Skip to content
EGAlpha

Methodology

What distinguishes a research operation is the protocol that decides whether a backtest becomes a position. Ours is formal, versioned, and applied identically to every study.

Research protocol and pre-registration

Every study runs under a versioned program constitution. Before any estimation, the hypotheses, test statistics, decision thresholds, and data partitions are committed. Data is partitioned into discovery, validation, and sealed out-of-sample blocks; the integrity of the out-of-sample seal is enforced programmatically by lock-file timestamps, so the confirmatory block cannot be inspected before the in-sample stage is finalized.

Results are adjudicated against the pre-committed criteria irrespective of outcome. Studies that fail to reject the null are retained in the research record rather than discarded — directly controlling the selection bias (the file-drawer problem and specification search) that inflates apparent performance when only favorable backtests are reported. Multiple pre-registered hypotheses in our program have returned null; they are recorded as such, and the corresponding features or strategies were not deployed.

Validation and statistical rigor

Selecting strategies from a universe of candidates inflates the false-discovery rate. We correct for this at pre-registration using Bonferroni and Holm family-wise error control for confirmatory tests, and Benjamini–Hochberg FDR control for screening stages. Reported performance is then deflated for the number of trials and non-normality of returns via the Deflated Sharpe Ratio (Bailey & López de Prado, 2014), and live performance is tracked against the validated baseline using Wald's sequential probability ratio test for early decay detection.

No edge is accepted on a single test. Each candidate must survive a robustness battery — walk-forward out-of-sample validation, leave-one-out cross-validation, rolling-window stability checks, block-bootstrap confidence intervals excluding zero at 95% and 99%, and random-null permutation tests — before advancing. Edges are decomposed across market regimes so conditional performance is known rather than averaged away, and every estimate passes through a capacity-constrained simulator enforcing position limits, concurrent-exposure caps, and realistic transaction costs. The backtest-to-live pipeline reproduces historical results to the exact value, and adversarial reviews target untested conditions before each phase transition.

Risk and execution architecture

  • —Layered monitoring — independent automated monitors for position-, strategy-, portfolio-, and macro-event risk, aggregated into one composite deploy / no-deploy determination.
  • —Hard kill-switches at position, strategy, and portfolio level, with thresholds set in advance.
  • —Decay early warning — live performance compared to each strategy's validated baseline; an alert is raised before any kill threshold is reached.
  • —Position sizing — fractional-Kelly under hard caps, with shrinkage-based covariance estimation (Ledoit & Wolf, 2004) for portfolio construction.

From hypothesis to position

  1. 01

    Hypothesis — A mechanistic thesis grounded in literature and first principles.

  2. 02

    Pre-registration — Methodology, statistics, thresholds, and partitions fixed before estimation.

  3. 03

    Discovery and validation — In-sample, independent validation, sealed out-of-sample, through the robustness battery.

  4. 04

    Adversarial review — Fresh-eyes audit against untested conditions.

  5. 05

    Engineering and risk — Production build with reproducibility guarantees, wrapped in the layered risk architecture.

  6. 06

    Shadow — Live signals, zero capital, validated against the backtest distribution (Kolmogorov–Smirnov).

  7. 07

    Staged deployment — Capital-controlled, operator-supervised; scaled only on accumulated evidence.

  8. 08

    Monitoring and retirement — Continuous decay detection; retirement on evidence.

Selected references

Methods and literature in active use.

Multiple-testing & false-discovery control

  • Benjamini, Y. & Hochberg, Y. (1995). Controlling the false discovery rate. JRSS-B.
  • Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scand. J. Stat.
  • White, H. (2000). A reality check for data snooping. Econometrica.
  • Romano, J. P. & Wolf, M. (2005). Stepwise multiple testing as formalized data snooping. Econometrica.
  • Harvey, C. R., Liu, Y. & Zhu, H. (2016). ...and the cross-section of expected returns. Review of Financial Studies.
  • Harvey, C. R. & Liu, Y. (2020). Lucky factors. J. Financial Economics.
  • Chen, A. Y. (2025). Most claimed statistical findings in cross-sectional return predictability are likely true. arXiv:2206.15365.

