Pre-launch· Token not live yet — no rewards or treasury until launchWhat this means
← Model
Adds global fees paid — retrained on new data

TRAIN v0.24

Adds global fees paid : total fees and Jito tips, fees per minute, fees relative to volume, transactions per paying wallet and failed-transaction share — then lets automatic discovery choose which change-signals to add on top. Retrained on the newest data.
Completed

Definition

Model type
Decision-tree ensemble
Signals used
… + global fees
Technical details
max_depth
3
subsample
0.8
n_estimators
300
learning_rate
0.03
min_samples_leaf
10
Random seed
1024
Config hash
1a7b9d0acbc2c587
Created
05 Oct 22:41
Completed
05 Oct 22:47
Artifact
models/v0.24/model.json
sha256 eda3dd609e804337…

Dataset & compute

Dataset
Solana Launch Dataset v16
8,939 snapshots · 1,794 launches · sha256 07b5c66cb3c8
Training runs
1
Compute
TRAIN training server (CPU)
GPU hours
—
no GPU used
CPU time
1,432.7 s
Provider cost
$0.00
nothing purchased

Benchmarks

held-out test set · change vs v0.23
Spotting crashes (−70% within 1h)
0.944
Spotting survivors (still traded after 1h)
0.976
Spotting 2× runs (within 1h)
0.848
Calling the 1h direction (down / flat / up)
56.7%
Spotting creator dumps (within 1h)
0.882
Spotting crashes (within 6h)
insufficient data
Spotting survivors (6h)
insufficient data
Calling the 6h direction
insufficient data
Spotting survivors (24h)
insufficient data
Spotting migrations (within 24h)
insufficient data

Validation metrics (during training)

collapse_1h
AUC 0.992
n=1064
collapse_6h
AUC 0.500
n=4
reach_2x_1h
AUC 0.908
n=1064
survival_1h
AUC 0.940
n=1064
survival_6h
—
n=4
survival_24h
—
n=0
migration_24h
—
n=0
trajectory_1h
acc 98.1%
n=1064
trajectory_6h
acc 25.0%
n=4
creator_exit_1h
AUC 0.844
n=245

Validation launches are separate from training launches and from the locked test set. Benchmarks above are the public numbers.

Signals this version discovered

45 candidate signals tested on held-out folds of the training launches. Cross-validated AUC 0.901 → 0.905.

What this version relies on

held-out validation launches
Collapse risk · 1h
Volatility, last 5 minutes−0.014
Net SOL into the curve−0.012
Trades per wallet−0.008
Trades, last 5 minutes−0.003
fee per min−0.001
Holder growth, last 5 minutes−0.001
Survival · 1h
Net SOL into the curve−0.179
Buy share of volume since launch−0.019
Share of tiny (<$3) trades−0.009
Top-5 holders' share−0.007
txs per payer−0.004
Creator's current holding−0.004
Reaches 2× · 1h
Price change, last minute−0.036
Holders (from trades)−0.017
Holder growth, last 5 minutes−0.017
Net SOL into the curve−0.016
Number of trades−0.008
Volume, last minute−0.003
Creator exit · 1h
Creator's current holding−0.096
Net SOL into the curve−0.071
Age of the launch−0.011
Creator's prior launches−0.011
Buyers seen in other launches−0.009
Volume, last 5 minutes−0.006

For each signal, how much the score drops when that signal is scrambled across launches. TRAIN learns these weights itself from the data; nobody hand-codes them.

Training runs

Attempt 1Succeeded
05 Oct 22:41 · 00:06:25 · 7,875 examples · $0.00

Timeline

  1. 05 Oct 22:47
    TRAIN v0.24 not deployed; v0.19 stays live
    Reason: average score across 5 tests 0.843 vs 0.845 for the live v0.19.
  2. 05 Oct 22:47
    Benchmark completed: TRAIN v0.24
    Crash-spotting score regressed 0.955 → 0.944 vs live v0.19.
  3. 05 Oct 22:47
    Benchmark started: TRAIN v0.24
  4. 05 Oct 22:47
    TRAIN v0.24 discovered 1 useful new signal
    Change between snapshots: share of tiny (<$3) trades. Score on held-out training launches 0.901 → 0.905.
  5. 05 Oct 22:47
    Training completed: TRAIN v0.24
    00:06:25 · trained on 7875 examples · $0 compute cost
  6. 05 Oct 22:41
    Training started: TRAIN v0.24
    Solana Launch Dataset v16 · decision-tree ensemble · TRAIN training server
  7. 05 Oct 22:41
    TRAIN v0.24 queued
    Adds global fees paid : total fees and Jito tips, fees per minute, fees relative to volume, transactions per paying wallet and failed-transaction share — then lets automatic discovery choose which change-signals to add on top. Retrained on the newest data.