IPDA (interbank price delivery algorithm)

Behind the vocabulary of gaps and blocks sits a claim about what moves price, and the interbank price delivery algorithm is its name.

The premise, as the teaching states it

In ICT teaching, price is not simply discovered by buyers and sellers meeting; it is delivered, moved with intent from pools of resting orders on one side to imbalances such as fair value gaps on the other. Every move is read as doing one of two jobs: taking liquidity, or rebalancing an imbalance. The teaching attributes that delivery to an algorithm run in the interbank market, and adds that its direction shifts roughly every quarter.

None of that can be confirmed from public data. There is no published algorithm and no way to observe one from prices alone, and the premise is best read as the teaching’s account of why its objects should matter. What can be computed, and tested, are the data ranges that come with it.

The data ranges

The IPDA data ranges are rolling lookbacks on daily candles:

• The 20-day range: the highest high and lowest low of the last 20 trading days.
• The 40-day range: the same over 40 trading days.
• The 60-day range: the same over 60 trading days.

Because each longer window contains the shorter ones, the ranges nest: the 60-day high is never below the 40-day high, and the 60-day low is never above the 40-day low. Each range’s midpoint splits it into a premium half above and a discount half below, the same halves a PD array is sorted by. One published summary maps the three lengths to short, intermediate and long-term delivery.

A worked case, with constructed numbers, on daily EUR/USD. Over the last 20 trading days the high is 1.0960 and the low 1.0842; over 40, 1.0995 and 1.0842; over 60, 1.1030 and 1.0788. The midpoints are 1.0901, 1.09185 and 1.0909. With price at 1.0880, it sits in the discount half of all three ranges. The 20-day high at 1.0960 is the nearest buy-side extreme, and the low at 1.0842, shared by the 20 and 40-day ranges because the lowest price of the last 40 days printed inside the last 20, is the nearest sell-side one.

Reading price inside the ranges

Teaching reads the extremes as buy-side and sell-side liquidity: the orders resting above the range highs and below the range lows. A price in the discount half of the ranges is read as more likely to be delivered up toward the highs, and a price in the premium half as more likely to fall toward the lows, a reading that has to agree with the higher-timeframe structure before the method acts on it. The draw on liquidity is then chosen from these extremes and the other pools on the chart. The ranges supply candidates; they do not pick one.

The quarterly shift

The teaching also holds that delivery changes direction roughly every quarter. “Roughly” carries a great deal in that sentence: a shift that may arrive at any point within a few weeks of a date can be found after the fact in almost any price history. As a testable claim it needs a stated rule for what counts as a shift and when one is due, written down before the prices it is tested on.

The widest of the levels

The IPDA ranges are the widest of the reference levels the vocabulary measures a session against. Inside them sit the week’s and the day’s gaps, the new week opening gap and the new day opening gap; the opening range gap on index futures; and the central bank dealers range, which frames a single day. Read together they give a session a nested map, from a sixty-day range down to a six-hour one, and the daily bias is expected to agree with where price sits in the larger frames before the smaller ones are traded.

On lower timeframes

One published indicator also counts the same lookbacks in candles of a lower timeframe, such as the last 20, 40 and 60 hourly candles. That is the same arithmetic on a different clock. The ranges change with the timeframe, and a claim about lower-timeframe lookbacks is a separate claim from the one about daily ranges, with its own measurement. The teaching’s own statement of the ranges is in trading days.

Session conventions and the Sunday candle

A daily range depends on where each day starts and ends. Currencies and futures conventionally close the day at 17:00 New York time. Some currency feeds build daily candles on a server clock instead, and those feeds often print a short Sunday candle covering the first hours of the week. On such a feed twenty candles are not twenty trading days, and the ranges shift. The daily convention, and whether Sunday candles are merged or dropped, belong with any IPDA range quoted.

Computing it, step by step

1. Choose the feed and the daily convention, and merge or drop short Sunday candles.
2. Count back 20, 40 and 60 completed trading days, leaving out the day in progress.
3. Take each window’s highest high and lowest low, and each midpoint.
4. Extend the six extremes and three midpoints to the right.
5. Recompute each day, since the windows roll and an extreme drops out of a range when its day leaves the window.

Common mistakes

Counting calendar days instead of trading days. Including the day in progress, so that the ranges move during the session. Mixing feeds with and without Sunday candles. Treating the interbank algorithm as established, when it is the premise the ranges are meant to test. And crediting the ranges whenever price reaches an extreme, when any rolling high or low is reached some of the time.

What the ranges do not settle

The ranges are exact once the feed and the daily convention are stated. Whether price reaches their extremes, or turns at their midpoints, more often than it reaches the extremes of other lookbacks is a measurement, and the data ranges driver has a row in the claims ledger, where that verdict renders. The two lenses keep the arithmetic, which is checkable, apart from the premise of an algorithm, which is not.