bbgo_origin/config/optimizer-hyperparam-search.yaml
Raphanus Lo 67f8b1c32c optimizeex: hyperparameter optimization tool
Currently support the following search algorithms:
- Tree-structured Parzen Estimators (tpe, default)
- Covariance Matrix Adaptation Evolution Strategy (cmaes)
- Quasi-monte carlo sampling based on Sobol sequence (sobol)
- random search (random)

And the following objective function:
- profit
- volume
- equity
2022-07-29 17:09:54 +08:00

45 lines
1.2 KiB
YAML

# usage:
#
# go run ./cmd/bbgo optimizeex --config bollmaker_ethusdt.yaml --optimizer-config optimizer-hyperparam-search.yaml
#
---
# The search algorithm. Supports the following algorithms:
# - tpe: (default) Tree-structured Parzen Estimators
# - cmaes: Covariance Matrix Adaptation Evolution Strategy
# - sobol: Quasi-monte carlo sampling based on Sobol sequence
# - random: random search
# Reference: https://c-bata.medium.com/practical-bayesian-optimization-in-go-using-goptuna-edf97195fcb5
algorithm: tpe
# The objective function to be maximized. Possible options are:
# - profit: by trading profit
# - volume: by trading volume
# - equity: by equity difference
objectiveBy: equity
# Maximum number of search evaluations.
maxEvaluation: 1000
executor:
type: local
local:
maxNumberOfProcesses: 10
matrix:
- type: string # alias: iterate
path: '/exchangeStrategies/0/bollmaker/interval'
values: ["1m", "5m"]
- type: rangeInt
label: window
path: '/exchangeStrategies/0/bollmaker/defaultBollinger/window'
min: 12
max: 240
- type: rangeFloat # alias: range
path: '/exchangeStrategies/0/bollmaker/spread'
min: 0.001
max: 0.002
- type: bool
path: '/exchangeStrategies/0/bollmaker/buyBelowNeutralSMA'