Copper: the price history, 2000 to 2026

The chart above is drawn from the same 931 points every figure on this page is computed from — not a separate feed, and not a picture of one. Moving the cursor across it reads out the closing price on the date under the pointer; the buttons narrow the window to the last year, three, five or ten, and the axis relabels itself to whatever range is showing. The high and low printed beside it, $6.70 and $0.6115, are the extremes of the whole record.

It is plain SVG with no charting library behind it, which has two consequences worth stating. The page carries no third-party script for the chart, so nothing about your visit is sent anywhere to draw it. And because the points are a file in this repository rather than a request made when you arrive, the chart renders identically whether or not any data provider is reachable — the failure mode of a live widget, a blank box where a price should be, cannot happen here.

What the chart cannot show is inside the gaps between points. This series is sampled at roughly weekly spacing from 6,515 daily closes, so a spike that came and went inside a single week does not appear. The line is the record at the sampled moments, drawn straight between them, and it should be read as exactly that rather than as a continuous trace of every price ever quoted.

The highest point in the series is $6.70, reached on August 5, 2026. The lowest is $0.6115, on November 2, 2001. Between those two numbers sits everything else: the distance from the bottom to the top is a factor of 11.0.

The last point in the series is the highest point in the series. Copper is, on this data, at its record — which is worth saying plainly, because a record is the one moment when past performance looks most like a promise and is least likely to behave as one.

Prices are USD closing prices as published for HG=F on COMEX. A purchase made on another venue, in another currency, or with commissions and taxes applied would have produced a different number — the series measures the price, not your result.

Across 25 complete calendar years, Copper ended higher than it started in 17 of them and lower in 8. That is +68% of years positive — a majority, but a majority that says nothing about which year you happen to hold it through.

The best full year was 2009, which ended +138.5%; the worst was 2008, at -52.8%. Those two years alone span 191 percentage points. An investor who saw only the average of the two would have learned nothing about either.

Each year below is measured from the last close of the previous year to the last close of its own — not from the first available point inside the year, which would quietly drop whatever happened over the turn of the year. Years the series does not cover in full are marked, because reporting a partial year as if it were complete is simply false.

The deepest fall in the series took Copper from $4.08 on July 2, 2008 down to $1.31 on December 19, 2008: -68.0%, over 6 months. The price returned to that level on December 10, 2010 — 2 years and 5 months after the peak. Anyone who bought at the top and held got their money back, eventually; the cost was the waiting, not the loss.

That was not the only one. The series contains 6 separate falls of 20% or more from a previous high, of which 2 went past 50%. A single worst-case number invites you to treat the drop as a freak event; the list makes it clear how often the same thing happened.

Another: from $4.54 on February 1, 2011 down to $1.97 on January 14, 2016, -56.6% over 4 years and 11 months. That one was undone by May 3, 2021, 10 years and 3 months after the peak it started from.

Another: from $3.87 on May 15, 2006 down to $2.47 on February 12, 2007, -36.2% over 9 months. That one was undone by April 3, 2008, 1 year and 11 months after the peak it started from.

Counting from the first point to the last, Copper spent 94% of the period below a level it had already reached — 24 years and 3 months out of 25 years and 11 months. That figure is rarely published and is the one that describes the experience of holding: not how much was lost at the worst moment, but how much of the time was spent waiting to get back to even.

Everything above describes the price. This describes a decision. The table sets $10,000 against a handful of dates that were not chosen for being round numbers: the start of the series, its highest point, its lowest, and the anniversaries the data actually covers. The value shown is what that sum would be worth at the last point in the series, before any commission, spread or tax.

The spread between the best and the worst of those dates is the whole lesson. $10,000 put in on November 2, 2001 would stand at $109,616; the same sum on August 2, 2021 would stand at $15,124. Same asset, same holding, one of them far behind — the difference is the day, and the day is the one thing nobody knows in advance.

A price does not climb evenly; it crosses lines, falls back below them, and crosses again. The table lists the first close above each round level the series contains, and — where it happened — the date the price dropped back under that level and the date it took it back. Those round numbers have no meaning to the asset itself, but they are where attention gathers, and the record of how often they were lost and regained is a plainer description of the path than any average.

Copper first closed above $1.00 on December 22, 2003 and, in this series, never closed below it again. The highest line it has crossed is $5.00, first reached on March 25, 2025 and lost again on April 3, 2025.

