A chart pattern becomes a strategy once you've defined, in advance, exactly which instruments and timeframe you're watching, exactly what counts as your pattern, what context it needs to form in, what confirms it, where you enter, where it's invalidated, how much you risk, where you exit, and how you'll review it afterward. Spotting a shape on a chart is observation. This full sequence, decided before you ever place a trade, is what makes it a strategy.
Recognizing a pattern is one skill. Turning that recognition into a consistent, repeatable process is a different one, and it's the one that actually determines whether pattern analysis helps you or just gives you something to point to after a trade goes wrong. This guide walks through that full process, step by step.
Define your universe and timeframe
Decide which instruments you're actually watching, a specific watchlist of liquid stocks, a specific index, a specific sector, and which timeframe you're trading on, daily, 4-hour, 1-minute, before looking for a single pattern. A companion guide on how patterns change by market and timeframe covers why this choice affects everything downstream. Pick it first, and hold it steady, rather than switching timeframes mid-analysis to make a shape fit better.
Define your pattern criteria
Write down the specific, mechanical criteria for whichever pattern or patterns your strategy uses, minimum boundary touches, minimum formation length, volume behavior, using the recognition process and playbook criteria covered in earlier guides in this series. This page doesn't redefine those patterns, it assumes you've already got objective rules for them. The point here is committing to those rules in writing, so you're checking a real shape against a real standard, not eyeballing a vague resemblance.
Apply a context filter
Only take setups that match the market condition your chosen pattern actually needs. A pullback or breakout strategy needs a genuine trend. A range or reversal strategy needs sideways conditions. Check this before anything else, since applying the wrong pattern family to the wrong condition is one of the most common, avoidable strategy failures.
Require confirmation
Decide in advance that you will not enter on an anticipated pattern, only a confirmed one, a close beyond the relevant boundary, with the volume behavior your criteria call for. Should a pattern alone trigger a trade? No. The shape is one input. Confirmation, that the shape actually resolved the way its criteria require, is what turns an observation into an actual, real-time signal.
Define entry rules
Decide, before you're in the moment, exactly how you enter once confirmation happens, at the breakout close itself, or at a retest of the broken boundary. Both are legitimate, commonly used approaches. Pick one for your strategy and apply it consistently, rather than deciding case by case depending on how a trade "feels" once you're actually watching it unfold.
Define invalidation
State, in writing, the specific price or condition that proves your setup wrong, specific to the pattern you're trading, covered per-pattern in a companion playbook guide. This is what your stop-loss actually gets built from, not a round number chosen arbitrarily, but the real point where your original reasoning for the trade no longer holds.
Manage risk
Decide your risk budget before entering, commonly 1% to 2% of total capital, and calculate your position size from it directly: position size equals your risk budget divided by your stop-loss distance, the same formula covered in full, with a worked example, in a companion guide on risk management. This is where pattern recognition actually connects to capital protection, and skipping it is how a single wrong setup can do outsized damage regardless of how good your pattern criteria are.
Define your exit
Decide your target approach in advance too, a measured-move projection, the next meaningful support or resistance level, or a trailing stop that lets a winning position run. Whichever you choose, remember it's a planning convention, not a guarantee, covered honestly in a companion guide on pattern reliability. Price frequently falls short of or exceeds any specific projected target.
Journal every setup
Record every setup you act on, and arguably more usefully, every setup that met your criteria but that you skipped, with entry, exit, reasoning, and whether you actually followed your own rules. This is what turns a handful of trades into a real, reviewable sample, rather than a string of disconnected decisions you can't learn from afterward.
Backtest and review
Test your full rule set, criteria, context filter, confirmation, entry, invalidation, against historical data before trusting it with meaningful capital, using the expectancy, drawdown, and sample-size framework covered in a companion guide on backtesting. Review your actual results periodically against that same framework, not just whether your account balance went up.
A pre-trade checklist
- Instrument and timeframe defined in advance, not chosen after spotting a shape
- Pattern criteria met, checked mechanically, not just visually
- Market context actually fits the pattern type, trend for continuation and pullback setups, range for reversal setups
- Confirmation received, an actual close beyond the boundary, not an anticipated one
- Entry rule applied consistently, breakout or retest, decided before this specific trade
- Invalidation level written down before entry, not chosen after
- Position size calculated from your risk budget and stop distance, not from how confident you feel
- Exit approach decided in advance, target or trailing stop
- Trade logged immediately, win, loss, or skipped setup alike
How the decision sequence flows
Think of this as a series of gates, each one needing to pass before you move to the next. Does the instrument and timeframe match your defined universe? If not, stop. Does a shape meet your written pattern criteria? If not, stop. Does the market context fit the pattern type? If not, stop. Has the pattern actually confirmed, not just approached confirmation? If not, wait. Only once every gate passes do you calculate position size and place the entry, with invalidation and exit already decided before that entry order goes in.
Worked example: one trade that worked, one that didn't
- Trade one: A stock has been in a clear uptrend for six weeks, meeting the context filter. It pulls back to its 20-day moving average near ₹620 and consolidates for three sessions between ₹615 and ₹630 on declining volume, meeting the pattern criteria for a flag. It closes at ₹632 on volume 1.6 times its recent average, confirmation. Entry taken at ₹632. Invalidation set at a close below ₹610, the consolidation's low. On a ₹5,00,000 account risking 1%, ₹5,000, against a ₹22 stop distance, position size works out to roughly 227 shares. Nine sessions later, price reaches a reference target near ₹690, and the position is closed for a gain.
- Trade two: A different stock, also in an uptrend, also meeting the context filter. It pulls back to ₹410 and consolidates for four sessions between ₹405 and ₹415 on declining volume, meeting the same flag criteria. It closes at ₹417 on volume 1.3 times average, a weaker confirmation than trade one, but still meeting the written criteria. Entry taken at ₹417. Invalidation set at a close below ₹398. Same 1% risk, ₹5,000, against a ₹19 stop distance, position size roughly 263 shares. Three sessions later, the stock reverses and closes at ₹396, through the invalidation level. The position exits with a loss of roughly ₹5,523, slightly more than the planned ₹5,000, since the actual fill landed a little beyond the stop level itself, a realistic outcome, not a flaw in the process.
Both trades followed the exact same rules. One worked, one didn't. That's not a contradiction, it's what a real, rules-based process actually looks like applied consistently across more than one outcome.