The Ultimate Oscillator: Three Lookbacks Averaged Into One Line

Larry Williams created the Ultimate Oscillator to solve a specific complaint about momentum indicators: that the answer depends entirely on the lookback you picked. His solution was to use three lookbacks at once and weight them. It is a genuinely clever fix that does not remove the underlying problem.

Ultimate Oscillator at a glance

What it is and why it works

Start with the problem it was built for. Every momentum oscillator asks you to choose a lookback period, and the answer changes when you change it. A 7-period RSI and a 28-period RSI can say opposite things about the same market at the same moment, and there is no principled way to decide which is right. Larry Williams’ response was to stop choosing.

The building block is buying pressure. For each bar, take the close and subtract the lower of either the bar’s low or the previous close. That gives you how far price closed above the point from which it could have fallen. Divide that by the true range: the higher of this bar’s high or the previous close, minus the lower of this bar’s low or the previous close. The result is the fraction of the bar’s real movement that buyers were responsible for.

Then sum that buying pressure over three windows (7, 14 and 28 bars) and divide each sum by the total true range over the same window. You now have three ratios describing short, medium and long-term buying pressure. The final step weights them 4, 2 and 1 respectively, so the short window matters most, and scales the result to a 0–100 range.

What this genuinely achieves is worth acknowledging. Because it uses true range, it does not ignore gaps the way many oscillators do, which puts it ahead of the A/D Line family. And because it averages three lookbacks, a single-window artefact cannot dominate the reading, which makes it noticeably steadier than a short RSI. Where honesty is required is on the redundancy question: the Ultimate Oscillator is still a bounded 0–100 momentum measure built from price. It correlates heavily with RSI and with the stochastic, and running it alongside them is running three answers to nearly the same question. It is a better-constructed member of that family, not a different family.

How to trade it, step by step

1. Get an implementation and check it. MT5 does not include the Ultimate Oscillator, so you will need a custom indicator. Before trusting it, verify against a known reference that it uses true range rather than the raw high-low range, and that the weights are 4, 2 and 1. Community implementations vary, and a wrong one looks perfectly plausible on a chart.
2. Leave 7, 14 and 28 alone. The whole design rationale is that three separated lookbacks average out single-window artefacts. Bringing them closer together defeats the point and turns the indicator into a slow RSI. If you want it faster, use a lower timeframe rather than shorter periods.
3. Read 50 as the dividing line first. Above 50, buying pressure has accounted for more than half of true range across the blended windows. Below 50, selling pressure has. This is a cleaner state read than the 70/30 levels and it is what Williams himself emphasised.
4. Learn the specific bullish divergence rule, because it is the indicator’s real signal. Williams defined it precisely: price makes a lower low, the oscillator makes a higher low, and that oscillator low stays below 30. The trigger is the oscillator then rising above the high point it reached between the two lows. All three conditions matter: most traders use only the first two and get a much weaker signal.
5. Mirror it exactly for the bearish case. Price makes a higher high, the oscillator makes a lower high, and that oscillator high is above 70. Enter when the oscillator falls below the low point it made between the two highs. Again, the level condition is not optional.
6. Set the stop from price, not from the oscillator. Williams’ own guidance was to place the stop beyond the extreme that formed the divergence: the lower low for a long, the higher high for a short. That is a structural level with a defined invalidation, which is exactly what an oscillator reading cannot give you.
7. Exit when the oscillator reaches the opposite extreme or crosses back through 50. Choose one of these in advance and apply it consistently. The divergence rule generates an entry but says nothing about exits, and traders who improvise the exit lose the benefit of a well-defined entry.
8. Confirm the market is not ranging before you act on any of it. Divergence in a sideways market is continuous and meaningless, because price makes marginal new extremes in both directions all day. Establish the state from trends versus ranges before you look at the oscillator at all.

Size every one of those entries with the position size calculator and check the trade is worth taking with the risk/reward calculator before you commit.

The conditions it needs

At the end of an extended, tiring trend

The divergence rule is designed to catch exhaustion, and it works best where there has been a genuine sustained move to become exhausted from. A market that has trended for weeks and starts producing lower lows on visibly reduced buying pressure is the exact scenario the indicator was constructed around.

Instruments and timeframes with gaps

Because the denominator is true range rather than the bar’s own high-to-low range, the Ultimate Oscillator handles gaps correctly. On indices, shares or anything read across a weekend break, that is a real advantage over the oscillators that pretend gaps did not happen.

As a replacement for a short-period oscillator, not an addition

Its genuine benefit is stability. If you have been running a 7 or 9 period RSI and finding it too noisy, the Ultimate Oscillator does the same job with far fewer false extremes, because a single window cannot dominate the reading. Swap one for the other rather than running both.

