Adam Grimes: Testing Chart Patterns, Trading Edge and Risk

A breakout clears resistance, looks convincing for two sessions, then falls back into the range. The losing trade alone tells you very little about whether the setup was worth taking. Adam Grimes’s work puts a more useful question behind that familiar chart: what evidence supported the trade before its outcome was known?

His writing combines discretionary chart reading with statistical research. That combination matters because visual judgement can recognise structure while also finding patterns in noise. A tidy chart, a good explanation and a profitable result can occur together by chance. The harder task is defining a repeatable decision and measuring what happens across its successes and failures.

Grimes wrote The Art & Science of Technical Analysis, published in 2012, and later produced a separate course workbook. The useful thread through his work is a demand for precision: a claim about a setup needs a definition, a comparison and some account of the risk involved. The examples below illustrate that approach. Their prices and results are hypothetical, rather than records of his trades.

Adam Grimes and the difference between a chart and a test

In his explanation of how he trades, Grimes describes the difficulty of selecting historical patterns without letting the researcher influence the answer. He also stresses comparison with the underlying asset’s behaviour. A profitable stock strategy may still have added little to the return available from simply holding that market.

Consider a hypothetical range with resistance at 50.00. One definition could require a daily close above 50.00; another could count any intraday trade above it. A session that reaches 50.20 and closes at 49.80 qualifies under only the second definition. Mixing the two after seeing the result changes the experiment.

I find this example useful because one price bar exposes the ambiguity. The phrase “a clean breakout” leaves room for judgement that two researchers may apply differently. A recorded rule makes that disagreement visible. The same discipline applies to the range length, entry timing and exit assumption.

The site guide to market structure provides vocabulary for describing the setting. That vocabulary still needs an operational definition before it can support a test. A higher low could mean a completed swing identified several bars later, or a low that merely looks promising today. Those definitions make different information available at the decision point.

A 45% win rate can leave a positive average

Suppose a hypothetical set of 100 trades contains 45 winners averaging 2R and 55 losers averaging 1R. Here, R is a fixed unit of initial planned risk, and each loss is expressed as a positive magnitude for the calculation. Gross expectancy is 0.45 × 2R minus 0.55 × 1R, or 0.35R per trade. The set loses more often than it wins.

That arithmetic explains why judging a method from its latest losing trade is weak evidence. It also shows why win rate alone can’t describe performance. If the average winner falls to 1R with the same win rate and loss size, expectancy becomes minus 0.10R. The frequency of winners hasn’t changed; the economics have.

Costs belong in the calculation. An average cost of 0.10R would reduce the first example to 0.25R. Slippage, spreads and commissions can consume a substantial share of a small gross advantage. Planned risk is also an imperfect denominator when a market gaps through an exit level: the realised loss can exceed 1R.

These figures illustrate an identity, not an estimate of a Grimes strategy. They don’t establish that the next 100 trades will have the same distribution. A few unusually large winners, changing market conditions or inaccurate fill assumptions can make the historical average fragile.

Separate a pullback from a failure test

Grimes’s guide to fundamental trading patterns distinguishes a simple pullback, with one counter-trend leg, from a complex pullback with two or more. It also describes a failure test: price crosses an earlier pivot and then quickly reverses. These are different structures, even when a finished chart makes them look equally obvious.

For illustration, a market advances from 40.00 to 44.00 and retreats to 42.50. An attempted recovery to 43.50 followed by another decline to 42.20 adds a second counter-trend leg. Calling the whole sequence a simple pullback hides the failed first attempt to resume the advance. A study that keeps those cases separate can ask whether their subsequent behaviour differs.

A failure test poses another question. If an earlier high is 44.00, price reaches 44.20 and then returns below 44.00, the return through the old level is observable. The chart doesn’t reveal the identity or intentions of everyone trading there. Describing it as certain institutional distribution adds a claim the price sequence can’t establish.

The support and resistance framework helps locate that reference point. It supplies no universal probability that the level will hold. Even a clearly defined failure test needs evidence about its frequency, exits and losses in the market being studied.

