What TrendSpider auto-detects
TrendSpider's pattern recognition engine flags these patterns automatically, in real time, on any chart:
- Head and Shoulders (and inverse)
- Double / triple tops and bottoms
- Bull and bear flags
- Pennants
- Ascending, descending, symmetrical triangles
- Cup and handle
- Falling and rising wedges
- Channels (ascending, descending, horizontal)
Each detected pattern gets a confidence score (0-100) based on how cleanly the price action matches the textbook structure. Above 70 = strong; below 50 = noise.
How the detection works (in 60 seconds)
Behind the scenes, TrendSpider uses ML models trained on millions of historical charts. The model identifies:
1. Pivot points — significant local highs and lows
2. Geometric relationships — distances, angles, retracements between pivots
3. Volume confirmation — does volume match the expected pattern signature?
4. Time symmetry — patterns that took 20+ bars to form are stronger than 5-bar patterns
The output is a confidence score, not a binary detection. That's actually critical — it lets you filter for high-quality setups instead of trading every flag the platform sees.
Where to find pattern detection in TrendSpider
On any chart:
1. Right toolbar → toggle on "Chart Pattern Recognition"
2. Patterns appear as semi-transparent overlays with a label and confidence
3. Click a pattern to see the projected target (typical pattern measurement) and historical reliability for that pattern type on that ticker
You can also build a pattern scanner:
- Scanner → New
- Filter: "Pattern detected = Bull flag" + "Confidence > 75"
- Run on your watchlist or a broad universe
This wakes you up to a curated list of high-conviction patterns every morning.
How we trade auto-detected patterns
Not every pattern is tradable. Our filters before taking a trade:
1. Confidence > 75
Below 75 the false-positive rate is too high. Period.
2. Volume confirmation
The pattern's breakout (or breakdown) bar should print at > 1.2x the 30-day average volume. Without volume, patterns fail more often than they succeed.
3. Multi-timeframe alignment
Daily flag is much stronger if the weekly chart is also trending up. Use TrendSpider's multi-timeframe scanner to filter for this.
4. Stop placement
Place stops at the pattern's invalidation level (e.g., the swing low for a bull flag). TrendSpider draws this for you. If your stop is too wide for your risk per trade, skip the trade.
5. Target = pattern measurement
The classic "measured move" — the height of the pattern projected from the breakout point. TrendSpider draws this automatically. Don't override it without good reason.
Common patterns and their realistic reliability
Across our backtests on the S&P 500, 5-year window:
| Pattern | Win rate | Avg R-multiple |
|---|---|---|
| Bull flag | 62% | 1.4 |
| Cup and handle | 58% | 1.7 |
| Ascending triangle | 55% | 1.3 |
| Inverse head and shoulders | 53% | 1.6 |
| Double bottom | 51% | 1.5 |
| Falling wedge | 50% | 1.4 |
Take these as ranges, not exact numbers. Your universe, your filters, and your discretion will shift them ±10%.
Pattern recognition vs. discretionary chart reading
Auto-detection isn't a replacement for learning to read charts yourself — it's a filter and a confirmation tool. The traders who get the most value:
1. Already know what each pattern looks like and how to trade it
2. Use auto-detection to scan large universes for candidates
3. Apply discretionary judgment on top of the pattern flag
Traders who blindly trade every pattern TrendSpider flags will lose money. Traders who use the platform to surface high-quality setups they'd otherwise miss will compound an edge.
Plan requirement
Pattern recognition is on the Elite plan and above. With code SET25, that's $40.50/mo billed annually — and pattern recognition alone can save you 30-60 minutes a day of manual chart-flipping.
Start the 7-day free trial and run pattern recognition on your watchlist for a week. You'll see immediately whether it surfaces tradeable setups for your style.