Skip to main content
WeiMindWeiMind
ChineseView Course
Quant Radar
#Backtesting
18 JunThu
  1. Rob Carver：This Blog is Systematic60

    Equal-Weighted Averaging of Portfolio Weights Generated by Different Shrinkage Methods Fails to Outperform Single Shrinkage Approaches Across Multiple In-Sample and Out-of-Sample Backtests

    The author examines the effect of equal-weighted averaging of portfolio weights generated by different shrinkage methods in the Portfolio Optimization Series, comparing 9 shrinkage parameter combinations with 6 weight averaging combinations, totaling 15 approaches. The backtests use randomly selected 9 trading rules and random instruments, with 1-, 5-, and 10-year in-sample periods and 1- and 5-year out-of-sample periods, each group containing 4,000 random portfolios.

    Read original
15 JunMon
  1. Rob Carver：This Blog is Systematic62

    Comparing Portfolio Optimization Shrinkage Methods Using Real Return Data: Backtesting Conclusions for Monte Carlo, Bootstrap, and Double Shrinkage

    In the fourth installment of the Summer 2026 Portfolio Optimization Series, the author backtests Monte Carlo, bootstrap, double shrinkage, naive mean-variance, equal weight, maximum diversification, and EPO using real trading rule and instrument return data. Each test randomly selects 9 trading rules and one instrument, comparing portfolios over 1-, 5-, and 10-year in-sample periods and 1- and 5-year out-of-sample periods. The results show that under real data, correlation estimates are less stable than Sharpe ratios, requiring more correlation shrinkage rather than more Sharpe ratio shrinkage.

    Read original
8 JunMon
  1. Robot Wealth45

    How Independent Traders Should Allocate Resources for a Triangulated Statistical Arbitrage Business

    This article is the sixth part of a series on statistical arbitrage for independent traders. It explains the three core modules of triangulated statistical arbitrage and their resource allocation priorities. High-quality pair screening is the core foundation: using 24 months of historical data, the top 10% of pairs are screened monthly and divided into 5 quality tiers, with the best tier achieving a Sharpe ratio of 1.6 and the worst only 0.42. The author discloses only the general methodology and does not reveal specific implementation details.

    Read original
29 AprWed
  1. Robot Wealth36

    Data mining and AI-driven 'quant by feel' are essentially the same in quantitative trading; the core should focus on the structural source of trading edge.

    The article points out that data mining and AI-driven 'quant by feel' are identical in essence and philosophy, both attempting to skip the core step of understanding how markets work and profit directly from the market. A genuine trading edge requires identifying the counterparty being profited from, the reason they continue to pay, and the rationale for one's own ability to participate in the trade; such structural edge cannot be identified through backtesting optimization alone.

    Read original
29 MarSun
  1. Robot Wealth64

    Brave New Backtest: Why LLMs Have Three Structural Limitations in Trading Research

    Drawing on two academic papers, the author identifies three structural limitations of using LLMs in trading research: First, training data is saturated with trading common sense and lacks mechanism-driven edge thinking. Second, across 35+ models, accuracy in time-series value tracking declines logarithmically as the number of historical updates increases, and neither prompt engineering nor expanding the context window can fix this. Third, outputs from 25+ models on open-ended questions show 71%–82% similarity, and models with different architectures converge to the same common answers.

    Read original
23 MarMon
  1. Robot Wealth54

    AI Will Create Millions of Quantitative Traders, Most of Whom Will Lose Money

    The author notes that AI has dramatically lowered the barrier to developing quantitative strategies, enabling fully coded and backtested strategies to be generated within minutes. However, cheap experimentation amplifies the multiple-testing problem, producing a flood of false backtest results that appear effective but are actually overfit to noise.

    Read original
20 MarFri
24 DecWed