Qlib Configuration for TCTS Alpha360 Stock Prediction and Top-K Backtesting
Summary
This configuration describes a Qlib workflow for training a TCTS model on the CSI 300 universe using Alpha360 features. The data span 2008 to 2020, with training through 2014, validation over 2015–2016, and testing from 2017 to mid-2020. Feature processing applies robust z-score normalization and fills missing values; labels are cross-sectionally rank-normalized after missing labels are dropped. The model is set up to predict three forward-return labels, with mean squared error and Adam optimizers.
For portfolio analysis, the configuration uses a top-k dropout strategy that holds 50 names and allows five drops, with the Shanghai Composite 300 as benchmark. The backtest uses closing prices and specifies trading costs, a minimum fee, and a limit threshold. It also records signal analysis and portfolio analysis. This is an experimental setup rather than a report of findings: it contains no performance results, and the outcome depends on data quality, implementation details, and the assumptions encoded in the backtest.
Key ideas
- The workflow trains a TCTS model on Alpha360 data for the CSI 300 universe.
- Training, validation, and test periods are specified separately.
- The model targets three forward-return labels and uses mean squared error.
- Portfolio analysis applies a top-k dropout strategy with 50 holdings and five permitted drops.
- The configuration includes transaction costs and closing-price execution assumptions but reports no results.
Tags
Full text
# workflow_config_tcts_Alpha360.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: DropnaLabel
- class: CSRankNorm
kwargs:
fields_group: label
label: ["Ref($close, -2) / Ref($close, -1) - 1",
"Ref($close, -3) / Ref($close, -1) - 1",
"Ref($close, -4) / Ref($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: TCTS
module_path: qlib.contrib.model.pytorch_tcts
kwargs:
d_feat: 6
hidden_size: 64
num_layers: 2
dropout: 0.3
n_epochs: 200
early_stop: 20
batch_size: 800
metric: loss
loss: mse
GPU: 0
fore_optimizer: adam
weight_optimizer: adam
output_dim: 3
fore_lr: 2e-3
weight_lr: 2e-3
steps: 3
target_label: 0
lowest_valid_performance: 0.993
dataset:
class: DatasetH
module_path: qlib.data.dataset
kwargs:
handler:
class: Alpha360
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.