Quantitative Strategy Development and Portfolio Management Platform with Regime-Aware Multi-Stage Validation for Indian Equities

This paper presented S-RABLE, a web platform that unifies strategy development, regime classification, five-stage validation and simulated portfolio management for NSE equities. Its implementation comprises a React 19 front end, a FastAPI backend with 124 endpoints, a 26-table PostgreSQL database, five strategy families, a no-code builder, a three-layer sandbox, an event-driven backtester with slippage and brokerage, a regime engine combining a Gaussian mixture model with Random Forest and XGBoost classifiers, and delayed paper trading with replay. The software was verified by 1,334 backend, 464 front-end and 330 end-to-end test runs. On 125 out-of-sample NIFTY 50 days, the volatility and risk-sentiment classifiers exceeded their majority baselines (61.6% vs 43.2%, and 79.2% vs 75.2%), while the trend classifier did not (27.2% vs 51.2%). The main limitations are bar-close fills, no transaction taxes, margin or borrowing costs, simplified statistical-arbitrage and allocation rules, a weak trend classifier, short-lived access tokens and delayed data. A passed validation is evidence that a defined set of tests was met, not a guarantee of profit, and the platform is a research and educational tool, not investment advice. Future work will (i) add securities transaction tax, exchange charges, stamp duty, GST, margin and borrowing costs and report results gross and net of costs; (ii) offer next-bar-open execution and unify the two Sharpe implementations; (iii) add cointegration tests and hedge-ratio estimation to statistical arbitrage and a trend filter to tactical allocation; (iv) improve the trend classifier with longer histories, more features and probability calibration, and publish confusion matrices; (v) record the number of trials per user so that the deflated Sharpe ratio and the probability of backtest overfitting can be computed [7], [2]; (vi) compare the mixture model with hidden Markov models; (vii) add refresh tokens, token revocation and CORS configuration; and (viii) run extended out-of-sample and multi-month paper-trading experiments with confidence intervals. Real broker order routing would be considered only after compliance review, with paper trading remaining the default.

[1]  D. H. Bailey, J. M. Borwein, M. López de Prado, and Q. J. Zhu, “Pseudo-mathematics and financial charlatanism: The effects of backtest overfitting on out-of-sample performance,” Notices of the American Mathematical Society, vol. 61, no. 5, pp. 458–471, 2014. [2]  D. H. Bailey, J. M. Borwein, M. López de Prado, and Q. J. Zhu, “The probability of backtest overfitting,” Journal of Computational Finance, vol. 20, no. 4, pp. 39–69, 2017. [3]  H. White, “A reality check for data snooping,” Econometrica, vol. 68, no. 5, pp. 1097–1126, 2000. [4]  R. Sullivan, A. Timmermann, and H. White, “Data-snooping, technical trading rule performance, and the bootstrap,” The Journal of Finance, vol. 54, no. 5, pp. 1647–1691, 1999. [5]  S. Kaufman, S. Rosset, C. Perlich, and O. Stitelman, “Leakage in data mining: Formulation, detection, and avoidance,” ACM Transactions on Knowledge Discovery from Data, vol. 6, no. 4, pp. 1–21, 2012. [6]  M. López de Prado, Advances in Financial Machine Learning. Hoboken, NJ, USA: Wiley, 2018. [7]  D. H. Bailey and M. López de Prado, “The deflated Sharpe ratio: Correcting for selection bias, backtest overfitting, and non-normality,” The Journal of Portfolio Management, vol. 40, no. 5, pp. 94–107, 2014. [8]  C. R. Harvey, Y. Liu, and H. Zhu, “… and the cross-section of expected returns,” The Review of Financial Studies, vol. 29, no. 1, pp. 5–68, 2016. [9]  J. D. Hamilton, “A new approach to the economic analysis of nonstationary time series and the business cycle,” Econometrica, vol. 57, no. 2, pp. 357–384, 1989. [10]      A. Ang and G. Bekaert, “International asset allocation with regime shifts,” The Review of Financial Studies, vol. 15, no. 4, pp. 1137–1187, 2002