Performance measurement & backtesting

  • Bailey, D. H. & López de Prado, M. (2012). The Sharpe Ratio efficient frontier. J. Risk.
  • Bailey, D. H. & López de Prado, M. (2014). The Deflated Sharpe Ratio. J. Portfolio Management.
  • Bailey, D. H. et al. (2014). Pseudo-mathematics and financial charlatanism: the effects of backtest overfitting. AMS Notices.
  • Arnott, R. D., Harvey, C. R. & Markowitz, H. (2019). A backtesting protocol in the era of machine learning. J. Financial Data Science.
  • Pham, T. A., Nguyen, B. C. & Nguyen, N. T. (2026). AlgoXpert Alpha Research Framework: a rigorous IS-WFA-OOS protocol for mitigating overfitting. arXiv:2603.09219.
  • Yin, D. et al. (2026). Implementation risk in portfolio backtesting: a previously unquantified source of error. arXiv:2603.20319.

Momentum & cross-sectional strategies

  • Jegadeesh, N. & Titman, S. (1993). Returns to buying winners and selling losers. J. Finance.
  • Novy-Marx, R. (2012). Is momentum really momentum? J. Financial Economics.
  • Asness, C. S., Moskowitz, T. J. & Pedersen, L. H. (2013). Value and momentum everywhere. J. Finance.
  • Da, Z., Gurun, U. G. & Warachka, M. (2014). Frog in the pan: continuous information and momentum. Review of Financial Studies.
  • Barroso, P. & Santa-Clara, P. (2015). Momentum has its moments. J. Financial Economics.
  • Daniel, K. & Moskowitz, T. J. (2016). Momentum crashes. J. Financial Economics.

Mean-reversion & microstructure

  • Kyle, A. S. (1985). Continuous auctions and insider trading. Econometrica.
  • Poterba, J. M. & Summers, L. H. (1988). Mean reversion in stock prices: evidence and implications. J. Financial Economics.
  • Lo, A. W. & MacKinlay, A. C. (1990). When are contrarian profits due to stock market overreaction? Review of Financial Studies.
  • Cont, R., Kukanov, A. & Stoianov, S. (2014). The price impact of order book events. J. Financial Econometrics.
  • Anantha, A. N. & Jain, S. (2024). Forecasting high frequency order flow imbalance. arXiv:2408.03594.
  • Epstein, E. L. et al. (2025). Attention factors for statistical arbitrage. arXiv:2510.11616.

Risk, portfolio construction & statistical methods

  • Wald, A. (1945). Sequential tests of statistical hypotheses. Ann. Math. Stat.
  • Kelly, J. L. (1956). A new interpretation of information rate. Bell System Technical Journal.
  • Hamilton, J. D. (1989). A new approach to the economic analysis of nonstationary time series and the business cycle. Econometrica.
  • Künsch, H. R. (1989); Politis, D. N. & Romano, J. P. (1994). The block / stationary bootstrap.
  • Ledoit, O. & Wolf, M. (2004). Honey, I shrunk the sample covariance matrix. J. Portfolio Management.
  • Maillard, S., Roncalli, T. & Teiletche, J. (2010). The properties of equally weighted risk contribution portfolios. J. Portfolio Management.
  • López de Prado, M. (2018). Advances in Financial Machine Learning. Wiley.
  • Fischer, L. & Ramdas, A. (2024). Improving Wald's (approximate) sequential probability ratio test by avoiding overshoot. arXiv:2410.16076.
  • Oliveira, D. C., Guzman, G. & Firoozye, N. (2025). (Non-parametric) bootstrap robust optimization for portfolios and trading strategies. arXiv:2510.12725.
  • Parra-Diaz, M. & Castro-Iragorri, C. (2025). Deep declarative risk budgeting portfolios. arXiv:2504.19980.

Crowding, short interest & regime conditioning

  • Boehmer, E., Jones, C. M. & Zhang, X. (2008). Which shorts are informed? J. Finance.
  • Rapach, D. E., Ringgenberg, M. C. & Zhou, G. (2016). Short interest and aggregate stock returns. J. Financial Economics.
  • Lou, D. & Polk, C. (2022). Comomentum: inferring arbitrage activity from return correlations. Review of Financial Studies.
  • Zhang, Y. et al. (2025). RegimeFolio: a regime aware ML system for sectoral portfolio optimization in dynamic markets. arXiv:2510.14986.
  • Deep, G., Deep, A. & Lamptey, W. (2025). Interpretable hypothesis-driven trading: a rigorous walk-forward validation framework. arXiv:2512.12924.
  • Sun, J. et al. (2026). Synthetic American option pricing via Jump-HMM-driven Heston implied volatility. arXiv:2605.13998.

What we keep proprietary

We describe our methodology in depth and report our results honestly. We do not publish the deployable parameters of our strategies — the universes, thresholds, sizing rules, and signal definitions that constitute the edge. A real edge is finite and is treated accordingly. We are transparent about how an edge is established and how the capital trading it is protected.

Contact