Most TSI misjudgments do not come from calculation errors, but from users treating "signal line crossovers" as independent entry signals. TSI is designed to filter noise and show smoothed momentum direction, with lag as the unavoidable tradeoff. TA-Lib documentation clearly states that TSI output does not include a signal line, which users need to generate manually with EMA, since "there is no universal consensus on the signal line period". This means TSI signal lines you see on different platforms may not be the same at all, and the delay brought by smoothing makes crossover signals appear only after the price has already moved a certain distance.
A leading global cryptocurrency platform,suitable for both beginners and experienced traders.
New user benefit: 20% off trading fees upon registration!！
Calculation: The double smoothing order cannot be reversed
The TSI formula is 100 × EMA(EMA(price change, first period), second period) / EMA(EMA(|price change|, first period), second period). When William Blau proposed this indicator in 1991, the default first period was 25, and the default second period was 13. TA-Lib specifically reminds that the order of the two EMA operations cannot be swapped, because each EMA is initialized with the simple moving average of its own input. Smoothing with the longer period first then the shorter one gives a completely different sequence than the reverse order.
The numerator is the directional price change (today's close minus yesterday's close), and the denominator is the absolute value of the change. Both go through the same double smoothing process, so TSI is constrained between -100 and +100. If the price does not change at all for a period of time, both numerator and denominator are zero. TA-Lib handles this case by directly outputting 0 instead of throwing an error. If you write the TSI code yourself, you need to explicitly handle this edge condition.
There is an easily overlooked implementation difference: TA-Lib's default parameter order is "25 first then 13", but some platforms or scripts may name the parameters "fast line/slow line" and apply them in reverse order. The TSI curve shape you see may look similar, but the values near turning points and crossover timings will differ.
Signal Line Crossover: Most Frequent, and Most Prone to Misjudgment
The most widely cited TSI trading signal is the TSI line crossing its signal line. Investopedia's description notes that a TSI line crossing above the signal line can be treated as a buy signal, and crossing below as a sell signal. But the same article immediately points out that signal line crossovers occur very frequently, and should only be used as part of a combined judgment, not as a standalone signal.
StockCharts' ChartSchool puts this more directly: signal line crossovers are "the most common signals, which means you will get good ones, bad ones, and ugly ones". It suggests raising the TSI or price chart parameters to reduce the number of signals and filter noise. Third-party trading resources also reach the same conclusion: double smoothing reduces some noise, but TSI tends to generate false crossovers in sideways ranging markets.
A practical filtering rule comes from Quantum Algo's trading guide: if ADX is below 25 and falling, treat TSI crossovers as noise instead of valid signals. The guide claims that when ADX is below 25, the failure rate of TSI crossovers is around 68%. This figure comes from third-party backtests, not official statistics, but the logic it points out is reasonable: TSI is a momentum tool, and in non-trending markets, momentum oscillator crossovers do not have sustained directional follow-through.
Zero Line Crossover: "Purer" Than Signal Line Crossover, But More Lagged
StockCharts calls zero line crossovers the "purest signal". A TSI value above zero means the double-smoothed price change is positive, while a value below zero means it is negative. Some traders use the zero line to determine directional bias: only consider long positions when TSI is positive, and only consider short positions when TSI is negative.
The problem is that double smoothing makes zero line crossovers arrive very late. Quantum Algo's guide estimates that a zero line crossover at the end of a long-term trend may appear when the price is 5% below the peak. This is not an exact official number, but it describes the inherent cost of double smoothing: when TSI confirms a trend reversal, the price has already moved a significant distance.
If you use zero line crossovers as "direction confirmation", they are very useful; if you treat them as "entry timing signals", they are almost always lagged. A more reasonable usage is: use the zero line to determine which direction you should trade in, then use signal line crossovers or price structure patterns to find entry points in that confirmed direction.
Divergence: The Most Valuable TSI Feature, And The Easiest To Misread
Divergence is the most widely discussed TSI signal. Investopedia's definition is: when price hits a new high but TSI does not hit a new high, it is bearish divergence; when price hits a new low but TSI does not hit a new low, it is bullish divergence. But the same article immediately notes that divergence is a poor timing signal, and should only be used in combination with other signals.
Quantum Algo's guide gives a more specific rule: divergence is the "earliest reading" of TSI, which leads zero line crossovers, but the zero line crossover is the required confirmation. This means if you enter a trade as soon as you see divergence, you are trading without confirmation; if you wait for the zero line crossover to confirm, the confirmation arrives, but the price may have already reversed for a large move.
A commonly overlooked detail is that divergence can persist for a long time while price continues moving in its original direction. In a strong trend, TSI can keep showing bearish divergence while the price keeps rising. If you take a reverse position every time you see divergence, you will get stopped out repeatedly.
A Verifiable Check Workflow You Can Follow
If you get a TSI signal and are not sure if it is reliable, follow these three steps in order.
Check parameter order. Confirm whether the TSI on your used platform uses the "long period first then short period" or "short period first then long period" order. TA-Lib's default is 25 then 13, but some TradingView scripts may use 13 then 25, or completely different periods. Different parameter orders will change the exact candlestick where the crossover occurs. If signals you see on two platforms are inconsistent, check this first.
Check current market status. If ADX is below 25 and falling, or the price is in an obvious sideways range, TSI signal line crossovers have very low reference value. In ranging markets, TSI may repeatedly cross the signal line or zero line without bringing any sustained price movement. In this case, either do not use TSI for directional judgment, or treat it as a mean reversion tool (trade in reverse direction at extreme levels) instead of a trend following tool.
Fix your divergence definition rule. Do not casually label a pattern on the chart as "divergence" at first glance. Recheck it with the same consistent set of rules: how do you define price highs (close price or highest high?), how do you define TSI highs (main line or signal line?), how many candlesticks do you look back at. If the divergence disappears when you switch to another reasonable definition, this divergence is not robust. Quantum Algo's guide treats divergence as TSI's "strongest signal", but also acknowledges it needs confirmation from zero line crossovers. Treating divergence as a warning, and zero line crossover or price structure breakout as the action condition, is far more practical than relying solely on divergence.
A leading global cryptocurrency platform,suitable for both beginners and experienced traders.
New user benefit: 20% off trading fees upon registration!！
References
- TA-Lib·True Strength Index (TSI), page publish/update date: 2026-10-08; verification date: 2026-10-19.
- Investopedia·True Strength Index (TSI): Formula, Calculations, and Interpretation, page publish/update date: 2005-06-14; verification date: 2026-10-19.
- Gate Academy·What is TSI Indicator? How True Strength Index Measures Momentum, page publish/update date: 2026-08-02; verification date: 2026-10-19.
- StockCharts ChartSchool·True Strength Index, page publish/update date: 2026-09-13; verification date: 2026-10-19.
- NexusFi Academy·True Strength Index (TSI): The Double-Smoothed Momentum Oscillator Built for Futures Traders, page publish/update date: 2026-05-26; verification date: 2026-10-19.
- Quantum Algo·True Strength Index (TSI): Formula, Settings & Signals, page publish/update date: 2026-09-23; verification date: 2026-10-19.
