True Strength Index: How to Read Blau’s Double-Smoothed Momentum Line

Most momentum oscillators make you choose between a line that reacts fast and a line you can actually read. The true strength index (TSI) tries to dodge that trade-off by smoothing price changes twice before it tells you anything. The result is a momentum line that often hugs the shape of the price chart itself, which is why some traders draw trend lines on it the way they would on a stock.

We’ll go through where the indicator came from, the formula in plain terms, a full worked example with numbers you can check on a calculator, and the four ways people read it. We’ll also be honest about where it misleads, because every momentum tool does.

Where the true strength index comes from

The TSI was created by William Blau. According to Wikipedia’s entry on the indicator, he first published it in 1991 in Technical Analysis of Stocks & Commodities magazine, and the entry also cites his 1995 Wiley book on momentum, direction and divergence. StockCharts’ ChartSchool describes it simply as a momentum oscillator built on a double smoothing of price changes.

The thinking behind it is neat. Raw momentum, today’s close minus yesterday’s, is the purest form of momentum you can get, but it jumps around wildly from day to day. A moving average calms it down but lags. Blau’s idea was to put the moving average on the momentum itself, then do it again, so the line stays responsive to the direction of change while shedding most of the daily noise.

Unlike the RSI, the TSI doesn’t split the day’s move into separate “up” and “down” buckets. It works with the signed change directly, and uses the absolute change only to scale the result. That’s a small design choice with a big effect on how the line behaves, as we’ll see.

The TSI formula, step by step

The calculation has three parts. Using the defaults that both StockCharts and Wikipedia quote, 25 periods for the first smoothing and 13 for the second, it goes like this:

1. Price change (PC): current close minus the previous close.
2. Double-smoothed PC: take a 25-period exponential moving average (EMA) of PC, then a 13-period EMA of that EMA.
3. Double-smoothed absolute PC: do exactly the same two smoothings on the absolute value of PC, ignoring the sign.
4. TSI: 100 × (double-smoothed PC ÷ double-smoothed absolute PC).

Most platforms also plot a signal line, which is just an EMA of the TSI itself. StockCharts’ examples use settings such as TSI(40,20,10), where the last number is the signal line period.

Look at what the division does. The top half carries the sign: if the smoothed changes are mostly negative, the TSI is negative. The bottom half is always positive and acts as a ruler. If every single day in the window were an up day, the two numbers would be equal and the TSI would sit at +100. If every day were down, it would hit −100. In practice, mixed days pull the numerator towards zero much faster than the denominator, so readings near the extremes are rare.

The TSI is really a ratio of direction to distance: how much of the market’s recent travel actually went somewhere, versus how much was just back-and-forth.

One practical point. An EMA starts from a simple average and then builds on itself, so StockCharts notes that you need extra history before the values settle to their “true” level. With 25 and 13 periods, the first few dozen TSI readings on a fresh data series aren’t fully reliable. Your charting software handles this by loading more history than it displays.

A worked example you can check by hand

Doing 25 and 13 periods by hand is tedious, so we’ll shrink the settings to a 5-period first smoothing and a 3-period second smoothing, with a 3-period signal line. The mechanics are identical; only the lag changes. Here’s a hypothetical stock over 15 sessions. These prices are made up for illustration and don’t belong to any real company.

Step 1: the first smoothing

The first 5-period EMA of PC is seeded with a simple average of the first five changes: (12 − 4 + 12 + 11 − 4) ÷ 5 = 5.4. That value belongs to day 6. From there, each new value uses the EMA multiplier, 2 ÷ (5 + 1) = 0.333. On day 7, PC is +13, so the EMA becomes 0.333 × 13 + 0.667 × 5.4 = 7.93.

The absolute series gets the same treatment. The first five absolute changes are 12, 4, 12, 11 and 4, which average to 8.6. On day 7 it becomes 0.333 × 13 + 0.667 × 8.6 = 10.07.

Carry both forward and you get, for days 6 to 15:

• First-smoothed PC: 5.40, 7.93, 9.29, 5.19, 7.46, 3.97, −0.02, −2.34, −0.23, −3.82
• First-smoothed |PC|: 8.60, 10.07, 10.71, 8.14, 9.43, 7.28, 7.52, 7.35, 6.23, 7.82

Step 2: the second smoothing

Now we run a 3-period EMA over each of those lines (multiplier 2 ÷ 4 = 0.5), again starting with a simple average of the first three values. For the signed line that’s (5.40 + 7.93 + 9.29) ÷ 3 = 7.54, landing on day 8. For the absolute line it’s (8.60 + 10.07 + 10.71) ÷ 3 = 9.79.

For days 8 to 15 the double-smoothed values come out as:

• Double-smoothed PC: 7.54, 6.37, 6.91, 5.44, 2.71, 0.18, −0.02, −1.92
• Double-smoothed |PC|: 9.79, 8.97, 9.20, 8.24, 7.88, 7.62, 6.92, 7.37

Step 3: the TSI and its signal line

Divide and multiply by 100. On day 8, that’s 100 × 7.54 ÷ 9.79 = 77.0. Running through the rest:

(Small differences in the last decimal can show up if you round at each step; we carried full precision and rounded only for display.)

What the example tells us

Several things happen in this short stretch, and they line up with the classic TSI readings.

A bearish divergence. The stock made a higher close on day 10 (₹1,061) than on day 8 (₹1,052). The TSI didn’t follow: it printed 75.2 on day 10 against 77.0 on day 8. Price pushed higher, momentum didn’t. That’s a warning, not a sell button, but in this case the stock never got back above ₹1,061 in the sessions shown.

