◈ DISCLOSURE · AI-assisted editorial summary. As an Amazon Associate, ClearValue Books earns from qualifying purchases.

◈ BOOK REVIEW · INVESTING

Quantitative Trading Systems

Name: Quantitative Trading Systems
Author: Dr Howard B Bandy

by Dr Howard B Bandy

Who this is for

For retail algorithmic traders building rule-based systems in AmiBroker who want a rigorous statistical framework for development and validation — and for any systematic trader who wants to understand walk-forward testing methodology regardless of platform.

Buy on Amazon →🎧 Listen on Audible →

Curated by Brian Kim, CPA — every pick gets a plain-English summary and the key takeaways.

Brian Kim, CPA · 2.89M YouTube Subscribers →

◈ KEY TAKEAWAYS

What this book actually teaches

01Walk-forward optimization — optimizing on one historical window, then testing immediately out-of-sample on the next — is the book's core methodology for avoiding overfitting to historical noise.
02Position sizing determines not just expected return but variance of outcomes and probability of ruin; Bandy draws on Kelly criterion and fixed fractional sizing to frame this as a system-design decision, not an afterthought.
03Standard performance metrics (total return, Sharpe ratio) are insufficient for evaluating optimized systems; the book introduces profit factor, maximum adverse excursion (MAE), and bootstrapped confidence intervals as more honest measures.
04The distinction between exit signal types — profit targets versus trailing stops — has material effects on system performance characteristics and deserves as much attention as entry signals.
05Deep AmiBroker/AFL integration limits direct applicability for Python or other platform users; the statistical concepts translate, but the code examples do not.

◈ THE SUMMARY

What's in this book

Scored against ClearValue's published methodology · Jun 28, 2026

Quantitative Trading Systems (2007) by Howard B. Bandy is a technical manual for developing, testing, and evaluating algorithmic trading systems using the AmiBroker platform and AFL (AmiBroker Formula Language). The book's central argument is that discretionary trading — where buy and sell decisions depend on a trader's judgment in real time — introduces psychological biases and inconsistency that systematic rule-based approaches can eliminate. Bandy argues that the value of a trading system lies not in its signal-generation logic but in the rigor of its development and validation process, and that most retail traders fail because they mistake curve-fitted historical performance for predictive power.

The methodology Bandy teaches is built around walk-forward optimization: rather than finding the parameter set that performs best across all historical data (which overfits to noise), the developer optimizes parameters on one historical window and tests immediately out-of-sample on the next, then repeats this process rolling forward. This approach gives a more honest picture of how a system would have performed in real trading conditions, where the future is unknown. The walk-forward framework is the book's most practically valuable contribution and applies regardless of the specific platform used.

Bandy covers the design of trading system components in sequence: entry signals, exit signals (including the mathematically important distinction between profit targets and trailing stops), position sizing, and portfolio-level aggregation. The position sizing chapters draw on concepts from Kelly criterion and fixed fractional position sizing, explaining how the allocation decision affects not just expected return but also the variance of outcomes and the probability of ruin — a dimension many system developers underweight.

The book's statistical framework for evaluating trading system performance is one of its strongest sections. Bandy explains why standard performance metrics like total return and Sharpe ratio are insufficient for evaluating systems developed through optimization, and introduces concepts like the profit factor, the maximum adverse excursion (MAE), and bootstrapped confidence intervals for system metrics. These tools help distinguish systems that have genuine predictive structure from those that merely describe historical noise well.

The weaknesses are structural. The deep integration with AmiBroker's AFL means that readers using other platforms — Python with backtrader or zipline, TradeStation's EasyLanguage, or more modern frameworks like Lean/QuantConnect — must translate concepts manually. The 2007 publication date predates the dominance of machine learning approaches in quantitative trading, and the book does not engage with factor-based systematic investing as practiced institutionally. The retail-trader focus also means the book does not address market impact, execution quality, or co-location — considerations that matter as position sizes grow.

For retail traders building rule-based systems in AmiBroker or for anyone who wants a rigorous introduction to walk-forward validation methodology, this book covers the statistical discipline behind system development more honestly than most competing texts.

Summary as of June 28, 2026AI-assisted summary.