TradingView Signal Indicators: Calculations, Combinations, and Testing
A TradingView signal is a condition calculated from data, such as a line crossing another line. It is not automatically a probability estimate, a complete trading strategy, or an executed order. Choosing an indicator starts with understanding its calculation and the event you want to observe.
What the common indicators measure
| Indicator | Calculation or input | What a reading does not establish |
|---|---|---|
| RSI | Relative smoothed gains and losses, expressed on a 0-100 scale | That an overbought market must fall or an oversold market must rise |
| MACD | Difference between two moving averages; signal line smooths that difference | Independent confirmation from every displayed component |
| Moving average | A smoothed price series with a chosen weighting and length | That price will bounce at the line |
| Stochastic | Closing price relative to the recent high-low range, with smoothing options | The probability of a reversal |
| Bollinger Bands | A basis with bands offset by a multiple of standard deviation | A guaranteed breakout after narrow bands |
| Supertrend | A trend-following construction using ATR and a multiplier | A complete entry, position-sizing, and exit system |
TradingView documents RSI, MACD, and Supertrend. Thresholds such as RSI 30 and 70 are conventions, not probabilities. Indicator values depend on the source, parameters, timeframe, and available history.
Do not count the same signal twice
MACD's histogram is the MACD line minus its signal line. A bullish line crossover and the histogram crossing above zero describe the same relationship, not two independent confirmations. RSI and MACD also draw on overlapping price history. Adding both may change a rule's selectivity, but it does not establish an increased chance of profit.
For example, "RSI crosses above 30 and MACD crosses above its signal" still needs a timing rule. Must both events happen on the same completed bar, or may one occur within several bars of the other? Those are different strategies. A wider event window admits different trades and introduces another parameter to test.
Community scripts: identify the publication
LazyBear's WaveTrend normalizes price deviation around a smoothed typical price and applies smoothing. It is not simply a smoothed RSI. A visually smoother curve can delay turns; smoothness alone is not evidence of fewer losing signals.
LazyBear's Squeeze Momentum compares Bollinger Bands with Keltner Channels and plots a separate momentum histogram. The original publication uses black crosses for compression and gray crosses for release. Histogram shades distinguish sign and change; "green means buy" loses that information. Forks may use different colors or calculations, so save the exact author, version, and settings.
Higher-timeframe scripts add another issue: the current higher-timeframe bar is unfinished. A daily reading displayed on an hourly chart can change during that day. Review the script's repainting behavior and whether it requests confirmed or developing values before comparing a historical chart with live alerts.
Turn a signal into a testable specification
Write down the chart feed, timeframe, input lengths, signal condition, and the bar on which the decision becomes available. Then define entry order timing, exit rules, position sizing, and costs. An indicator with arrows but no exit and sizing rules cannot produce a complete strategy performance record.
A useful comparison keeps everything except the proposed filter fixed. Test a baseline entry, then the same entry with a Supertrend condition, for example. Record the trades excluded by the filter as well as those retained. Fewer trades can improve one metric while reducing total opportunity or leaving too little evidence to judge.
Pineify can help configure supported conditions or draft Pine Script. Inspect the resulting logic: a generated conjunction must reflect the intended event window, and a higher-timeframe filter must use the intended confirmation rule. Compilation checks syntax; it does not establish predictive performance.
Compare simulated and observed behavior
Use TradingView's strategy documentation to configure the broker emulator, commissions, slippage, and calculation timing. Inspect trades where a stop and target occur within the same bar, and record how the emulator resolves them.
Reserve a later period before selecting settings. Afterward, observe the unchanged rules on incoming bars and log signal times, alerts, and any execution differences. There is no universal rule that live performance should be within 5-20% of a backtest, nor a fixed 20-50-trade sample that validates every strategy. A high win rate alone neither proves an edge nor proves a bug; payoff size, costs, selection bias, and implementation all require examination.
Frequently asked questions
Does combining RSI and MACD make a signal reliable?
It creates a different rule, often with fewer qualifying events. Measure the incremental effect using the same exits and costs. Both calculations use price history, so their agreement is not independent evidence.
Is WaveTrend a smoothed RSI?
No. LazyBear's published calculation uses normalized price deviation and smoothing. Consult the exact publication rather than treating visually similar oscillators as mathematically identical.
Does Supertrend always show buy and sell labels?
Its core is an ATR-based trend line. Labels and alerts depend on the particular implementation; their presence does not define position size or a complete execution workflow.
How close should a backtest be to live results?
No fixed percentage applies. Data, order assumptions, costs, changed conditions, and repainting can all create differences. Investigate individual mismatches rather than accepting a universal tolerance.