Pre-launch· Token not live yet — no rewards or treasury until launchWhat this means
← Model
Decision trees — retrained on new data

TRAIN v0.07

Same configuration as v0.04 (best so far on the headline benchmark), retrained on the newest dataset.
Completed

Definition

Model type
Decision-tree ensemble
Signals used
Market signals
Technical details
max_depth
2
subsample
0.8
n_estimators
120
learning_rate
0.05
min_samples_leaf
10
Random seed
1007
Config hash
c27631ee06284446
Created
05 Oct 07:12
Completed
05 Oct 07:12
Artifact
models/v0.07/model.json
sha256 e37acb99eee60f51…

Dataset & compute

Dataset
Solana Launch Dataset v6
2,492 snapshots · 500 launches · sha256 35c7f3706ceb
Training runs
1
Compute
TRAIN training server (CPU)
GPU hours
—
no GPU used
CPU time
6.8 s
Provider cost
$0.00
nothing purchased

Benchmarks

held-out test set · change vs v0.06
Spotting crashes (−70% within 1h)
0.934
Spotting survivors (still traded after 1h)
0.917
Spotting 2× runs (within 1h)
0.808
Calling the 1h direction (down / flat / up)
51.9%
Spotting creator dumps (within 1h)
0.868
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 0.996
n=305
collapse_6h
—
n=0
reach_2x_1h
AUC 0.993
n=305
survival_1h
AUC 0.782
n=305
survival_6h
—
n=0
survival_24h
—
n=0
migration_24h
—
n=0
trajectory_1h
acc 98.7%
n=305
trajectory_6h
—
n=0
creator_exit_1h
AUC 0.736
n=91

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

What this version relies on

held-out validation launches
This version was trained before TRAIN started measuring which signals each model relies on.

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 07:12 · 00:00:03 · 2,187 examples · $0.00

Timeline

  1. 05 Oct 07:12
    TRAIN v0.07 not deployed; v0.04 stays live
    Reason: average score across 2 tests 0.558 vs 0.602 for the live v0.04.
  2. 05 Oct 07:12
    Benchmark completed: TRAIN v0.07
    1h trajectory · balanced accuracy: 0.390 (n=94).
  3. 05 Oct 07:12
    Benchmark started: TRAIN v0.07
  4. 05 Oct 07:12
    Training completed: TRAIN v0.07
    00:00:03 · trained on 2187 examples · $0 compute cost
  5. 05 Oct 07:12
    Training started: TRAIN v0.07
    Solana Launch Dataset v6 · decision-tree ensemble · TRAIN training server
  6. 05 Oct 07:12
    TRAIN v0.07 queued
    Same configuration as v0.04 (best so far on the headline benchmark), retrained on the newest dataset.