A signal line cross. On day 10 the TSI (75.2) was just above its signal (74.4). On day 11 it dropped to 66.1, below the signal’s 70.2. That’s a bearish signal-line crossover, one session after the divergence.

A centerline cross. By day 14 the TSI slipped to −0.3, its first negative reading. Notice that day 14 was actually an up day (+₹4). The smoothing means one green candle can’t rescue the line once the recent run of changes has turned negative. On day 15 it fell to −26.1.

Look at the price, though. The stock was only about 2.4% below its day-10 high by day 15 (₹1,036 versus ₹1,061). The TSI had swung from the mid-70s to negative territory. That gap between a modest price move and a big oscillator swing is exactly why short settings are noisy, and why the defaults are much longer.

Four ways traders read the TSI

1. The centerline

StockCharts calls the zero-line crossover “the purest signal”. Above zero, the double-smoothed momentum is positive and buyers have the edge. Below zero, sellers do. Because of the heavy smoothing, these crosses come late, but they tend to mark real shifts in the trend rather than one-day blips. Wikipedia adds that the slope matters too: a rising TSI suggests an uptrend, a falling one a downtrend.

2. Signal line crossovers

This is the most frequent signal and, frankly, the least reliable on its own. StockCharts points out that on one weekly chart with TSI(40,20,10), the line crossed its signal at least 12 times between April 2007 and July 2012. Plenty of those were whipsaws. If you use crossovers, it helps to only take them in the direction of the centerline: bullish crosses while the TSI is above zero, bearish ones while it’s below. That’s essentially what StockCharts’ own suggested scans do.

3. Overbought and oversold zones

The TSI can technically run from −100 to +100, but Wikipedia notes most values fall between +25 and −25, and that Blau suggested treating those levels as overbought and oversold. StockCharts takes a more flexible view: the right levels depend on the stock’s volatility and your settings. In its example on the Nasdaq 100 ETF, a slow TSI(40,20,7) moved between roughly −20 and +44, while a fast TSI(13,7,7) swung between about −69 and +78, so the thresholds had to be set very differently for each.

An overbought TSI reading says the move has come far and fast. It doesn’t say the move is over. Wait for price to confirm.

So don’t copy a fixed ±25 band onto every chart. Pull up a year or two of history for the stock you’re watching, see where its TSI has tended to peak and trough, and set your bands from that. A sleepy utility and a volatile small-cap will look nothing alike.

4. Trend lines and support on the indicator

This is where the TSI is a little unusual. StockCharts notes that its peaks and troughs often match those in price, so you can draw trend lines, support and resistance directly on the oscillator. A break of a rising trend line on the TSI can come before the equivalent break on the price chart, giving you an early heads-up. It won’t always, but it’s a use most oscillators are too jumpy for.

Choosing settings for Indian stocks

There’s no magic set of numbers, and anyone selling one is guessing. The trade-off is the usual one, which StockCharts spells out: shorter periods give quicker signals with less lag but more false ones, while longer periods cut the whipsaws at the cost of later entries.

A sensible way to start:

• Positional trades on daily charts: the default 25 and 13 with a signal line around 7 to 13 periods is a reasonable base.
• Longer-term views on weekly charts: something slower, like the 40/20/10 combination StockCharts uses, keeps the line from flipping every few weeks.
• Intraday: shorter settings react faster, but expect a lot of noise. Pair them with a clear trend filter, such as price relative to a moving average, before acting on crossovers.

Whatever you pick, test it on the specific stock or index you trade. A setting that reads Nifty 50 nicely may be useless on a thinly traded small-cap where single block trades distort the price change.

Where the TSI lets you down

We like this indicator, but it has real weak spots.

• It lags. Two layers of EMA mean you’re always a bit late. In fast reversals, such as a sharp gap down on results, the TSI will confirm what already happened.
• Divergences can drag on. StockCharts warns that divergences can be misleading in a strong trend. A stock in a powerful rally can show a weakening TSI for weeks while price keeps climbing.
• Crossovers are noisy. As the 12-crossings example shows, the signal line alone generates many trades you’d rather not take.
• Gaps distort the math. One big overnight gap produces a huge single price change that sits in both EMAs for a while. Corporate actions like splits and bonuses also need adjusted price data, or the TSI will show a crash that never happened.

How the TSI fits with other tools

If you already use the percentage price oscillator or MACD, the TSI will feel familiar: a smoothed momentum line plus a signal line. The difference is that the TSI is scaled by the absolute size of moves, so it’s bounded and comparable over time in a way MACD’s raw values aren’t. Compared with the stochastic RSI, which is built to be very sensitive, the TSI sits at the calmer end of the spectrum.

A practical combo many traders use: let the TSI’s centerline and slope define the bias, use price structure (support, resistance, breakouts) for entries, and use a volatility-based stop. For more on how these pieces fit together, our guide to technical analysis for Indian stocks covers the basics in one place.

The short version: the true strength index is a slow, honest momentum gauge. It won’t catch tops and bottoms, but it does a good job of telling you whether the recent push has real direction behind it, and it’s one of the few oscillators clean enough to draw lines on.

Sources

• StockCharts ChartSchool: True Strength Index
• Wikipedia: True strength index

Disclaimer: This article is for information and education only and is not investment advice. The worked example uses hypothetical prices that don’t represent any real security. Technical indicators don’t guarantee future results; please consult a SEBI-registered adviser before making any investment decision.