FEE ESTIMATION[ OVERVIEW ]

Bitcoin transactions compete for limited block space by offering fees. This page shows ML-predicted optimal fee rates compared to Bitcoin Core's built-in estimator, using live mempool data.

BITCOIN FEE PREDICTIONLIVE_DATA
FEE ESTIMATION ENGINE[ LIVE ]
NEXT_BLOCK (1)
--- sat/vB
30_MIN (3)
--- sat/vB
1_HOUR (6)
--- sat/vB
ECONOMY (12)
--- sat/vB
LIVE FEE RECOMMENDATIONSMEMPOOL.SPACE
ML vs BITCOIN CORE[ COMPARISON ]
Target ML Estimate Bitcoin Core Difference
ML vs CORE BENCHMARKCOMPARISON
MEMPOOL STATE[ MONITORING ]
LIVE MEMPOOL STATEMEMPOOL.SPACE API
MODEL CARD[ LOADED ]
fee-lgbm

LightGBM fee estimator predicting optimal sat/vB for 1-block confirmation. Uses live mempool state features.

Last trained: 2026-03-01 06:09:36 UTC

Training Data

DATASET
Live mempool snapshots (mempool.space)
N FEATURES
13
N SAMPLES
110

Training Config

model: LGBMRegressor | objective: regression_l1 (MAE) | n_estimators: 300 | max_depth: 8

Metrics

Metric Value
MAE 0.0909 sat/vB
RMSE 0.2941 sat/vB
MAPE 0.0898
Median AE 0.0000 sat/vB

Feature Importance

[FEE-LGBM] REGISTRYR2_BACKED
TRAINING PIPELINE[ PIPELINE ]
COLLECT_MEMPOOL->FEATURIZE->TRAIN_LGBM->SAVE_TO_R2
[FEE-LGBM] PREFECTORCHESTRATION