Tools
Calculators That Show Every Step
Three working tools that take the inputs you already have on a chart and return the levels the rest of this reference refers back to. Nothing to install, nothing to sign up for, and every intermediate value shown.
These pages exist because of a specific complaint. Every indicator has a published formula, and yet two charting platforms will routinely disagree about the same indicator on the same data, not by rounding, but visibly. The disagreement almost never lives in the formula. It lives in a smoothing constant written one way rather than another, or in how much history the calculation was given before you looked at it.
A calculator that prints only its answer cannot help with that. These print the columns: the average gain and average loss for each bar, the relative strength before it becomes an index, the range a pivot was derived from. When your platform gives a different number, the difference is in one of those rows, and it takes a minute to find rather than an afternoon.
What all three tools have in common
Every level these calculators produce is a prior. It is computed from data that is already complete, yesterday's range, a finished swing, a series of closes, and then fixed. Nothing about it adapts to what price does next, which is unusual among the methods on this site and is the source of both its usefulness and its limits.
The consequence is that the informative event is never the touch. Price reaching a pivot resistance or a 61.8 per cent retracement says only that the prior arithmetic projected that far. What the level is actually good for is telling you where to look: it marks the place at which a question will be asked, and something else (volume expanding or thinning, breadth confirming or not) answers it. A level reached on heavy volume that then holds is a different market from the same level reached on the thinnest session of the month.
This is also why the number of levels matters. Classic pivots put seven lines on a chart and Camarilla puts nine; across a session price will touch several of them, and hindsight will always find the one that "worked". The discipline that makes these tools worth using is deciding which level matters before it is reached, and then reading the reaction rather than the arrival.
Which tool answers which question
The three calculators are not alternatives. They take different inputs and produce levels that mean different things, and choosing between them is a matter of what you already have on the chart.
| Tool | Needs | Returns | Valid for |
|---|---|---|---|
| Pivot points | One completed session: high, low, close | A pivot and three levels either side, in four formulas | The next session only |
| Fibonacci | A swing high and a swing low | Five retracements and three extension targets | As long as that swing is the relevant one |
| RSI | A series of closes, ideally a hundred or more | A reading per bar, with the average gain, average loss and relative strength behind it | The bar it was computed on |
Two of the three expire. Pivot levels are derived from a completed session and are replaced the moment the next one closes; a Fibonacci grid lasts only until the swing it was drawn from stops being the move the market is trading. Carrying yesterday’s pivots into next week, or leaving a retracement on the chart after price has broken well past the origin of its swing, is the most common way these tools are misused, and it is a matter of housekeeping rather than of method.
Checking a number rather than trusting it
A short procedure settles almost every disagreement between a calculator and a charting platform, and it is worth running in this order.
Start with the input. Confirm the prices are adjusted the same way in both places, that the precision matches, and that the session boundary is the same, on anything trading around the clock, where "the close" falls is a decision made by your data provider rather than by the market. A surprising share of disagreements end here.
Then check the convention. For anything smoothed, establish whether the platform uses Wilder’s 1/n weight or the 2/(n+1) exponential form, and how it seeded the first value. The RSI calculator prints both intermediate averages precisely so this comparison can be made row by row instead of argued about.
Finally check the history. An exponential average never fully forgets, so two implementations fed different amounts of data will differ for many bars and then converge: which looks like a bug and is not. If the values agree by the hundredth bar and disagree over the first twenty, nothing is wrong with either of them.
Levels
Support and resistance from two different starting points: a completed session, or a completed swing. They frequently disagree, and knowing why is more useful than picking one.
- Pivot point calculatorClassic, Woodie, Camarilla and Fibonacci pivots computed side by side from one session’s high, low and close.
- Fibonacci retracement calculatorRetracements and extensions from any swing, in either direction, with 50 % labelled as what it is, not a Fibonacci ratio.
Oscillators
Where the arithmetic is worth seeing rather than trusting, because the intermediate values are exactly where two platforms diverge.
- RSI calculatorWilder RSI over any period, with average gain, average loss and relative strength shown for every bar.