Higher timeframes, where the three windows span meaningful periods

On a daily chart, 7, 14 and 28 bars represent roughly a week, a fortnight and a month, genuinely different horizons that can genuinely disagree. On a 1-minute chart they represent three near-identical slices of the same few minutes, and the blending achieves very little.

When it fails

• Divergence in a range is constant and worthless. A sideways market produces marginal new highs and lows continuously, so the oscillator diverges against price several times a session. Without a state filter, the indicator’s main signal fires almost permanently and almost always wrongly.
• Averaging three lookbacks does not remove lag, it averages it. The blend is dominated by the 28-period component whenever the market changes character quickly, because the weighting reduces but does not eliminate its influence. The indicator is steadier than a short RSI and correspondingly slower to register a genuine turn.
• It is not independent from RSI or the stochastic. All three are bounded momentum measures derived from price, and they agree the great majority of the time. Adding the Ultimate Oscillator to a chart that already has one of them produces the feeling of confluence without the substance.
• Most traders use a truncated version of the signal. The published rule requires the divergence low to be below 30, or the divergence high above 70, and requires a specific trigger point. Dropping those conditions, which is what almost every online description does, leaves a generic divergence signal with none of the selectivity Williams designed in.
• The 4:2:1 weighting is an arbitrary choice. There is no derivation behind those weights; they were selected because they worked on the markets Williams was looking at in the 1970s. They are not a property of markets, and treating the resulting number as objective overstates what it is.
• No standard MT5 build means implementation risk. Because you must source a custom indicator, you are trusting someone else’s code. Versions using high-low range instead of true range, or the wrong weighting, are common and produce a line that looks correct while measuring something different.

Markets it is most informative on

• SPX500 (S&P 500): Extended trends punctuated by genuine exhaustion phases, and true-range handling deals with the overnight gaps.
• Gold (XAU/USD): Long directional runs that end in visible loss of pressure: the divergence rule’s intended setting.
• US30 (Dow Jones): Gaps between sessions are handled correctly here, unlike with high-low-based oscillators.
• EUR/USD: Deep and orderly enough on higher timeframes for the three-window blend to describe distinct horizons.

For different levels of experience

If you are brand new

This is not a first indicator, and the honest advice is to learn RSI properly before coming here. They measure much the same thing, and RSI is simpler and ships with your platform.

If you do want to understand the idea: every oscillator asks you to pick a number of bars to look back over, and the answer changes depending on what you pick. That is uncomfortable, because it means the indicator is partly reporting your choice rather than the market. The Ultimate Oscillator uses three different lookbacks at once and blends them, so no single choice dominates.

Read it in one way only as a beginner: above 50 means buyers have been doing more of the work recently, below 50 means sellers have. Ignore the divergence rules until you can reliably tell a trending market from a ranging one, because in a ranging market those rules will fire constantly and cost you money.

If your results are inconsistent

Two things separate people who get something from this indicator from people who do not. The first is using the complete signal. Williams’ bullish divergence requires three conditions (price lower low, oscillator higher low, and that oscillator low below 30) plus a specific trigger when the oscillator exceeds the peak it made between the two lows. Nearly every description online drops the level condition and the trigger. What is left is generic divergence, which is far less selective and is the version most people are unknowingly trading.

The second is the state filter. Divergence is a statement about exhaustion, and exhaustion only exists after exertion. In a range there has been no exertion, so every divergence is noise. Establish that the market has actually trended before you look for the signal, and you will take a fraction as many trades with a much better character.

One practical warning: since MT5 has no built-in version, check whatever you download. Compare a few values against a manual calculation on a handful of bars. Implementations that substitute the bar’s high-low range for true range are common, and they quietly remove the gap handling that is one of this indicator’s few genuine advantages.

If you are experienced

Structurally this is a ratio-of-sums rather than a sum-of-ratios, which matters more than it sounds. Because each window divides accumulated buying pressure by accumulated true range rather than averaging per-bar ratios, high-range bars dominate the reading in proportion to their range. That is defensible, it weights by activity, but it means the oscillator is not a clean momentum measure; it is momentum weighted by realised range, and it behaves differently across volatility regimes for that reason.

The multi-window blend is best understood as a crude approximation of a longer filter with a shaped impulse response. The 4:2:1 weighting is unmotivated, and if the goal is lookback robustness there are better-founded approaches: averaging a broader set of periods, or fitting the weights to the autocorrelation structure of the instrument. Treating 7/14/28 as canonical is treating one 1970s parameterisation as a market property.