Timeframe agreement needs its own evidence

A daily setup can point down while a weekly chart still suggests a possible recovery. It’s tempting to resolve that conflict by treating the longer timeframe as automatically superior. Grimes’s 2015 discussion of trading against his bias is more cautious. He says he had been unable to prove that aligning multiple timeframes increased the probability of his trades, and explains the need for a predefined response to conflict.

That distinction matters. A researcher could label each daily signal by whether a weekly filter agreed with it, then compare the two groups using the same exits and cost assumptions. The result might support the filter, undermine it or remain inconclusive. Assuming agreement helps before making the comparison builds the desired answer into the process.

The timing of the information matters too. A weekly close is available only after that week finishes. Using Friday’s completed weekly bar to classify a Tuesday entry would give the test information the trader didn’t have. A filter can look useful simply because its implementation looks ahead.

I separate “additional context” from “demonstrated improvement” for that reason. The weekly chart adds information, but the Tuesday-versus-Friday example shows how easily the apparent benefit can come from the way a test was assembled.

Record rule changes before judging their value

A trading journal can preserve the distinction between the planned method and the decisions actually made. In a hypothetical study, a rule might call for entering after a qualifying daily close. Entering early during the session is a different action, even if the trade eventually profits. Recording both the signal and the deviation makes that difference inspectable.

A useful comparison would retain the signal date, qualifying conditions, planned entry, actual fill, exit and costs. If an early entry improves the result, that observation can become a research question. If the journal records only profitable exceptions, it can’t support a fair comparison.

The Ed Seykota profile offers a related perspective on consistency and risk. A rule can be followed faithfully and still lose. Equally, a departure from a rule can make money once without improving the process. Keeping those possibilities separate makes the review more informative than a list of wins and losses.

There’s also a limit to what the comparison proves. A historical simulation and a live record may cover different markets, volatility and liquidity. Their difference can’t automatically be assigned to psychology. Data quality, changing conditions and execution assumptions deserve examination alongside behaviour.

Read the drawdown beside the return

Two hypothetical accounts finish a year up 40% and 15%. Their largest peak-to-trough declines are 35% and 8%, respectively. The first return is larger, but those four numbers alone don’t establish which strategy is better. The time spent underwater, exposure, financing costs and repeatability all matter.

The recovery arithmetic makes the drawdown tangible. An account falling from 100 to 65 needs a gain of about 53.8% to return to 100. An account falling from 100 to 92 needs about 8.7%. Those percentages describe recovery from the reduced balance; they aren’t forecasts.

The Ulcer Index adds another view by summarising percentage drawdowns over a chosen observation window. Repeated observations below a prior peak affect the measure, so the path matters alongside the deepest loss. A single return number leaves that experience hidden.

A short, favourable history can also understate risk. The worst drawdown observed so far places no ceiling on a future decline. This is especially relevant when several apparently separate trades depend on the same market move. Counting positions doesn’t measure the independence of their risks.

Two books with different jobs

The Art & Science of Technical Analysis: Market Structure, Price Action & Trading Strategies is the 2012 book. The verified edition linked here is hardcover. It suits readers looking for Grimes’s broader treatment of chart structure, trading methods and risk.

Disclosure: As an Amazon Associate, I earn from qualifying purchases.

(US) The Art & Science of Technical Analysis
(UK) The Art & Science of Technical Analysis
(DE) The Art & Science of Technical Analysis

The Art and Science of Trading: Course Workbook is a separate 2017 paperback. Grimes’s workbook FAQ explains that it contains exercises and supports his course; it isn’t a second edition of the earlier book. That distinction helps avoid purchasing one while expecting the other.

(US) The Art and Science of Trading: Course Workbook
(UK) The Art and Science of Trading: Course Workbook
(DE) The Art and Science of Trading: Course Workbook

Make the next chart claim specific enough to check

A journal entry saying “the breakout failed” leaves much unresolved. Recording that price closed above 50.00, returned below it the next session and produced a defined result under a stated exit rule creates something that can be compared with other cases. Grimes’s insistence on evidence becomes practical at that level of detail.

The discipline is accepting what the record can support. A profitable example can suggest a question. A consistent test can examine it. Neither removes uncertainty from the next trade.

Learn the pattern. Ride the trend. Keep the gains.

Educational content only. Not investment advice. Trading involves risk. You are responsible for your decisions.