The Ultimate Oscillator on short timeframes

[Fairman]

The Ultimate Oscillator on short timeframes

Larry Williams designed the Ultimate Oscillator to deal with one of the common complaints about momentum indicators: a single lookback period gives false signals because it only sees one time horizon. His solution was to combine three periods, usually 7, 14 and 28, weighted so that the shortest has the most influence.

The calculation uses buying pressure, which is the close minus the lower of the low or the previous close, against the true range. It sums these over each period, divides, and weights the three averages into one line from 0 to 100.

On M5 or M15, the three periods cover roughly half an hour, an hour and two hours. That gives you a momentum reading that takes the recent swing and the wider session into account at the same time. In practice it's smoother than RSI with a similar period and gives fewer extreme readings.

Williams had specific rules for divergence with this oscillator, involving a divergence below 30, then a break above the high of the divergence. Those rules are slow for scalping, so I don't use them as written.

Where I find it useful is as a check on whether a pullback has gone deep enough. In an uptrend on M15, a pullback that drops the oscillator into the 40 to 50 area and then turns up has often been a reasonable spot to look for entries. Below 30 the trend itself is in question.

[FXS HongKong Scalper]

The three timeframe blend is clever. On short timeframes one thing I noticed is that the default 7, 14 and 28 on M1 covers less than half an hour, which can make the line react to noise. Scaling the periods up, for example 14, 28 and 56, still keeps the idea of three horizons while smoothing the result. The divergence signals Williams described also need a clear prior trend to be meaningful. On a range bound chart, divergences appear all the time and tell you little.

[Fairman]

Changing the periods to suit scalping is common. Some people halve them to 4, 8 and 16 for M1 charts. The weighting stays the same. With shorter periods the oscillator becomes more reactive but noisier. I'd test a few combinations on your pair rather than just copying a setting, since how quickly a pair's momentum shifts varies. Short periods work better during active sessions. During the Asian session, the default periods may suit M5 better since moves are slower and smaller, and halved settings mostly react to noise. Fifteen to twenty minutes of looking through recent charts with different settings tells you a lot.

[Fairman]

Because it combines three periods, the oscillator can stay neutral while price makes a sharp short move. The longer periods hold it back. That's a strength if you want to avoid chasing spikes. It's a weakness if you're trying to catch the start of a fast move. It suits pullback trading more than momentum entries. Combining it with a fast indicator for entries is one option. The Ultimate Oscillator tells you whether the pullback is healthy, and something quicker like a short stochastic or simple price action tells you when to enter. That division of labour suits the way it's built.

[Fairman]

Not every platform includes it by default. MT4 doesn't have it as a built in indicator, but it's available as a free custom one. Check the code if you download one, since some versions use slightly different calculations. TradingView includes it as standard, and the default settings match the original formula. On TradingView you can also adjust the weights if you want to experiment, though the standard weighting of four, two and one for the short, medium and long periods works reasonably for most uses. For MT4, the custom versions usually come with source code, so you can compare the formula with Williams' original description before trusting the numbers.

[Fairman]

Comparing it to stochastic, the Ultimate Oscillator uses closes relative to the true range over several periods, while stochastic uses the close relative to the high low range of a single period. In practice the Ultimate Oscillator spends less time at extremes, so it's less useful for spotting overbought and oversold conditions. I'd use it for trend health more than reversals. For reversal trading, I'd look at stochastic or RSI instead. The Ultimate Oscillator's strength is showing whether momentum across several periods agrees, which makes it a better filter for trend trades than a tool for catching turns.

[Shadow Trader]

Few things sting like a limit order that sits one tick short of the level, then watches price reverse into the profit zone without it.

The brain files that as a loss, even though no money was spent. It has the same flavor as a stopped-out trade. The urge that follows is to enter at market, chasing the move, at a worse price than the original plan.

It's a bias of its own: regret for an action not taken, treated as a real cost. In short timeframes, it comes up dozens of times a week.

The cleaner way to treat it is mechanical. An order that wasn't filled is not a trade. The setup may be gone, and a new decision at the current price needs its own justification, separate from the earlier plan.

A rule that helps: when an order misses by a tick, the setup is canceled, and the next signal is awaited, with no chasing.

How is regret over unfilled orders handled when price reverses in the right direction without the trade?