Qlib Localformer Alpha158 Workflow for CSI 300 Forecasting
Summary
This configuration defines a Qlib equity forecasting workflow using the CSI 300 universe and the Shanghai CSI 300 index as benchmark. It sets a data window from 2008 through 2020, with training through 2014, validation over 2015–2016, and testing over 2017–2020. The dataset uses Alpha158 features, a 20-step sequence, and a forward close-price return label spanning the next two sessions relative to the next session.
The feature pipeline filters a specified subset of Alpha158 fields, applies robust cross-sectional normalization with outlier clipping, and fills missing values; labels are filtered and cross-sectionally rank-normalized. A Localformer model is specified, with signal analysis and portfolio backtesting records. The portfolio strategy selects the top 50 names and drops five each rebalance, with transaction costs and a price limit configured. This is an experiment recipe, not a report of results: it provides no metrics, and its findings would depend on data quality, model implementation, rebalancing assumptions, and trading costs.
Key ideas
- The workflow trains a Localformer model on CSI 300 stocks using Alpha158 features.
- The target is a short-horizon forward close-price return.
- Feature preprocessing includes robust normalization, outlier clipping, and missing-value filling.
- The portfolio simulation holds up to 50 top-ranked stocks and drops five positions at a time.
- The configuration specifies a test period and trading assumptions but reports no performance results.
Tags
Full text
# workflow_config_localformer_Alpha158.yaml
```yaml
qlib_init:
provider_uri: "~/.qlib/qlib_data/cn_data"
region: cn
market: &market csi300
benchmark: &benchmark SH000300
data_handler_config: &data_handler_config
start_time: 2008-01-01
end_time: 2020-08-01
fit_start_time: 2008-01-01
fit_end_time: 2014-12-31
instruments: *market
infer_processors:
- class: FilterCol
kwargs:
fields_group: feature
col_list: ["RESI5", "WVMA5", "RSQR5", "KLEN", "RSQR10", "CORR5", "CORD5", "CORR10",
"ROC60", "RESI10", "VSTD5", "RSQR60", "CORR60", "WVMA60", "STD5",
"RSQR20", "CORD60", "CORD10", "CORR20", "KLOW"
]
- class: RobustZScoreNorm
kwargs:
fields_group: feature
clip_outlier: true
- class: Fillna
kwargs:
fields_group: feature
learn_processors:
- class: DropnaLabel
- class: CSRankNorm
kwargs:
fields_group: label
label: ["Ref($close, -2) / Ref($close, -1) - 1"]
port_analysis_config: &port_analysis_config
strategy:
class: TopkDropoutStrategy
module_path: qlib.contrib.strategy
kwargs:
signal: <PRED>
topk: 50
n_drop: 5
backtest:
start_time: 2017-01-01
end_time: 2020-08-01
account: 100000000
benchmark: *benchmark
exchange_kwargs:
limit_threshold: 0.095
deal_price: close
open_cost: 0.0005
close_cost: 0.0015
min_cost: 5
task:
model:
class: LocalformerModel
module_path: qlib.contrib.model.pytorch_localformer_ts
kwargs:
seed: 0
n_jobs: 20
dataset:
class: TSDatasetH
module_path: qlib.data.dataset
kwargs:
handler:
class: Alpha158
module_path: qlib.contrib.data.handler
kwargs: *data_handler_config
segments:
train: [2008-01-01, 2014-12-31]
valid: [2015-01-01, 2016-12-31]
test: [2017-01-01, 2020-08-01]
step_len: 20
record:
- class: SignalRecord
module_path: qlib.workflow.record_temp
kwargs:
model: <MODEL>
dataset: <DATASET>
- class: SigAnaRecord
module_path: qlib.workflow.record_temp
kwargs:
ana_long_short: False
ann_scaler: 252
- class: PortAnaRecord
module_path: qlib.workflow.record_temp
kwargs:
config: *port_analysis_config
```Shown in full with attribution under the source's licence. Licence: MIT
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.