Dollar-cost averaging

The case for averaging rests on a fact about bitcoin's price record that is easy to state and hard to internalise: the asset spends most of its time far from its trend. The drawdown record shows peak-to-trough declines of eighty per cent or more in every cycle so far, and recoveries that took between two and three years. An investor who commits a large sum on a single day is therefore making a bet whose outcome depends heavily on which day they chose. An investor who spreads the same sum across two years is making a bet on the average price over those two years, which is a far more stable quantity.

The second reason is arithmetic rather than psychological. Because the fixed sum buys more units at low prices, the average cost of a schedule is always at or below the arithmetic mean of the prices paid. This is sometimes called the harmonic-mean effect, and it is the reason a schedule can show a profit in a period where the price ends lower than it began. It is a real effect, but it is modest, and it is not a reason to expect averaging to beat a well-timed purchase. It is a reason to expect averaging to beat a badly timed one.

The third reason is behavioural, and for most people it is the decisive one. A schedule removes the decision. There is no moment at which the investor must judge whether the price is attractive, because the rule has already answered the question. That matters more in bitcoin than in most markets, because the asset's volatility is designed to provoke exactly the two errors that destroy returns: buying in the middle of an advance because the advance feels permanent, and selling in the middle of a decline because the decline feels terminal. A schedule that runs through both is a commitment device against both.

The comparison that matters is between a schedule and a lump sum invested at the start of the same window. The table below sets out the opening and closing prices for the eight most recent completed calendar years, through 2025, which is the raw material for that comparison. Read it as a description of the terrain rather than as a result: the spread between the best and worst entry years in this window is wider than the total return of most equity indices over the same period.

Averaging is a risk-management and habit tool, not a return-maximising one. That distinction is worth stating plainly because the marketing around the method tends to blur it. If an asset rises over the long run, a lump sum invested at the beginning of the period will usually beat a schedule spread across it, because the lump sum is exposed to the asset for longer. Averaging wins only when the entry date is poor, and it wins by losing less rather than by gaining more. An investor who averages is choosing a narrower distribution of outcomes, not a higher average one.

The method also does nothing about the quality of the asset. A schedule applied to something that declines forever produces a steadily growing loss, purchased at a steadily improving average price. Averaging is agnostic about what is being bought; it assumes the long-run direction is upward and manages only the path. That assumption is doing all the work, and it is an assumption rather than a result.

There are practical costs as well. Each purchase is a separate transaction, so exchange fees, spreads and — in most jurisdictions — a separate tax lot accumulate with every contribution. A monthly schedule over ten years produces a hundred and twenty lots, each with its own cost basis, which is a real administrative burden at disposal time. For small contributions the fixed component of trading fees can consume a meaningful share of the amount invested, which is why low-cost venues matter more for a schedule than for a single purchase.

Finally, averaging is not a substitute for position sizing. The method controls when money is invested, not how much of a portfolio it represents. An investor who averages into a position that is too large for their tolerance will still sell at the bottom, schedule or no schedule. The volatility record is the place to start that judgement, and it is a separate question from the one this page answers.

The interval is the one parameter an investor actually controls, and the evidence on it is less dramatic than the debate suggests. Moving from monthly to weekly contributions shortens the period over which the average is taken, which slightly reduces the benefit of averaging across a full cycle while slightly reducing the risk of buying everything at a local peak. The difference between the two is small relative to the difference between either and a single purchase. What matters far more is that the interval is long enough to be sustainable and short enough to complete within the horizon the investor actually has.

The more consequential choice is the horizon. A schedule that runs for six months is not really averaging; it is a single purchase spread thinly, and its outcome will be dominated by the price level of that half-year. A schedule that runs for several years crosses at least one full cycle and is genuinely exposed to the averaging effect. The market-cycle record gives a sense of how long those episodes have run, and therefore how long a schedule needs to be to span one.