Its correlation with RSI on the same effective horizon is high enough that it should be treated as a substitute rather than an addition in any factor set, and the true-range denominator is the specific reason to prefer it: it makes the series robust to gaps in a way the RSI family is not, which matters on anything with a session break. The divergence rule itself has the usual problem of all divergence rules; it is defined retrospectively on confirmed swings, so any backtest must construct swings using only information available at the time, or the results are meaningless.

Risk management for this strategy

The Ultimate Oscillator’s main signal is a counter-trend entry, and counter-trend entries are where accounts are damaged fastest. That single fact should set the risk approach.

Williams’ own stop placement is the right starting point: beyond the price extreme that formed the divergence. That is often a considerable distance, particularly after a sustained move, and the correct response is a smaller position rather than a closer stop. Work the size from the actual stop distance with the position size calculator, and if the resulting position feels too small to be worth taking, that is the trade telling you something.

The specific failure to guard against is adding to a losing divergence trade because the divergence has become “even stronger”. In a genuinely strong trend, divergence deepens for weeks while price continues. A rule that says never to add to a position taken against the prevailing direction removes the worst outcome this indicator can produce.

Where Market Structure Pro fits

Everything that makes the Ultimate Oscillator useful depends on one prior judgement: has the market actually trended enough for exhaustion to be a meaningful concept? Get that right and the divergence rule is selective. Get it wrong and it fires all day in a range, offering a counter-trend trade every time.

Market Structure Pro makes that judgement explicitly rather than leaving it to interpretation. Its dedicated ranging and chop filter exists to say NO TRADE when conditions are directionless, and the verdict comes with a confidence percentage, an A/B/C grade and a plain-English explanation of what is driving it. That converts the hardest part of using this indicator from a subjective call into a stated state.

There is a second, less obvious fit. The Ultimate Oscillator was designed to reduce dependence on a single lookback: an attempt to avoid the trap of one tool giving one answer. MSP takes the same instinct considerably further, fusing 27 genuinely different tools into a single verdict rather than three weightings of the same measurement. And because the state locks on the closed bar and does not repaint, a divergence you acted on is still recorded exactly as it appeared, which is the only way to review whether the full Williams conditions were actually met or whether you took the shortened version.

Frequently asked questions

What does the Ultimate Oscillator measure?

It measures buying pressure as a proportion of true range, calculated over three lookback periods at once and blended into a single line scaled from 0 to 100. Buying pressure is the distance from the close down to the lower of the bar's low or the previous close, so the oscillator describes how much of the real movement buyers accounted for.

Why does the Ultimate Oscillator use three periods?

Because a single lookback makes any momentum indicator partly a reflection of the setting rather than the market, and a short and a long period can say opposite things at the same time. Larry Williams blended 7, 14 and 28 bar windows, weighted 4 to 2 to 1, so no single window can dominate the reading.

Is the Ultimate Oscillator in MetaTrader 5?

No. It is not among the standard MT5 oscillators and requires a custom indicator, either from the MQL5 community or coded yourself. Because implementations vary, it is worth checking that the version you install uses true range rather than the bar's high-low range, since that difference removes its gap handling.

What is the Ultimate Oscillator divergence rule?

For a buy, price must make a lower low while the oscillator makes a higher low, and that oscillator low must be below 30. The entry trigger is the oscillator then rising above the peak it made between the two lows, with a stop beyond the price low. The sell rule mirrors it with a high above 70.

What are the best Ultimate Oscillator settings?

The original 7, 14 and 28 with 4:2:1 weighting is the standard and there is good reason to keep it, because the design depends on the three windows being genuinely separated. Bringing them closer together turns the indicator into a slow RSI and removes the robustness it was built to provide.

How is the Ultimate Oscillator different from RSI?

RSI uses a single lookback and measures the size of up closes against down closes, while the Ultimate Oscillator blends three lookbacks and measures buying pressure against true range. The practical differences are that it is steadier than a short RSI and, because it uses true range, it accounts for gaps between bars.

Is the Ultimate Oscillator good for day trading?

It can be used intraday, but its design works better on higher timeframes where 7, 14 and 28 bars represent genuinely different horizons. On very short timeframes the three windows cover nearly the same few minutes, so the blending achieves little and you are left with a slow oscillator.

Why does the Ultimate Oscillator give false signals?

Almost always because it is being used in a ranging market. Its divergence rule is designed to detect exhaustion at the end of a sustained trend, and in sideways conditions price makes marginal new extremes constantly, so divergence appears continuously without meaning anything. A trend filter removes most of the problem.

Related reading

• RSI: The simplest member of the same family: learn it first, and treat the two as substitutes.
• Stochastic Oscillator: Another bounded momentum measure that largely overlaps in what it says.
• Williams %R: Larry Williams’ earlier and far simpler oscillator.
• Divergence: The core mechanic of this indicator, and its most misused concept.