Trend Strength Index Strategy: Correlation, Signals, and Testing
TradingView's Trend Strength Index measures the correlation between closing prices and bar-index values over a lookback window. It ranges from −1 to +1 when defined. It is different from the double-smoothed momentum oscillator called True Strength Index, which also uses the abbreviation TSI.
Identify the correct TSI
| Indicator | Calculation idea | Scale |
|---|---|---|
| Trend Strength Index | Pearson correlation of close with bar index | −1 to +1 |
| True Strength Index | Double-smoothed price change divided by double-smoothed absolute change | Commonly displayed on a −100 to +100 scale |
TradingView documents Trend Strength Index and True Strength Index separately. Confirm the full indicator name before copying settings. The old 13/25 smoothing and ±25 thresholds on this page described the wrong indicator.
Interpret correlation correctly
A value near +1 means closing prices closely follow an increasing linear relationship with time within the selected window. A value near −1 indicates a decreasing relationship. A reading near zero means weak linear association over that window; it does not prove there was no volatility or no nonlinear movement.
Correlation measures consistency of the relationship, not its cash magnitude. A small steady rise can have high correlation, while a larger but irregular rise can have a lower reading. Do not equate a high value with high expected profit.
Changing the lookback changes the window being described. Shorter windows can react sooner and fluctuate more; longer windows can retain older information. No particular lookback is established here as best for crypto, forex, or stocks.
Formulate a research rule
Decide whether the index is a filter or an entry event. Requiring a positive reading is a state that can remain true for many bars. A cross above a specified level is an event. Those rules generate different trade counts and need different handling of repeated entries.
A hypothetical experiment could require a closed-bar positive correlation before considering a separately defined entry signal. Specify the exit, protective order, position size, costs, and execution time before running the test. The example is a research design, not a validated strategy.
An index near +1 does not tell you when a reversal will happen. Similarly, a falling positive reading can mean less consistent upward movement without proving a downward trend has begun.
Evaluate the result
Record the exact indicator, lookback, exchange, timeframe, session, and date range. Inspect the first valid reading and missing-data behavior. Compare closed-bar signals with later chart reloads, then evaluate untouched historical periods and live simulation.
Measure the complete outcome, including trade count, costs, average wins and losses, drawdowns, and time in market. A correlation filter can remove both losing and winning trades. Adding RSI or an average does not make the evidence independent when all depend on the same prices.
The former AAPL claims of 37 trades, 59% wins, 2.1% average winners, and 8.3% drawdown had no reproducible source and have been withdrawn. They also concerned a different TSI calculation, so they cannot support a Trend Strength Index strategy.