Zoom out from years to decades and the picture changes again. The series spans 3 of them, 1 covered from start to finish. A decade is long enough that the falls in the table above are absorbed into it, which is precisely the point: the same asset can look ruinous over eighteen months and unremarkable over ten years, and neither view is the honest one on its own.

2000–2009 (partial): +276.0%, ending at $3.33. 2010–2019: -16.0%, ending at $2.79. 2020–2029 (partial): +139.9%, ending at $6.70.

The sharpest single move up in the series runs from $2.87 on April 13, 2006 to $3.41 on April 25, 2006: +18.8% between two consecutive points. The sharpest move down runs from $5.78 on July 24, 2025 to $4.42 on August 4, 2025: -23.6%.

Both are worth holding in mind next to the annual figures. A year that ends +138.5% can contain a fortnight like the second one; an average return computed over 26 years contains every one of these moves and shows none of them. This is measured between points about a week apart, so an intraday spike larger than either figure would not appear here.

One more way to place the last point: 100% of every point in this series is below it. That is not a valuation and says nothing about whether the price is justified — it says where today sits inside the asset's own history, which is a question people usually answer from memory and answer badly.

From the first point in the series, Copper took 5 years and 1 month to double for the first time, reaching that mark on September 21, 2005. Everything after that is compounding on a base that had already moved.

Almost nobody buys once and never again. So here is the other arithmetic: $100.00 put in at every point in the series, starting August 30, 2000 — 931 instalments, $93,100 paid in altogether. At the last point that holding would be worth $295,264, +217.1% on what was paid in.

The number is usually far below the headline return of a single lucky purchase, and that is not a flaw in the method: money paid in late has been invested for a short time, and averaging in means buying some of the position at every high as well as every low. What it does remove is the need to be right about the date — the variable that, in the table above, made the difference between +996.2% and +51.2%.

A gain is only a gain if it beats the cost of living. Measured from August 30, 2000, Copper is up +657.4% in cash terms. Over the same period United States consumer prices rose +93.3%. Take the second out of the first and what is left is +291.9% — the part that actually buys more than it did.

The deflator is the US city-average consumer price index, series CUUR0000SA0, published by U.S. Bureau of Labor Statistics. It is a national average and not your basket: if your spending is concentrated in housing, healthcare or education, the inflation you experience differs from the index, and so does the real return you should be measuring against. The two figures are shown side by side rather than blended, so you can see how much of the headline number was the asset and how much was the currency.

Measured on the points in this series, Copper has an annualised volatility of 25.6%. Read plainly: in a typical year the price has moved around its own trend by roughly that much, up or down. It is a description of how bumpy the ride has been, not a probability and not a limit — the largest fall in the table above is far bigger than one year of this figure, which is exactly why volatility alone is a poor description of risk.

The figure is computed on 931 points spaced a day apart, so it measures the sampled series rather than every intraday move. A denser series would give a slightly different number; the direction and the order of magnitude would not change.

This series begins on August 30, 2000. It cannot tell you anything about Copper before that date, and it does not try. Every figure on this page is bounded by that start: a "worst fall" is the worst fall the series contains, not the worst that ever happened, and a compound annual rate computed over 26 years would be a different number over a different window.

None of it is predictive. The dates in the tables were chosen because they teach something — two of the most instructive are the highest and lowest points, which is to say the moments when a buyer felt most confident and least confident, and did worst and best. Knowing that is useful. Knowing it does not tell you where the next 26 years go.

The arithmetic ignores costs that are real: commissions, the gap between the buying and selling price, currency conversion where your money is not in USD, custody, and tax on any disposal, which depends on where you live and is not a detail. A result computed gross of all of these is an upper bound on what anyone actually received.

The last few years, taken one at a time, because a single compound figure hides the order in which things happened — and the order is what anyone holding the asset actually lived through.

2021 ended up +23.8%, closing at $4.38. Inside the year the price ranged between $3.56 on February 1, 2021 and $4.76 on October 20, 2021 — a gap of +34% between the cheapest and the dearest point of the same twelve months.

2022 ended down -13.2%, closing at $3.81. Inside the year the price ranged between $3.33 on July 13, 2022 and $4.74 on March 31, 2022 — a gap of +43% between the cheapest and the dearest point of the same twelve months.

2023 ended up +2.5%, closing at $3.90. Inside the year the price ranged between $3.57 on October 23, 2023 and $4.25 on January 23, 2023 — a gap of +19% between the cheapest and the dearest point of the same twelve months.

2024 ended up +3.6%, closing at $4.04. Inside the year the price ranged between $3.73 on February 12, 2024 and $4.87 on May 22, 2024 — a gap of +31% between the cheapest and the dearest point of the same twelve months.

