Leveraged ETFs Explained: The Daily Reset and What It Costs

A fund built to deliver twice the daily return of an index fell 6 percent over four months in which that index gained 2 percent. Nothing inside the fund failed. It hit its stated objective on every one of those trading days, and the gap opened anyway. The figures are recorded by the SEC, and they are the plainest evidence that the multiple printed in a fund name describes a single session and makes no claim about anything longer.

The arithmetic behind that gap is short enough to work through by hand, and it points somewhere most explanations stop short of: the same reset that turned a rising index into a loss can also hand a holder more than the stated multiple. Which of the two arrives depends on the path the index takes, not on the fund.

What a Leveraged ETF Actually Promises, and Over What Period

A 2x fund undertakes to move twice as far as its benchmark between one closing price and the next. A 3x fund undertakes three times. The undertaking ends at that second close, and the following morning it starts again from wherever the fund happens to be.

That period is the whole of the promise, and it is the piece that goes missing when the product is described. A page that opens with one-year performance rankings has already invited the reader to read the multiple as a property of the year, which is the one reading the prospectus does not support.

The SEC states the position directly: these funds are built to hit their objective on a daily basis, and results over any longer stretch can differ substantially from the stated multiple. Read the name as a description of one session and the rest of the behaviour stops being surprising.

None of this makes the product broken. It makes it a different instrument from the one most people assume they are buying when they set out trading an index with a view measured in months.

The Daily Reset, Worked Through Two Days

To hold a constant multiple, a fund has to change its exposure whenever its own value changes. After a winning day it holds more; after a losing day it holds less. The exposure is rebuilt at each close so that the next session starts at exactly the stated multiple of the new, smaller or larger, asset base.

Take an index that gains 4 percent and then gives back 4 percent. It does not return to where it began: 1.04 multiplied by 0.96 is 0.9984, a loss of 0.16 percent. A 2x fund on the same two sessions moves plus 8 percent and then minus 8 percent, giving 1.08 multiplied by 0.92, or 0.9936. The fund is down 0.64 percent.

Twice the index loss would have been 0.32 percent. The fund lost double that again. No fee, no slippage and no tracking error is responsible; the entire difference comes from the second day being calculated on a base the first day moved.

This is the same compounding that makes gains and losses asymmetric anywhere else leverage appears, and it behaves the same way wherever leverage is tiered against position size. What changes here is that the rebuilding is automatic and happens every session whether the holder acts or not.

Why a Flat Market Can Still Cost You

Stretch those two days into six. Suppose an index alternates plus 5 percent and minus 5 percent for three full cycles. It ends down 0.75 percent, close enough to flat that a chart would show a sideways month.

A 2x fund over those same six sessions ends down 2.97 percent, against the 1.50 percent that twice the index return implies. A 3x fund ends down 6.60 percent against an implied 2.24 percent. The index barely moved and the 3x holder is down almost seven percent.

Now reverse the conditions. Suppose the index simply gains 5 percent on five consecutive sessions, ending up 27.63 percent. Twice that return is 55.26 percent, but the 2x fund ends up 61.05 percent. Three times is 82.88 percent, and the 3x fund ends up 101.14 percent. Both funds beat their multiple.

That second case is the part routinely left out, and leaving it out changes what the product looks like. The reset is not a fee dressed as arithmetic. It is a mechanism that pays a holder who is right about direction and continuity, and charges one who is right about direction but wrong about the route.

The drop from a peak behaves the same way. A fund carrying three times the exposure reaches a given maximum drawdown on a much smaller index move, and the deeper the hole the larger the index recovery needed to climb out of it.

Inverse Funds Are the Same Mechanism Pointed the Other Way

An inverse fund undertakes to move against its benchmark by a stated multiple, again between one close and the next, and again rebuilding afterwards. Nothing about the arithmetic changes because the sign changed.

What does change is the size of the effect a holder is exposed to. The reset works against a position whenever the benchmark reverses, and an inverse position is normally opened in exactly the conditions where reversals are frequent and large. The mechanism is at its most costly in the market that made the position attractive.

The SEC figures make the scale visible. Over the four months in which an index rose 2 percent and the 2x fund on it lost 6 percent, the 2x inverse fund on that same index lost 25 percent. On the second index, up around 8 percent, the 3x inverse fund lost 90 percent.

A 90 percent loss against an index that rose 8 percent is not a mistracked fund. It is compounding applied to a position that was reset against the benchmark every session for four months.

The Costs That Sit On Top of the Reset

Discussions of these funds tend to treat the expense ratio as the cost and the reset as a curiosity, and comparison tables ranking funds by fee reinforce that. The two sit on top of each other, and the fee is usually the smaller of the two.

The exposure is built with swaps, futures and other derivatives rather than by holding the index outright, which the SEC sets out plainly. Those instruments carry their own financing, and the rebuilding at each close means the fund trades every session regardless of what the holder does.

So the drag on a position has three parts. The published expense ratio, the financing embedded in the derivative exposure, and the path effect worked through above, which is the only one of the three that can also work for the holder. A comparison built on the first part alone measures the smallest term.

Anyone who has held a leveraged position financed overnight will recognise the second part; it is the same overnight cost that appears in how a CFD works, arriving through a fund structure instead of a broker statement.

Holding Periods: Where the Product Fits and Where It Stops Fitting

The useful question is not how long these funds may be held. It is what has to be true for the holding period to work, and that condition is about the shape of the move rather than the number of days.

A holder needs direction and continuity together. Direction alone is not enough, as the four-month figures show. FINRA puts the boundary at one trading session, stating that daily-reset funds are typically unsuitable for retail investors planning to hold beyond it, and singling out volatile markets as where this bites hardest.

Read the table by row rather than looking for a permitted number of days. A week of unbroken trend and a week of choppy grind produce opposite outcomes from an identical fund, and the calendar cannot tell them apart. The exposure carried also has to be small enough to survive the reversal that has not happened yet, which is the ordinary discipline of sizing a position rather than anything specific to funds.

Who These Are Not For

Anyone who cannot watch the position through the session. The product resets whether or not the holder is at the screen, and the reversal that does the damage does not wait for a convenient moment.

Anyone buying a view measured in months. The four-month figures on record are the direct answer to that plan, and no amount of conviction about the destination changes what the route costs on the way.

Anyone selecting between these funds on fee alone. The published expense ratio is real, and it is the smallest of the three drags described above, so a table ranked by it is answering a minor question.

Anyone relying on a past return to set an expectation. A one-year figure on a daily-reset product is a record of one specific path, and a different path over the same index move produces a different number entirely.

So the decision comes down to a single question, asked before the position is opened rather than after. If the plan depends on the index arriving somewhere, and the route it takes is not something you are watching and prepared to act on, the mismatch is with the instrument itself and no holding period fixes it.

If instead the plan is about one session, in one direction, with the position closed at the end of it, the product is doing the job it was designed for.