Matthias
a7ee45a0ba
refactor: Update timezone.utc to UTC
2025-07-04 18:22:42 +02:00
Matthias
ae9073885f
chore: Update log wording, only log "duplicate parameters" once
2025-05-31 16:05:39 +02:00
viotemp1
12d31c4acb
keep INITIAL_POINTS only in hyperopt_optimizer.py
2025-05-31 08:21:44 +03:00
viotemp1
14cfdb13c5
add back INITIAL_POINTS (default 30) for Samplers which support this
2025-05-30 18:17:15 +03:00
viotemp1
dfae7ca2ec
fix duplicate params in same batch also
2025-05-29 15:41:47 +03:00
viotemp1
53383f3184
add up to 5 retries for ask in case of duplicate params
2025-05-28 09:35:20 +02:00
viotemp1
b51c937e87
fix hyperopt repeated parameters between batches
2025-05-27 13:38:03 +02:00
viotemp1
28e5efc902
fix docs
2025-05-17 01:15:10 +03:00
viotemp1
5b92af6a90
fix some spelling errors
2025-05-10 19:45:16 +03:00
viotemp1
442a1ba50d
add early stopping for hyperopt
2025-05-10 19:36:48 +03:00
Matthias
7cfecab7e2
Merge branch 'develop' into optuna
2025-05-08 19:25:33 +02:00
Matthias
8f8da51808
feat: round hyperopt results to 13 digits (this removes floating point errors)
2025-05-08 19:24:31 +02:00
viotemp1
5d2f5ec12f
change SKDecimal low/high to be rounded by decimals
2025-05-07 21:57:56 +03:00
viotemp1
2d2dc7f14a
Merge remote-tracking branch 'refs/remotes/origin/optuna' into optuna
2025-05-07 17:25:14 +03:00
viotemp1
43bd2a060c
fix optuna userwarning the range is not divisible by step
2025-05-07 17:20:17 +03:00
Matthias
6f4ffa0758
chore: minor cleanup
2025-05-04 19:32:09 +02:00
Matthias
8b8bf6f97d
refactor: automatically load detailed bt data
2025-05-03 20:01:13 +02:00
viotemp1
9eea958e17
remove unwrap
2025-05-03 16:55:41 +03:00
Matthias
1013c32316
fix: duplicate generate_optimizer to have a non-delayed alternative
2025-05-03 09:19:50 +02:00
viotemp1
d9ed7e1fb2
remove backtest from assign_params
2025-05-03 09:49:34 +03:00
viotemp1
dd613ac86c
fix formatting
2025-05-02 21:07:34 +03:00
viotemp1
73c28890d7
move @delayed and @wrap_non_picklable_objects in hyperopt_optimizer.py
...
one test with analyze_per_epoch is failing
2025-05-02 21:03:51 +03:00
Matthias
149133cc44
refactor: remove _get_params_dict method
2025-05-01 12:03:04 +02:00
viotemp1
ced1ce340a
fix some formatting issues
2025-04-29 17:09:16 +03:00
viotemp1
5c47a75f63
move back delayed and wrap_non_picklable_objects from hyperopt_optimizer to hyperopt.
...
There are tests failing when using delayed and wrap_non_picklable_objects as decorator.
until I'll find a solution to run generate_optimizer standalone for analyze_per_epoch=True
2025-04-29 08:16:44 +03:00
viotemp1
c32f8e972a
move delayed and wrap_non_picklable_objects from hyperopt to hyperopt_optimizer
2025-04-28 18:50:09 +03:00
viotemp1
04492e75b2
remove # Suppress optuna ExperimentalWarning from skopt
...
with warnings.catch_warnings():
from optuna.exceptions import ExperimentalWarning
warnings.filterwarnings("ignore", category=FutureWarning)
this should be when importing sampler
2025-04-27 22:02:39 +03:00
viotemp1
8d0ca7f5c1
remove backtesting from generate_optimizer
2025-04-27 21:26:06 +03:00
Matthias
fb64ac942b
chore: raise freqtrade exception so upstream handling is in place
2025-04-23 20:18:58 +02:00
Matthias
3fc40f45b3
chore: simplify diff in hyperopt-tinterface
...
Use aliases where possible.
2025-04-23 20:15:24 +02:00
Matthias
ca5ccc8799
chore: cleanup some code
2025-04-23 20:07:40 +02:00
Matthias
ed22789a1a
chore: cleanup unused import
2025-04-23 19:53:07 +02:00
Matthias
2abf22e37b
chore: simplify usage of data_pickle_file
2025-04-23 19:50:17 +02:00
Matthias
83cdf76636
chore: simplify import/exports
2025-04-23 19:50:17 +02:00
Matthias
f545113840
chore: improve EstimatorType type
2025-04-23 19:50:17 +02:00
viotemp1
20fca07d8f
fixed mypy errors
...
freqtrade/optimize/space/optunaspaces.py:39: error: Argument 1 to "__init__" of "IntDistribution" has incompatible type "int | float"; expected "int" [arg-type]
freqtrade/optimize/space/optunaspaces.py:39: error: Argument 2 to "__init__" of "IntDistribution" has incompatible type "int | float"; expected "int" [arg-type]
remove all references for ExtraTreesRegressor and skopt.space
2025-04-13 12:39:44 +03:00
viotemp1
35c3868c56
change for SKDecimal and othercomments
2025-04-13 08:58:42 +03:00
Matthias
4fcc9dd587
feat: use floatDistribution for SKDecimal
2025-04-12 12:10:48 +02:00
Matthias
05f19d574a
chore: remove commented skopt usages
2025-04-12 10:13:42 +02:00
Matthias
7a51c9d540
types: slightly improved typing
2025-04-12 10:05:11 +02:00
Matthias
4986f356ee
refactor: Type backtest results
2025-04-07 19:28:43 +02:00
viotemp1
90aaaa50fc
fix increasing memory usage.
2025-04-02 18:45:49 +03:00
viotemp1
3fcf6559ab
change from skopt.space.Real to optuna.distributions.FloatDistribution
2025-03-31 13:48:12 +03:00
viotemp1
85f4a8daea
fix formatting
2025-03-31 00:10:52 +03:00
viotemp1
2595479e43
change CategoricalParameter and IntParameter in parameters.py to use optuna.distributions CategoricalDistribution and IntDistribution instead of skopt
2025-03-30 21:13:15 +03:00
viotemp1
6b78b1c882
fix formatting
2025-03-26 21:30:18 +02:00
viotemp1
ee3d46c8fa
change default optimizer to NSGAIIISampler - best results so far
2025-03-26 21:25:56 +02:00
viotemp1
59e52bb601
fix type-errors by declaring the type of o_dimensions in convert_dimensions_to_optuna_space
2025-03-26 16:55:43 +02:00
viotemp1
553dbccec7
simplify get_optimizer
2025-03-26 16:42:09 +02:00
viotemp1
eb03382b2d
remove SKOPT_MODEL_QUEUE_SIZE comments
2025-03-26 08:36:55 +02:00
