agent believes.

Live Mainnet Data · Epoch Labs is a research platform, not a financial adviserFour features cannot forecast a market. This measures survival, not price.
@EpochLabsHQ