How to Use the Awesome Oscillator
To use the awesome oscillator, read it as the gap between a fast and a slow average of each bar's midpoint. Above zero means the fast average is higher, which is a directional statement, and the histogram's shape is a second-order reading of whether that gap is widening.
The awesome oscillator subtracts a slow moving average from a fast one, both calculated on the midpoint of each bar rather than its close. The result is displayed as a histogram around zero. That is the entire construction.
Before you start
An understanding that it is the difference between two moving averages, and nothing more. The name suggests something novel; the arithmetic does not.
A decision about whether you are using the zero cross or the histogram’s shape. One is a direction signal, the other a momentum one, and they disagree regularly.
An acceptance that it lags, because both of its inputs do. A moving average describes bars that have already happened, and subtracting two of them does not change that.
The steps
1. Read above and below zero as direction
Above zero means the fast average sits above the slow one, which means recent bars have averaged higher than older ones. That is a directional fact about the past.
2. Use the zero cross as the filter
Crossing zero is a dated event you can act on. Sitting above zero is a state that can persist for dozens of bars.
3. Read the histogram’s shape separately
Bars growing means the two averages are separating; bars shrinking means they are converging. That is a second-order reading and it turns before the zero line does.
4. Do not treat shrinking bars as a reversal
The averages converging means the move is decelerating, which happens constantly inside continuing trends. On this site’s shared series direction runs average 2.01 bars.
5. Require structure before acting on either reading
A signal at a level price has already respected is a different event from one in open space, and the indicator has no way to distinguish them.
6. Do not run it alongside its close-based relative
Both are the difference between a fast and a slow average. Having both looks like confirmation and is the same arithmetic run on slightly different inputs.
7. Take the stop from the chart
Where the idea is wrong is structural. That distance sets the position size, and the oscillator plays no part in either decision.
How to tell it worked
The reading you use was chosen as either the zero cross or the histogram shape, not both.
0 trades came from shrinking bars alone, treated as a reversal.
The chart carries 1 difference-of-averages indicator, not two.
And every stop came from structure, so 1 adverse move costs your intended risk.
What the midpoint input changes
It uses the average of each bar’s high and low rather than the close. That makes it slightly less sensitive to a sharp close and slightly more representative of where the bar spent its time.
The practical difference is small. On most bars the midpoint and the close are close together, and the readings diverge mainly on bars with long wicks — which is a real distinction and a narrow one.
Against the close-based version
The other indicator has a signal line and this one does not. That is the largest functional difference: a signal line adds a crossover event, which is either useful smoothing or added lag depending on what you want.
The other uses exponential averages of the close; this uses simple averages of the midpoint. Exponential weighting responds faster to recent bars, so the two lead and lag each other in ways that depend on the market rather than being consistently ordered.
Neither is better and running both is the mistake. They are the same idea with different parameters, and two versions of one calculation will agree far more often than two independent measures would — which reads as confirmation and is not.
Using it as a filter and nothing else
Long setups only while it is above zero, short setups only while below. That is a single rule, it is easy to check, and it removes half your trades in the direction the recent averages do not support.
The setup still comes from structure. A level, a pattern, a tested area — something the market actually did, rather than a value on a histogram.
The oscillator’s job in that arrangement is to say no. A filter that has never stopped you taking a trade is not filtering, and the test of whether this one is working is whether you can name a trade you declined because of it.
Everything past that is asking a subtraction of two averages for information it does not hold. It knows where recent bars averaged relative to older ones. That is one fact, and most of what gets built on this indicator needs several more.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 53 mention this indicator in the
title, at a median of 10,321 views across 43 channels, and 60% of those titles are instruction-shaped.
Its close-based relative appears in 173 instruction-shaped titles at 12,724. The counts come from
site/corpus_count.py and site/rank_howto.py.
53 videos at 10,321 against 173 at 12,724 for the close-based version. A third of the coverage for nearly the same audience per video, which suggests the two are competing for the same viewers with almost the same content.
The answer to the question on that chart is that shrinking bars mean deceleration, not reversal. The two averages are converging, which happens repeatedly inside trends that continue — and on this site’s series 54% of 566 ten-bar windows finished higher, so the default is continuation.
When it fails
The failure is reading the histogram as a forecast, and it exits good trades early and often. Bars shrink whenever the fast average stops pulling away from the slow one, which happens during every pause in every trend. Read as a warning, that produces an exit on the first quiet stretch of a move that has much further to run — and the indicator was never saying anything about what comes next, only that the gap between two backward-looking averages had narrowed.
The second failure is trading the level instead of the cross. Above zero is a condition.
A third is running it with its close-based relative. They are the same calculation.
A fourth is expecting it to lead. Both inputs are averages of past bars.
A fifth is acting without structure. The oscillator sees no levels.
And a sixth is expecting the name to mean something. It is a subtraction.
Related
Awesome oscillator covers the calculation. MACD is the close-based relative with a signal line. And momentum indicator is the family both belong to.
The name promises more than the calculation delivers, which is worth saying plainly. It is a difference of two averages of the bar midpoint. Understanding that removed the temptation to look for something in it that is not there, and made it usable as the simple directional filter it actually is.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.