Technical Analysis: The Complete Resource for Financial Market Technicians
Charles D. Kirkpatrick II and Julie Dahlquist · 2011 · 700 pages
This is the closest thing technical analysis has to a college textbook: every method gets a definition, a formula, a chart, and a citation to the academic study that tested it. It hands you the vocabulary and the tools but stops short of telling you which combination actually makes money.
The one idea
The book's real argument is not any single indicator, it is that a trend is a temporary imbalance of power between buyers and sellers, and the job of the technician is to identify that imbalance early, ride it, and get out before it reverses. Everything else in the 700 pages (support and resistance, chart patterns, moving averages, Elliott wave, cycles) is a different lens for answering the same two questions: is a trend in force, and is it ending.
The authors spend the first four chapters defending technical analysis against the Random Walk Hypothesis and Efficient Markets Hypothesis before teaching a single indicator. Their case rests on Andrew Lo and Craig MacKinlay's 1988 finding that stock returns are not proportionally distributed the way a true random walk requires, and on Cheol-Ho Park and Scott Irwin's 2003 review of 92 post-1986 academic studies, in which 58 found technical strategies produced positive results against 24 that found negative results. Neither study proves any specific rule works, only that the door is open.
Once past the theory, the book systematically catalogs the field: trend and trading-range identification (Chapter 12), breakout and stop methods (Chapter 13), moving averages (Chapter 14), bar and point-and-figure chart patterns with Thomas Bulkowski's failure-rate statistics attached to each one (Chapters 15 to 17), volume and breadth confirmation (Chapters 8 and 18), cycles and Elliott/Fibonacci/Gann (Chapters 19 and 20), and, in the last two chapters, how to actually design, test, and risk-manage a system built from any of it. That last section is where the book is most honest about how little of the preceding 550 pages has been proven to work in real time.
What it actually teaches
The book does not sell one system, so the closest thing to a repeatable process is the sequence it walks a reader through, chapter by chapter, to build and test a trend-following approach from its own toolkit.
IDENTIFY THE TREND WITH PEAKS AND TROUGHS
Chapter 12 defines an uptrend as a series of higher peaks and higher troughs, a downtrend as lower peaks and lower troughs, and a trading range as peaks and troughs clustering at the same levels. To find a significant reversal point, the book offers several filters: the DeMark/Williams method (a trough needs a set number of bars on each side with higher lows), the Gann two-bar swing method (a low bar followed by two bars with higher highs confirms a support point), and a percentage method (commonly 1 percent, the larger the percentage the more significant but less frequent the reversal).
DRAW SUPPORT, RESISTANCE, AND WAIT FOR A BREAKOUT
Horizontal lines through significant peaks and troughs mark support and resistance zones. The book argues range trading is 'almost impossible for the nonprofessional' because of execution costs and imprecise boundaries, and instead recommends trading the breakout: buy when resistance is broken, sell when support is broken. The Donchian four-week breakout system (buy the highest high of the past four weeks, sell the lowest low) is cited as a rule that 'still appears to work' per a 2003 Active Trader Magazine study, especially in commodity futures.
CONFIRM THE BREAKOUT BEFORE ACTING
Because intra-bar penetrations are frequently false, Chapter 13 lists five confirmation filters: waiting for the closing price to hold beyond the level, a point-or-percent filter (the book names 3 percent as the most common), a time filter (usually two bars), a volume increase, and an Average True Range (ATR) filter, commonly 1.5 times the 14-day ATR added beyond the level. None is presented as reliably better than the others; the trade-off is always speed of entry against certainty that the breakout is real.
SET A STOP BASED ON PRICE ACTION, NOT AN ARBITRARY NUMBER
Protective stops belong at the level where the original analysis would be proven wrong, not at a round percentage. The book's worked example: a stock breaks out at $20, the investor decides $17 invalidates the trade, so the risk is $3 a share. With $100,000 of capital and a 10 percent maximum single-position risk, that caps the position at 3,300 shares. For trailing stops on a running position, the Chandelier Exit places the stop a multiple (the book suggests 2.5 to 4 times) of the 14-day ATR below the highest high since entry; a $50 stock with a $2.50 ATR and a 3x multiple gets a stop at $42.50, which then only moves up as new highs are made.
SIZE THE TRADE WITH A MINIMUM 3-TO-1 REWARD/RISK RATIO
The price objective after a breakout is usually the next support or resistance zone. The book states the reward/risk ratio 'should be a minimum of three to one,' which means the position can still be profitable even if two out of every three trades lose, provided the winner delivers the full three-unit gain.
TEST THE SYSTEM OUT OF SAMPLE BEFORE TRADING IT LIVE
Chapter 22 splits historical data into an in-sample set (70 to 80 percent) used to optimize parameters and an out-of-sample set (20 to 30 percent) used only to check the result. The book warns that out-of-sample performance will 'invariably' be worse than the optimization suggested, and cites Hill, Pruitt, and Hill's rule of thumb to 'generally expect half the profits and twice the drawdowns' shown in backtesting once a system goes live.
