Development of a Support/Resistance Level System
Algorithmic determination of support and resistance (S/R) levels is a challenge every trader handling 50+ instruments faces. Manual markup takes up to 20 hours per week, and a missed breakout on ETHUSD costs between $500 and $2000 per trade. We built a system that automatically identifies significant levels with 75%+ accuracy and reduces markup time by 80%. This saves $3000 to $8000 per month on manual labor (depending on the number of instruments) and further reduces losses from false breakouts by 25%.
Our experience of 5+ years in trading systems development guarantees robust performance. The system is tested on 50+ crypto pairs and shows 4x faster level detection compared to manual methods.
Algorithmic S/R Determination
How Are Pivot Points Clustered?
The main method is pivot point clustering. The algorithm:
- Find all pivot highs/lows over a rolling window (default 50 bars)
- Group close prices into clusters (tolerance ±0.5%)
- Evaluate each level's strength: touch count, volume, recency
from scipy.cluster.hierarchy import linkage, fcluster def cluster_levels(price_points, tolerance=0.005): prices = np.array([p[1] for p in price_points]).reshape(-1, 1) Z = linkage(prices, method='single') labels = fcluster(Z, t=tolerance * np.mean(prices), criterion='distance') clusters = {} for i, label in enumerate(labels): clusters.setdefault(label, []).append(price_points[i]) return {k: np.mean([p[1] for p in v]) for k, v in clusters.items()} Clustering filters noise: instead of hundreds of useless lines, you get up to 20–30 significant ones. In our ETHUSD backtest, signal accuracy is 25% higher than with simple rolling max/min.
Volume Profile and Round Numbers
Additional sources:
- Volume Profile / Point of Control (POC) — the price with maximum volume over a period. More details on Volume Profile.
- Round numbers (50000, 100000 for BTC) — psychological levels with a base weight.
- Bill Williams fractals — stricter local extremes.
Level Strength Scoring and Filtering
Each level receives a score based on several metrics:
| Metric | Description | Weight |
|---|---|---|
| Touch count | How many times price tested the level | High |
| Volume at level | Volume near the level | High |
| Recency | How recently the level is relevant | Medium |
| Bounce strength | Strength of the bounce | Medium |
| Timeframe confluence | Visibility across multiple timeframes | High |
A level visible on daily and 4h receives a 40% higher score — a key filtering factor. Additionally, we filter by volume: if a level has less than 5% of average session volume, it is ignored.
Dynamic Levels and Breakouts
Breakout Handling and Polarity Change
When a candle closes outside the zone (±ATR/4), the level is marked as broken. Support becomes resistance and vice versa. Status updates in real time via WebSocket — critical for trading bots.
Dynamic Levels
Statics are complemented by dynamic ones:
- EMA 20/50/200
- Bollinger Bands (upper/lower band as S/R)
- VWAP and anchored VWAP
For example, daily EMA 200 is a powerful support level for BTCUSDT.
Practical Setup and Visualization
Configuring the Level System
- Connect the data source (REST/WebSocket) — exchange API or broker.
- Specify the instrument list (e.g., 50 crypto pairs).
- Adjust parameters: clustering tolerance, score threshold, timeframes.
- Run a backtest on 6 months of historical data.
- Deploy the core on a server (Python) and connect the frontend.
- Receive breakout alerts via Telegram or through the UI.
Visualization
Levels are drawn as zones (rectangles) on the chart. Border thickness and color saturation are proportional to score. Zones instead of lines are more realistic: the market tests a range, not a point. Automatic update on breakout and polarity change.
Tech stack: Python (pandas, scipy, numpy), PostgreSQL, WebSocket, React + TradingView Lightweight Charts.
Timing and Cost
Development time ranges from 4 to 8 weeks depending on integration complexity. Typical cost: $8,000–$15,000 for a complete system. We offer a satisfaction guarantee: free bug fixes for 3 months post-delivery. Order a free backtest of your strategy with our system — our engineers will respond within a day. Contact us for a consultation.
What’s Included in the Work
Click to see deliverables
- **Documentation**: detailed algorithm description, API reference, configuration guide - **Access**: full source code, deployment scripts, database schema - **Training**: 2 hours of live walkthrough for your team - **Support**: 3 months of bug fixes and minor tweaks - **Optional**: integration with your existing trading platformCommon Mistakes in DIY Implementation
- Using a single method (e.g., only fractals) — low representativeness.
- No volume filter — false levels.
- Ignoring dynamic levels — missing trend moves.
- Incorrect tolerance: 0.1% gives 100+ lines, 2% gives one level.
Our system avoids these pitfalls through multi-method fusion and rigorous filtering. With 5+ years of experience and a proven track record on 50+ instruments, we deliver a robust solution. The method is based on the Volume Profile approach by J. Dalton (Mind Over Markets, 1991).