- Rate of change calculatorThe percentage change other indicators smooth, in both its forms, with the lookback bar printed beside the current one.
- Moving average calculatorSimple, exponential and weighted from the same closes, with the seed and the smoothing constant stated rather than assumed.
The numbers behind the tools
Reference pages rather than calculators: what the inputs mean, which baseline a comparison needs, and how far a published threshold can be trusted. Every figure on them is computed when the page is built, and labelled with what it was computed from.
- Technical analysis and what it supportsThree kinds of claim travel under one name. Includes a computed demonstration of what a parameter sweep looks like when it is presented as a test.
- Average volume as a baselineThe denominator under every claim that a session was heavy: which window, whether today belongs in it, and why the mean sits above an ordinary day.
- Inside an advance/decline countWhat the five published fields contain, which issues distort them, and how two numbers become four measures that can disagree.
- Reading a new-high / new-low chartThe twelve-month window that suppresses new highs for a year after a peak, counted issue by issue on a synthetic basket.
- McClellan oscillator across erasWhy a raw reading from an earlier decade sits on a different scale, and what the ratio adjustment fixes.
Frequently asked questions
Why show the intermediate steps?
Because a number you cannot reproduce is a number you cannot check, and "my platform gives a different answer" is the most common question about every indicator here. The answer is almost always in the intermediate values, a different smoothing constant, or a series that started too recently for the smoothing to have converged. Printing the columns turns an argument into an inspection.
Do these send my prices anywhere?
No. Every calculator runs as ordinary JavaScript in your browser and there is no server-side component to receive anything. You can verify it: load a page, disconnect from the network, and the calculators still work. The privacy page covers this in more detail.
Which decimal precision should I use?
Enough to match your data source, and this matters more than it sounds. The standard RSI verification series returns 70.53 at full precision and 70.46 when the same prices are rounded to two decimals, a difference large enough to start an argument, produced entirely by the input. If your figures disagree with a platform, check the precision of what you fed in before checking anything else.
Can I use these for intraday data?
The arithmetic is indifferent to the timeframe. The interpretation is not: pivot levels depend on where your data provider puts the session boundary, which on a 24-hour instrument is a convention rather than a fact, and volume-based measures are distorted by the open and the close. Each tool’s page states the caveat that applies to it.
How much price history should I paste in?
For the pivot and Fibonacci calculators, only the handful of values they ask for. They compute from a completed session or a completed swing and nothing else. For the RSI calculator, considerably more than the period suggests: Wilder smoothing carries its seed forward for ever with a decaying weight, so a 14-period RSI fed 20 bars is still substantially a function of where the data began. Give it a hundred bars or more if you intend to compare the output with a platform.
Why do two platforms disagree about the same indicator?
Four causes account for nearly all of it. A different smoothing weight, 1/n as Wilder specified against 2/(n+1) for a conventional exponential average. A different seed for the first smoothed value. A different amount of history before the point you are looking at. And a different input series, most often adjusted against unadjusted prices around a split or a dividend. The columns these tools print let you find which of the four it is in a minute or two.
Do the levels from two different tools ever agree?
Occasionally, and it is worth noticing when they do. A pivot resistance and a Fibonacci retracement level are computed from unrelated inputs — one session’s range against a multi-week swing — so when they land within a few ticks of each other, two independent methods have projected the same price. That is a coincidence rather than a confirmation in any statistical sense, and practitioners treat such a confluence as a place to look more carefully rather than as a stronger signal.
Should I round the inputs?
No, feed the tools the same precision your data source carries, and round only the output. Rounding the inputs propagates: on the standard RSI verification series, two-decimal inputs shift the result by roughly seven hundredths of a point, which is small in isolation and enough to make you think a correct implementation is wrong. If you are comparing with a platform, match the input precision first and the disagreement often disappears on its own.
Is there a calculator for volume-based measures?
Not yet, and the reason is data rather than arithmetic. A volume calculator is only useful if you can paste in a volume series alongside the prices, which is a good deal more typing than a swing high and a swing low, so the volume measures on this site are presented as computed figures on their own pages instead, where the series is in the source and can be inspected. The indicator library sets out the formulas in full for anyone working from their own data.