Strategy #0000000104 · Machine learning and asset pricing
Machine-learning cross-sectional return ranking
Enter predicted and assumed realised long-short spreads, portfolio turnover and cost per turnover.
Machine-learning models can fit nonlinear return relationships from firm characteristics and rank a long-short portfolio by predicted returns. Results depend on point-in-time data, out-of-sample testing, feature engineering, portfolio construction, benchmarks and costs; leakage and repeated tuning can create false performance. This scenario compares manually entered predicted and realised spreads with turnover costs.
No model, firm characteristics or prices are connected; no forecast accuracy, statistical significance or out-of-sample performance is claimed.
Three ways in
Understand the story. Then inspect the risk.
Start with a familiar situation, check the payoff, then test how the numbers change.
The idea · in two minutes
How do a model-ranked prediction spread, supplied realised spread and turnover costs differ?
Machine-learning models can fit nonlinear return relationships from firm characteristics and rank a long-short portfolio by predicted returns. Results depend on point-in-time data, out-of-sample testing, feature engineering, portfolio construction, benchmarks and costs; leakage and repeated tuning can create false performance. This scenario compares manually entered predicted and realised spreads with turnover costs.
Enter predicted and assumed realised long-short spreads, portfolio turnover and cost per turnover.
No model, firm characteristics or prices are connected; no forecast accuracy, statistical significance or out-of-sample performance is claimed.
- Input assumptionsEnter predicted and assumed realised long-short spreads, portfolio turnover and cost per turnover.
- Model outputNo model, firm characteristics or prices are connected; no forecast accuracy, statistical significance or out-of-sample performance is claimed.
Illustrative research examples only. They are not live signals or a recommendation.
The mechanics · payoff and exposure
Spread_{net}=R_{top}-R_{bottom}-\frac{Turnover\cdot Cost_{turnover}}{100}
The predicted spread describes a model output only; realised long-short return and turnover are manual scenarios, not backtest data.
Greeks: the directional intuition
- 訊號
- Measures sensitivity to implied volatility.
- 樣本
- Measures sensitivity to implied volatility.
- 執行
- Measures sensitivity to implied volatility.
Spread_{net}=R_{top}-R_{bottom}-\frac{Turnover\cdot Cost_{turnover}}{100}
Practical context · published observations
Machine-learning cross-sectional return ranking: read the record with its caveats.
The chart uses the published figures or explicitly labelled study inputs below. A historical index or model replay is not the same as an investor’s realised account return.
No model, firm characteristics or prices are connected; no forecast accuracy, statistical significance or out-of-sample performance is claimed. Source ↗ · Methodology ↗
Change assumptions and inspect the scenario
Inputs illustrate the strategy mechanics; they are not live quotes, forecasts or investment advice.
Variants · same family, different risk levers
Data quality, sample selection and execution assumptions affect results.
Compare signal, sample and implementation choices.
| Related strategy | Relationship | What changes | New cost or risk |
|---|---|---|---|
| Base scenario | Manual inputs | Machine-learning models can fit nonlinear return relationships from firm characteristics and rank a long-short portfolio by predicted returns. Results depend on point-in-time data, out-of-sample testing, feature engineering, portfolio construction, benchmarks and costs; leakage and repeated tuning can create false performance. This scenario compares manually entered predicted and realised spreads with turnover costs. | No model, firm characteristics or prices are connected; no forecast accuracy, statistical significance or out-of-sample performance is claimed. |
| Conservative version | Lower exposure | Reduce exposure and raise sample thresholds. | May narrow strategy coverage. |
| Validation version | Out-of-sample review | Freeze signal definitions before time-ordered evaluation. | Needs more clean data and governance. |
No model, firm characteristics or prices are connected; no forecast accuracy, statistical significance or out-of-sample performance is claimed.
Read beyond the summary
References
Primary sources and research used to describe the strategy and its limits.