True Strength Index (TSI): What the Settings Actually Do
The True Strength Index turns price changes into a reading that cannot leave a fixed range. Where inside that range a trader should start paying attention depends entirely on two smoothing periods, and no platform vendor sets those periods for you. The explanations in circulation state them as though the question were settled. They are not settled, and the pages that state them do not agree with each other.
Key takeaways
- The reading is a ratio: smoothed price change divided by smoothed absolute price change, scaled by one hundred.
- The limit of plus and minus one hundred is arithmetic. It follows from that ratio and needs no source to hold.
- The often repeated claim that most readings sit between plus and minus twenty-five is not arithmetic. It describes one pair of smoothing periods and stops being true when either number changes.
- The thinkorswim study reference names three period inputs and publishes no default value for any of them, while the two explainer pages that do give numbers give different ones.
- The MQL5 reference lists the indicator functions MetaTrader provides, and no True Strength Index is among them.
Table of contents
What the True Strength Index Measures
The calculation starts with the change in price from one bar to the next. That signed series is smoothed twice. The same double smoothing is then applied to the absolute value of the same series, and the reading is the first result divided by the second, multiplied by one hundred.
Two things follow from that shape. The numerator keeps direction, so a run of higher closes leaves it positive and a run of lower closes leaves it negative. The denominator throws direction away and keeps only size, so it grows with movement of any kind.
The ratio therefore asks what share of recent movement pointed one way. It is momentum expressed as a proportion of activity, not as a distance in price. A reading of forty on a quiet instrument and a reading of forty on a violent one describe the same balance between directional and total movement, which is why the number is comparable across instruments in a way that a raw price change is not.
The thinkorswim study reference describes the same construction as a double smoothing and names the two lengths that control it, without setting either one.
Why the Reading Cannot Leave Plus or Minus 100
The limit is not a clamp applied after the calculation, and it is not a convention borrowed from other oscillators. It falls out of the ratio itself.
Each smoothing step is a weighted average whose weights are positive and add up to one. Apply such an average to a series that contains both positive and negative values, and the opposing values cancel. Apply the identical average to the absolute values of that same series, and nothing cancels, because every term is already positive. The second result can therefore never be smaller than the size of the first.
Divide the first by the second and the answer sits between minus one and plus one. Scale it by one hundred and the reading sits between minus one hundred and plus one hundred. The encyclopedia entry states this bound and cites nothing for it; the arithmetic holds it up regardless.
Knowing why the limit exists also tells a reader what it does not mean. The extremes are reached only when nearly every recent bar moved the same way, which is a description of a clean trend rather than of an exhausted one. Fixed thresholds inside the range are a separate question, and the general problem with them is set out on our page about overbought and oversold levels.
The Typical Range Is Not a Property of the Indicator
The same encyclopedia entry that states the bound also states that most readings fall between plus and minus twenty-five, and cites nothing for that either. The two statements look alike and are not the same kind of statement at all.
The bound is arithmetic, as the previous section shows. It holds for every instrument, every timeframe and every pair of smoothing periods anyone could choose. The narrower range is an observation, and an observation made on one series smoothed one particular way.
Change either period and the distribution moves. Longer smoothing lets more opposing bars into each average, so more cancellation occurs in the numerator while the denominator keeps accumulating, and readings compress toward zero. Shorter smoothing does the reverse: fewer bars enter each average, less cancels, and readings spread out toward the bound.
So a sentence about where most readings sit is a sentence about settings. It is not a fact about the True Strength Index, and it stops being true the moment a reader changes either number. None of the pages consulted for this article separates the two claims, and several present both in the same breath.
The practical consequence is direct. A threshold copied from an explanation whose settings you did not check is a threshold measured on somebody else’s chart. Before treating any level as meaningful, read the two periods off your own indicator and accept that the level has to be established there.
No Vendor Publishes a Default Period Pair
The obvious way to settle the question is to ask the platform. That does not work here, because the platform declines to answer.
The thinkorswim study reference lists three period inputs for this study, called long length, short length and signal length, and describes what each one controls. It publishes no default value for any of the three. It also exposes the type of average as a further input, which means two charts agreeing on all three periods can still draw different lines.
The pages that do give numbers do not agree. The encyclopedia entry gives twenty-five and thirteen and describes them as typical, with no citation attached. A forex education page read for this article runs five and eight instead, also with nothing behind it.
| Source consulted | Period values stated | Citation given |
|---|---|---|
| Platform study reference | None; three inputs named only | Not applicable |
| Encyclopedia entry | Twenty-five and thirteen, called typical | None |
| Forex education page | Five and eight | None |
| Charting service learning page | No numeric values captured | None |
Because no official document sets these numbers, none of them appears on this page as a recommendation. The averaging method is a variable in its own right, and what changes when a platform lets you switch it is worked through on our page about the smoothing method a platform applies.
Where TSI Is Not Built In
The MQL5 reference enumerates the technical indicator functions MetaTrader provides, from the accelerator oscillator through to the volumes indicator. That list includes the relative vigor index, the force index, the Chaikin oscillator and the variable index dynamic average. There is no True Strength Index function in it.
The consequence is narrow and worth stating precisely. A trader who wants this indicator on MetaTrader installs something written by a third party, and the periods, the averaging type and the signal line in that copy are decisions made by its author rather than by the platform. Since no vendor publishes defaults in the first place, there is nothing to check the installed copy against.
What that situation does to a reader more generally, and how to work with a tool the platform does not document, is covered on our page about an indicator that exists only as a third-party build. The point here is confined to this one absence.
Reading the Signal Line Without Overreading It
The thinkorswim reference describes crossovers between the main plot and its own average. That average is a third smoothing applied to a series that has already been smoothed twice.
Each smoothing step delays the response to a change in price, and three of them in sequence delay it more than one. A crossover is therefore a statement about what already happened, and the lag grows with every period a trader lengthens.
This is the trade a reader is actually making when choosing periods. Short settings produce more crossovers, earlier, and more of them reverse. Long settings produce fewer and later, and each one carries more of the move behind it. Neither is correct, and no source consulted here offers evidence for a preferred pair, which is consistent with the absence of any published default. Where a platform does decide the signal line for you, as it does for the Relative Vigor Index, the weighting is fixed and documented instead.
Who Should Skip This Indicator
Anyone who wants a tool with a checkable specification should look elsewhere, because the periods that decide every reading are set by whoever wrote the copy on the chart. Anyone intending to apply a threshold read from an article, without confirming the settings behind it, will be reading a number that means something different on their own screen.
If the interest is the underlying idea rather than this particular arrangement of it, the smoothing and momentum components are each documented on their own, and our guide to how these tools fit together is the better starting point.
Three checks settle what a reading here can and cannot tell you. Open the settings and write down the two smoothing periods and the signal period actually in use.
Then check whether your platform exposes the averaging type, and record which one is selected. Last, establish where readings on that instrument, with those settings, have turned before, rather than importing a level from a page whose settings were never stated.
Risk notice. This page is educational and describes how one technical indicator is constructed and what documentation exists for it. Nothing here is a recommendation to buy or sell any instrument, no reading of any indicator is a forecast, and no figure above is presented as a result anyone should expect. Leveraged trading carries a high risk of loss.