What it looks like Monday morning
A reader who has never formalized a trading approach walks away with a concrete checklist: pick a trend-identification rule (peaks and troughs, or a moving-average crossover), pick a breakout confirmation filter (the book leans toward ATR-based filters as the most volatility-aware), place a stop at a price level where the thesis is actually wrong rather than at a fixed percentage, and size the position so no single loss exceeds a set fraction of capital.
The more important habit the book pushes for Monday morning is deferred gratification: before risking money on any of its 20-plus methods, split available history into in-sample and out-of-sample data and test the exact rule, not a close approximation of it, on the out-of-sample portion before trading it. The book is explicit that skipping this step is why most amateur systems fail live even when they looked great in a backtest.
| Measure | Rate | What it means |
|---|---|---|
| Pattern fails to break the confirming trough at all | 64% | Most apparent double tops never become a valid, tradable pattern; the breakout never happens |
| Failure rate once the breakout occurs | 11% | Of the patterns that do complete with a valid breakout, only 11% then fail to move at least 5% in the breakout direction |
| Bulkowski's overall performance rank | 2nd of 21 patterns | Composite score based on failure rate, average profit, pullback/throwback rate, and percent of trades reaching a price target |
The gap between the two failure numbers is the book's clearest illustration of why waiting for confirmation matters: acting on an unconfirmed double top fails most of the time, but acting only after the breakout completes cuts the failure rate to 11%.
The more a system is optimized or curve-fitted, the less reliable it will be in the future.Charles D. Kirkpatrick II and Julie Dahlquist, Technical Analysis
Where it fails
- It is a survey, not a strategy, and it says so. The book covers over 20 distinct methods (Dow Theory, sentiment, breadth, moving averages, a dozen chart pattern types, cycles, Elliott wave, Gann, point-and-figure) without ranking which combination is worth trading. A reader who wants a finished, tested system has to build it themselves from parts the book only individually validates; the authors' own conclusion on chart patterns is that Bulkowski's statistics are 'relative only' and 'cannot be assumed to be profitable in the future.'
- Elliott Wave and Gann get a full chapter with none of the statistical scrutiny given to chart patterns. Chapter 20 lays out Elliott's six 'inviolate' rules and the golden ratio's Fibonacci math in detail, but unlike the bar chart pattern chapters, which lean on Bulkowski's 12,385-pattern dataset with failure rates for every formation, Elliott wave and Gann's methods are presented with no comparable sample size or failure-rate data. The book itself acknowledges elsewhere that pattern interpretation is subjective and that 'misinterpreting any of the waves' between long-term and short-term counts is a real risk, without ever quantifying how often that misinterpretation happens.
- The pivot point formula the book documents does not work in its own example. Chapter 13 walks through the standard pivot point calculation (previous high, low, close averaged, then offset for R1/S1/R2/S2) and shows it applied to three days of Dow Jones intraday data in Figure 13.5. The book's own conclusion from that chart: 'the pivot points did not provide much guidance,' and it adds that when several alternative pivot formulas (Woodie's, Camarilla, DeMark's) are compared, 'not one seems to be consistent.'
- The Random Walk and Efficient Markets rebuttal chapter argues from academic citations, not from the book's own trading results. Chapter 4 builds its entire case for technical analysis's legitimacy on outside academic papers (Lo and MacKinlay 1988, Park and Irwin 2003, Sornette 2003) rather than on any system tested inside the book. That is intellectually honest, but it means a reader gets a philosophical defense of the field, not evidence that the specific rules taught in Chapters 12 through 20 have live track records.
- It assumes a reader who already has charting software, historical data, and screen time, none of which the book supplies. Every example chart is built in TradeStation with named tickers (JNJ, DIS, TYX, X) and specific date ranges the reader cannot reproduce without a paid data feed and platform. The system-testing chapter assumes access to enough clean historical data to run an in-sample/out-of-sample split, which is a nontrivial cost and setup barrier the book never addresses.
- Beta and standard deviation, two of the book's three named volatility filters, are described as flawed in the same paragraph that introduces them. The book states beta 'has diminished' as a valid risk measure and doesn't work for commodities, and that standard deviation of price is compromised because it includes the security's trend rather than just volatility around it. It recommends ATR instead, but readers who only skim the volatility section could walk away using the two methods the authors themselves discount.
Who it is for
Buy it if
You want one reference that maps the entire field of technical analysis, complete with real formulas, named academic studies, and Bulkowski's actual failure-rate statistics for each classic chart pattern, and you are willing to do the testing work yourself before trading any of it live. It suits a reader building a system from scratch who wants the raw materials and the discipline framework, not a finished answer.
Skip it if
You want a single validated strategy handed to you with a track record, or you are looking for a short read. At 700 pages covering more than 20 distinct methods with no ranking of which combination actually produces an edge, this is a multi-week study project, and several chapters (Elliott, Gann, pivot points) get equal shelf space to methods the book's own data suggests are weaker.
As an Amazon Associate I earn from qualifying purchases. It costs you nothing extra, and it does not change what gets recommended or what the verdict says.