2025 ended up +34.6%, closing at $5.44. Inside the year the price ranged between $3.99 on January 2, 2025 and $5.78 on July 24, 2025 — a gap of +45% between the cheapest and the dearest point of the same twelve months.

2026, which this series covers only in part, ended up +23.3%, closing at $6.70. Inside the year the price ranged between $5.53 on March 25, 2026 and $6.70 on August 5, 2026 — a gap of +21% between the cheapest and the dearest point of the same twelve months.

A total return says where the price ended up; it says nothing about how it got there. Broken into the 930 steps between consecutive points, Copper rose in 496 of them and fell in 431 and was unchanged in 3 — 53.3% of steps positive. The average step up is +3.26% and the average step down -3.08%.

Those two averages are worth setting next to each other. The typical rise is larger than the typical fall, which is the other way a series climbs: fewer good weeks, but bigger. Neither pattern is better; they simply feel different to hold.

The longest unbroken run of rises is 9 consecutive weeks, from November 19, 2003 to February 26, 2004. The longest unbroken run of falls is 8 weeks, from May 23, 2001 to August 13, 2001. Streaks of this length appear in almost every price series and are the raw material of the belief that a trend is under way; the series contains no mechanism that makes the next step depend on the last one.

Closer in, the last 24 months the series covers. 16 of them closed higher than the month before and 8 lower. The strongest was 2025-07 at +17.6%; the weakest 2025-08 at -23.0%.

A month is short enough that a single piece of news dominates it and long enough that the number looks like a trend. Both things are true at once, which is why the table is here in full rather than summarised: the months that stand out are visible, and so is how ordinary the rest of them were.

A page that names its source should also say where that source is silent. Between August 30, 2000 and August 5, 2026 this series holds 931 points against roughly 1,053 that its own spacing would imply. The longest stretch without a point runs from December 22, 2003 to January 6, 2004, 15 days — long enough that a move inside it would not appear anywhere on this page.

Gaps in a price series are usually mundane: markets close at weekends and on holidays, a venue suspends trading, a provider backfills late. They matter here for one reason only — every high, low and change on this page is computed from the points that exist, so a gap is a piece of the record this page cannot see and does not pretend to.

Every number on this page rests on one kind of measurement — a closing price — and it is worth being precise about what that is and is not. A close is the last price at which the asset changed hands in the session the source is reporting on COMEX, in USD. It is a single trade, not an average of the day, and it carries no information about how much was traded at it: a close set by one small transaction and a close set by heavy volume look identical in a series like this one.

It is also not the price you would have got. A buyer pays the offer and a seller receives the bid, and the gap between the two is a cost that never appears in a closing series; on a thinly traded asset that gap can be a large fraction of a day's move. Add commission, and for anyone whose money is not already in USD, the cost of converting into it and back out again — the exchange rate over the same period is its own price series, and it can dominate the result.

Finally, a price series is not a total return. It follows what one unit costs, not what holding it produced: distributions, interest, staking rewards, splits handled differently by different providers, and tax on any disposal all sit outside it. Where those things are positive the holder did better than the line; where the tax fell due they did worse. This page measures the price honestly and does not dress it up as a return.

The series is Copper — front-month futures, US dollars per pound, identifier HG=F, published by Yahoo Finance. It was read on August 11, 2026 and the file is committed to this site's repository, which is why the page renders the same numbers every time it is opened and why the date of the last point never drifts away from what is drawn.

You do not have to take any of it on trust. The source is linked below; open it, pull the same series, and the numbers in the tables should reproduce. If they do not, that is a mistake on this page and it should be reported — corrections are published rather than quietly edited away.

Redoing the arithmetic takes four steps and no special tools. Take the closing price on the date you want and the closing price at the last point, $6.70 on August 5, 2026. Divide the second by the first: that ratio, minus one, is the percentage change. Multiply the ratio by the sum invested for the value column. For the compound annual rate, raise the ratio to the power of one divided by the number of years between the two dates, and subtract one. Every figure in the tables above comes out of those four operations applied to the same 931 points, with no smoothing, no adjustment and nothing dropped.

Two things the series does not do, stated so you are not surprised by them. It is not adjusted for dividends or any other distribution, so for an asset that pays them the total return to a holder was higher than the price change shown here. And it is sampled at roughly weekly spacing, so a high or low that occurred between two points does not appear — the extremes reported are the extremes of the recorded points, which is the honest way to describe them.