Something most people never notice
Whether a moving average rises or falls is already decided today. It has nothing to do with what happens tomorrow.
That sounds strange, but it is just arithmetic. The 60-day line is the mean of the last 60 closes. To compute tomorrow's value, two things happen: the close from 60 days ago gets dropped, and tomorrow's close is added.
The bar about to be dropped is the deduction value.
So:
- Deduction value below the current price → drop a small number, add a bigger one → the 60MA must rise
- Deduction value above the current price → drop a big number, add a smaller one → the 60MA must fall
Written as a formula it is even clearer:
Tomorrow's MA = today's MA + (tomorrow's close − deduction value) ÷ 60
Knowing the deduction value lets you make the call today. No forecast required.
What it is good for: estimating how much time is left
The most common use of a moving average is as support. A stock runs up, then goes sideways, and the 60-day line slowly climbs toward price. The moment it catches up, support is sitting right at the price — and the market usually has to pick a side: either break out with volume, or lose the line and end the move.
So "how many sessions until the 60MA reaches this price" is really answering:
> How much longer can this consolidation last?
That is what the MA Deduction page computes. Pick the index or a stock, enter a price to watch (leave it blank to use the latest close), and it tells you how many sessions the 60-day and 120-day lines each need.
How to read the page
The top two rows are the conclusion:
- Flat: assumes price stays where it is. This is the *slowest* case for the average to catch up, so the number is an upper bound.
- Continuing the 20-session slope: assumes price keeps moving at its recent pace.
The two numbers bracket a reasonable range.
⚠️ Be careful when the slope is negative. The trend row will show *fewer* days — but that is not because the average is catching up faster; it is because price would fall to the average. The meaning is the opposite. The page prints an extra note when it detects a negative slope; when you see it, read the flat row instead.
Below that is the detail explaining where the number came from: the current MA level, tomorrow's deduction bar, and whether that bar sits below or above the target (i.e. whether the average is about to rise or fall).
Three outcomes get three different labels: "already deducted" means the average is already there; "more than 120 sessions" means it cannot get there within the computable range; otherwise you get the day count.
A concrete example
Say a stock trades at 100, the 60MA sits at 95, and tomorrow's deduction value is 88.
88 < 100, so the 60MA will rise. How fast? If price simply stays at 100, each day drops a value near 88 and adds one at 100, lifting the average by roughly 0.2 per day. Closing a 5-point gap therefore takes about 25 sessions — around five weeks.
So you have roughly five weeks of room to consolidate before the 60-day line arrives.
Flip it around: if the deduction value were 105 (above price), the average would be falling toward a price of 100 — support pressing down from above, which means something entirely different.
⚠️ This is arithmetic extrapolation, not a forecast
This has to be stated plainly, because it is the easiest way to misuse the tool.
The deduction value itself is certain — that bar is already in the record and will not change. But "how many days until it catches up" rests on the assumption that price moves by the same amount every day, and real markets do not.
Read the number as "if this pace holds, roughly this long" — not "guaranteed this many days".
While we are here: about breaking a moving average
Since we are on the subject, one common misreading is worth clearing up. We backtested this on both the TAIEX and the Nasdaq Composite, and the answer splits into two very different questions — the answers differ by an order of magnitude.
Question 1: "It just broke the line today — what happens next?"
Treating every break as an independent event, the median subsequent decline is only about −2%, and seven to nine out of ten breaks are recovered within 20 sessions.
In other words, a single break is usually noise. Selling everything on a break gets you shaken out most of the time.
Question 2: "In this correction, how much is left from the break to the low?"
Here you first identify a complete cycle — index makes a cycle high, weakens, breaks the long-term average, and falls to the cycle low — and take only the last break before that low. On that basis, Taiwan's last decade (4 cycles) has a median of −23.4%, and the US last decade (6 cycles) around −17% to −20%.
⚠️⚠️ Both numbers are correct, but they answer different questions and must never be mixed.
Question 2 uses the knowledge of *which break was the last one* — and that is only knowable after the fact. In the moment you cannot tell whether the break in front of you is the final one. So the −20% figure is for setting expectations about the shape of a real correction, not for deciding whether to sell today.
Seeing "prices fall another 20% after breaking the half-year line" and liquidating is exactly the error of treating question 2 as question 1 — and the honest answer to question 1 is "seven to nine times out of ten it recovers".
Three practical suggestions
1. Treat deduction as a timetable, not a signal. It tells you *when* the average arrives, not whether to buy or sell.
2. Use it alongside a view you already hold. If you already like a stock and are waiting for an entry, "the 60MA needs 20 more sessions" means you have time. If you already want to trim, "the 60MA arrives in 3 sessions" is a reason to move faster.
3. Look a little further down the deduction queue. The page shows tomorrow's deduction bar, but you can scan ahead yourself: if the next several bars to be dropped are all high prices, the average will flatten or turn down — even if it is still rising today.
Taiwan vs US
Both sites have this feature, with different moving averages:
- Taiwan uses the 60-day and 120-day lines, the local convention.
- US Stock Coffee uses 50, 100 and 150 days. The 50-day holds the position Taiwan's 60-day does; the 100-day is what decides whether the market has truly turned to a headwind; the 150-day is the trend-template line.
⚠️ Do not carry the Taiwan day counts straight over. The conventions differ, and a page that says "quarterly line" while computing something else is worse than not having the feature at all.