Qlib TRA Configuration for Alpha158 CSI 300 Forecasting
Summary
This configuration specifies a Qlib experiment using Alpha158 features for CSI 300 equities, with the Shanghai Composite 300 index as benchmark. It divides the history into training, validation, and test periods, applies robust feature normalization and missing-value filling, and ranks labels cross-sectionally. The target is a forward close-to-close return over the next daily interval, as encoded by the configured shifted prices.
The model is a recurrent TRA setup using an LSTM, three latent states, and a 60-step input sequence. It is trained with a 2008–2014 training period, validated on 2015–2016, and evaluated on 2017–2020 data. Portfolio analysis uses a top-k dropout strategy that holds up to 50 names and drops five, with close-price execution and specified transaction costs and price-limit assumptions. This is an experiment recipe, not a report of findings: no predictive performance, portfolio returns, or robustness results are included, and the outcome depends on data and implementation details.
Key ideas
- The experiment uses Alpha158 features for CSI 300 stocks and the CSI 300 index as its benchmark.
- Features are robustly normalized and missing feature values are filled before modeling.
- A recurrent LSTM TRA model processes sequences of 60 observations and learns three states.
- The portfolio configuration selects up to 50 stocks and can replace five holdings at a time.
- The file specifies a backtest setup but reports no model or portfolio results.
Tags
Full text
# workflow_config_tra_Alpha158_full.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: RobustZScoreNorm
kwargs:
fields_group: feature
clip_outlier: true
- class: Fillna
kwargs:
fields_group: feature
learn_processors:
- class: CSRankNorm
kwargs:
fields_group: label
label: ["Ref($close, -2) / Ref($close, -1) - 1"]
num_states: &num_states 3
memory_mode: &memory_mode sample
tra_config: &tra_config
num_states: *num_states
rnn_arch: LSTM
hidden_size: 32
num_layers: 1
dropout: 0.0
tau: 1.0
src_info: LR_TPE
model_config: &model_config
input_size: 158
hidden_size: 256
num_layers: 2
rnn_arch: LSTM
use_attn: True
dropout: 0.2
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: TRAModel
module_path: qlib.contrib.model.pytorch_tra
kwargs:
tra_config: *tra_config
model_config: *model_config
model_type: RNN
lr: 1e-3
n_epochs: 100
max_steps_per_epoch:
early_stop: 20
logdir: output/Alpha158_full
seed: 0
lamb: 1.0
rho: 0.99
alpha: 0.5
transport_method: router
memory_mode: *memory_mode
eval_train: False
eval_test: True
pretrain: True
init_state:
freeze_model: False
freeze_predictors: False
dataset:
class: MTSDatasetH
module_path: qlib.contrib.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]
seq_len: 60
input_size:
num_states: *num_states
batch_size: 1024
n_samples:
memory_mode: *memory_mode
drop_last: True
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.