StratevaEvidence
Published work, including the parts that did not pass.
Every case here is public and reproducible. None of them is a client engagement, and none of them is a success story — which is rather the point of publishing them.
Two kinds of case, and they are not equivalent
- Validation audit
- An independent audit of a claim, taken through the full validation protocol and ending in a formal outcome.
- Research example
- A strategy built by Strateva and published for transparency, to show how work is documented and reviewed. Some were run through the full protocol and carry an outcome; those that were not are research, not validated strategies.
These are not client references. No client work is published, no returns are advertised and nothing here is a recommendation to trade.
Public case studies
RF100 — Can a Random Forest beat simple momentum?
- Market
- US equities, spot
- Strategy type
- Cross-sectional machine learning
What was checked
A causal cross-sectional Random Forest found statistically positive predictive structure but failed to beat a simple momentum benchmark under equivalent execution assumptions.
Main findingPredictive structure is not the same as incremental, monetizable alpha.
EWMA63 — Does beta selection add value beyond de-risking?
- Market
- US equities, spot
- Strategy type
- Volatility-based allocation overlay
What was checked
The audit showed that most of the historical Sharpe and drawdown improvement came from lower market exposure, while the incremental value of the EWMA63 allocation layer was approximately zero.
Main findingThe correct control can invalidate an apparently successful strategy.
Historical research — outside the current commercial scopeKept for the methodology it demonstrates, not as an example of something you could send today. Prediction markets and derivative-led strategies are not accepted.
F2 Anticipator V2 — Can a late-book signal anticipate 5-minute BTC Up/Down markets?
- Market
- Polymarket prediction markets
- Strategy type
- Order-book microstructure signal
What was checked
A causal, book-only model estimated the final probability of Up and traded only when expected edge exceeded costs. The temporal OOS showed that the predicted edge did not survive realistic execution assumptions.
Main findingPredicted edge is not monetizable alpha when it is poorly calibrated, fails to persist and deteriorates after the model freeze.
Read the case study — F2-ANTICIPATOR-V2View repository on GitHub — F2-ANTICIPATOR-V2 (opens in a new tab)This market is not accepted today. The CTA below is for strategies inside the current scope: cash equities and ETFs, spot crypto and spot FX.
Volatility Term-Structure Strategy
- Market
- SPY ETF, spot
- Strategy type
- Volatility term-structure regime filter
What was checked
A transparent SPY allocation example combining a lagged VIX term-structure filter, realized-volatility targeting, leverage limits and explicit turnover costs.
Main findingLower volatility and drawdown do not automatically imply benchmark outperformance or alpha.
Have something that needs an independent check?
Describe the strategy, the backtest or the model. The scope and the price are confirmed with you before any work starts.
Nothing runs, and nothing is charged, before you approve the scope.
The information you submit will be used to assess your request. Read the privacy policy.