WEIRSolana Mainnet

WEIR has been at this window since day one, counting every Solana token that clears $10K. The gate only opens when the evidence is provably good enough.

$WEIRthe ladder · model validation & weights
01Continuous learning progress

Active model weights & validation

Real-time aggregate data collected across Solana. Machine learning weights update continuously as token market caps cross established evaluation thresholds.

ONLINE · SOLANA

ML Validation Engine

5-Fold Stratified · d=37Target Peak MC ≥ $30K
AGENT #0342GATE 13.5%
Kip the Salmon (WEIR Agent #0342)
Total Tokens CountedLIVE SCAN
269
Entry Threshold:> $10K Peak MC
Class Distribution35.3% Clear Rate
95cleared $30K0 held
Model Accuracy & VC FloorROC-AUC
0.5480Measured
VC Bound Floor (ε=0.723):0.513
Feature Dimension Vectord = 37
37signal parameters
Evaluation Window:Launch Hour 0 – 24

Active Data Streams & Ingestion Pipeline

STREAMING

Continuous ingestion monitoring new Solana token mints, decentralized exchange pairs, and holder accounts.

Entered the river (peak ≥ $10K)
Solana ingest worker
269
LIVE
Cleared the falls (peak ≥ $30K)
Positive class
95
LIVE
Held below (age ≥ 48h)
Negative class
0
LIVE

Model Validation & Four Locks Status

Lock 1 · Sample Volume (N ≥ 2,000)
269 / 2,000● HOLDING
Lock 2 · Cleared Target Class (N_pos ≥ 200)
95 / 200● HOLDING
Lock 3 · Variance Bound (σ_AUC < 0.05)
σ = 0.0215● HOLDING
Lock 4 · Out-of-Sample Time Split (gap ≤ 0.04)
NOT YET● HOLDING

Learned Pattern Weights (Feature Importance)

37 FEATURES TRACKED

The relative impact of extracted data signals calculated by WEIR's training algorithm to distinguish tokens that cleared $30K from those held below.

No model run yet. Feature weights appear after the first training run.

Gate Water Level
Proven floor 0.5134 vs Target 0.600
13.5%
HELD BY LOCK 1
02Vapnik–Chervonenkis proof & capacityThe gate opens on proof, not on time.WATER LEVEL: 13.5%
WEIR Water Gauge
Drag to turn
Water Level
13.5%
Proven Floor
0.500
Target
0.600
Why the water rises slowly

A model can look good by luck on a small sample. Vapnik–Chervonenkis theory puts a number on that luck: with n tokens and a model of capacity d, the true score can sit below the measured one by at most ε, with 95% confidence.

ε=37(ln⁡2⋅34037+1)+ln⁡80340=0.662\varepsilon = \sqrt{\frac{37\left(\ln\frac{2\cdot 340}{37}+1\right)+\ln 80}{340}} = \mathbf{0.662}
AUC‾−ε=0.548−0.662=−0.114\overline{\mathrm{AUC}} - \varepsilon = 0.548 - 0.662 = -0.114

WEIR lets water rise with AUC − ε, never with the raw score. Good model, thin sample: the water stays low. That is the point.

measured AUC
0.548
penalty ε
0.662
proven floor
0.500
340
0.548
37
Measured 0.548, but ε takes 0.662. Floor is 0.500.
03The math on this page

Generalization theory & Four Locks

All equations →
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.

Bootstrap bound(3.2)
AUC‾ boot  =  Q0.025({AUC(Db∗)}b=12000)\underline{\mathrm{AUC}}_{\,\mathrm{boot}} \;=\; Q_{0.025}\Big(\big\{\mathrm{AUC}(\mathcal{D}^{*}_{b})\big\}_{b=1}^{2000}\Big)

Resample the data two thousand times; keep the score that 97.5% of worlds beat.

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.

Four Locks(4.1)
L  =  ⋀k=14ℓk,ℓ1:n≥2000ℓ3:σAUC<0.05ℓ2:n+≥200ℓ4:∣AUCcv−AUCtime∣≤0.04L \;=\; \bigwedge_{k=1}^{4} \ell_k, \qquad \begin{aligned} \ell_1 &: n \ge 2000 & \ell_3 &: \sigma_{\mathrm{AUC}} < 0.05 \\ \ell_2 &: n_{+} \ge 200 & \ell_4 &: \big|\mathrm{AUC}_{\mathrm{cv}} - \mathrm{AUC}_{\mathrm{time}}\big| \le 0.04 \end{aligned}

Enough data, enough survivors, stable folds, no look-ahead.