Qlib Alpha158 DNN Workflow for CSI 300 Portfolio Backtesting
Summary
This configuration describes a Qlib workflow that trains a PyTorch deep neural network on Alpha158 features for CSI 300 stocks, using the Shanghai 300 index as its benchmark. It defines training, validation, and test periods, drops the VWAP feature, fills missing feature values for inference, and applies missing-data handling and label normalization during learning.
The portfolio simulation uses a TopkDropout strategy that targets 50 holdings and replaces five positions, with transaction costs, minimum fees, a price limit threshold, and closing prices specified. The model uses mean squared error with Adam and a defined learning rate, batch size, training steps, weight decay, and GPU setting. The configuration shows intended setup rather than results: it gives no reported returns, risk statistics, or evidence of out-of-sample performance. Its historical sample and assumptions also limit conclusions about current conditions or live trading.
Key ideas
- The workflow trains a PyTorch DNN on Alpha158 features for CSI 300 instruments.
- The data is divided into training, validation, and test segments, with fitting limited to the training period.
- The portfolio strategy targets 50 holdings and rotates five positions at a time.
- The backtest specifies transaction costs, minimum fees, a price threshold, and closing-price execution.
- The configuration contains no performance results, so it cannot establish strategy effectiveness.
Tags
Full text
# workflow_config_mlp_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" : "DropCol",
"kwargs":{"col_list": ["VWAP0"]}
},
{
"class" : "CSZFillna",
"kwargs":{"fields_group": "feature"}
}
]
learn_processors: [
{
"class" : "DropCol",
"kwargs":{"col_list": ["VWAP0"]}
},
{
"class" : "DropnaProcessor",
"kwargs":{"fields_group": "feature"}
},
"DropnaLabel",
{
"class": "CSZScoreNorm",
"kwargs": {"fields_group": "label"}
}
]
process_type: "independent"
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: DNNModelPytorch
module_path: qlib.contrib.model.pytorch_nn
kwargs:
loss: mse
lr: 0.002
optimizer: adam
max_steps: 8000
batch_size: 8192
GPU: 0
weight_decay: 0.0002
pt_model_kwargs:
input_dim: 157
dataset:
class: DatasetH
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]
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.