Statistics Summary

Core idea

The Statistics Summary topic is the back-of-the-book reference dataset that makes the entire Encyclopedia tradeable. Each preceding topic described one pattern's performance in isolation; this topic sorts and ranks all 75 patterns side-by-side across four cuts: bull-up, bear-up, bull-down, bear-down. Two parallel sets of tables exist — one for performance (average move after breakout, where higher is better) and one for failure rates (5% failure rates, where lower is better). The patterns are listed alphabetically and ranked.

The summary is what lets you answer comparative questions: "What is the best-performing bullish pattern in a bear market?" "What are the lowest-failure-rate bearish patterns?" "Which patterns rank top-ten in both performance and reliability?" Without this topic, the book is 75 isolated studies; with it, the book becomes a comparative pattern-selection tool.

Bulkowski's framing: Performance and failure rate measure different things. A high-performance pattern can also be a high-failure pattern (big winners, lots of misses). A low-failure pattern can be a low-performance one (small wins, but very reliable). The right pattern depends on your edge: trade size, win rate, frequency, and capital constraints.

Why it matters

The Encyclopedia's 75 topics describe patterns one at a time. A trader scanning a chart sees three or four candidate patterns simultaneously and must choose. That choice is impossible without a relative ranking — the Statistics Summary is the structural answer to "of these candidates, which one historically rewards the trade?"

Why performance and failure rate are split

Performance answers "how much do I make when it works." Failure rate answers "how often does it work at all." A pattern with a 30% average rise and a 25% failure rate has a different expected value profile than a pattern with a 20% average rise and an 8% failure rate. The latter is usually preferable for sized trading because the variance is lower — but a high-conviction trader with tight stops may prefer the former for the bigger payoffs. Both tables matter; reading only one is a mistake.

Why four cuts (bull/bear x up/down)

Patterns behave differently in different regimes. The same head-and-shoulders top has a 16% bull-market decline but a 23% bear-market decline. The same triple bottom has a 13% bull-market failure rate but ~25% in bear markets. Lumping all data together hides this. Bulkowski's four-quadrant separation lets you ask the regime-specific question: "in a bear market with downward breakouts, what works?"

Why scale and pattern-class footnotes matter

Some patterns use the weekly or monthly scale rather than daily; their move-magnitudes are not directly comparable to daily patterns. Harmonic / Fibonacci patterns (Bat, Crab, Butterfly, Gartley, Wolfe Wave) use a special measure rule — same caveat. The asterisk footnotes in the tables exist for a reason: comparing a daily-scale pattern's 14% rise to a weekly-scale pattern's 32% rise is apples-to-cars-to-oranges.

Key takeaways

Mental model

Copy sourceZoom

Practical application

The Statistics Summary is a reference, not a topic to read end-to-end. The practical workflow:

Step 1: Identify your regime

Before opening the tables, classify the current market. Is the broad index in a confirmed bull or bear regime? Is the candidate pattern's expected breakout direction up or down? This determines which of the four cuts (bull-up, bear-up, bull-down, bear-down) you consult.

Step 2: Choose performance OR failure rate based on trading style

If you trade with tight stops and need a high hit rate, prioritize the failure rate tables. If you trade with wider stops and need bigger payoffs to justify the risk, prioritize performance. Most traders should look at both.

Step 3: Look up your candidate patterns

For each pattern showing on the chart, find its rank in the relevant cut. Rank 1 is best; rank N is worst (N = number of patterns in that table). Filter to the top tercile or top decile depending on selectivity.

Step 4: Use the alphabetical list as a backstop

When ranking is close, the alphabetical list shows the raw percentage. A pattern ranked 8th with a 22% average rise is meaningfully different from a pattern ranked 12th with a 21% rise — the rank order obscures small differences. Look at the underlying number when the call is tight.

Example

A trader is reviewing a small-cap industrial stock in a confirmed bull market. The chart shows three potentially valid patterns:

1. Symmetrical triangle with upward breakout pending
2. Cup with handle in late-formation stage
3. Pennant off a recent flagpole

All three suggest an upward breakout. Without the Statistics Summary, the trader picks intuitively. With the Statistics Summary, the trader looks up bull-market, upward breakout rankings:

• Cup with Handle: top 5 in performance, top 5 in failure rate (low-failure, high-performance)
• Symmetrical Triangle: mid-pack in both
• Pennant: top-10 in performance, mid-pack in failure rate

The Cup with Handle wins on both metrics. The trader sizes the trade on that pattern and treats the other two as confirmation rather than independent setups. This is the Summary topic's entire purpose: turning a list of "possible patterns" into a defensible ranking.

Then the trader re-runs the same exercise the next month, when the market has rolled into a confirmed bear phase. The same Cup with Handle setup would now be checked against the bear-market, upward breakout table — where it has too few samples to rank. Conclusion: skip the trade. That regime-aware filtering is also the topic's job.

Related lessons