An old-fashioned scale with a glowing golden Bitcoin coin on one pan, and translucent ledger pages with percentage symbols floating on the other pan, tipping the balance, against a dark background with candlestick charts behind
Original illustration generated with AI (Adobe Firefly) — not a representation of real data.

Taxes as a Strategy Cost: Why Deferral Is a Structural Edge, Not a Footnote

NodeWitness Team 5 min read

In the 2,255-strategy tournament, rotating positions turned gross winners into net losers. Gold is the extreme case (+1.1% gross, -0.6% net), but the median tax drag across champion strategies ran 19-22% of gross return. Taxes are a real strategy cost, not a footnote.

Almost no backtest circulating online subtracts taxes. The trading strategy tournament covered in this series did, with a tax model fixed in writing before computing anything, and the result reshapes the underlying argument: buy-and-hold, or DCA, aren't just simpler than rotating positions. They're also cheaper on taxes, in a way that can wipe out a more active strategy's gross edge entirely.

The tax model, explained before the numbers

The experiment applied Spanish progressive savings-income brackets (19% up to €6,000, 21% up to €50,000, 23% up to €200,000, 27% up to €300,000, 28% above that) over a calendar tax year, netting every gain and loss realized in a given year before applying the bracket. Net losses from a year carry forward up to 4 years against future gains, always consuming the oldest losses first; once that window passes, whatever remains uncompensated expires. Buy-and-hold and DCA get their own rule: gains stay unrealized, untaxed, until the end of the partition, at which point a hypothetical full liquidation is assumed, so the comparison with rotating strategies is apples-to-apples. Both gross AND net results are always reported, never just one. More on this in NodeWitness's own guide to Bitcoin taxes in Spain, though the tournament models a general tax regime, not one specific to crypto.

The gold case, and why it happens

The two best-performing strategies on gold closed validation at +1.1% and +1.0% gross return. Net: -0.6% and -0.7%. A tax drag of 158% and 174% of the total gain. The mechanism is simple once you see it: the strategy got taxed on gains realized in its good years, while losses realized in the bad years that followed carried forward without fully offsetting them inside the partition's window. This isn't a calculation error or a cherry-picked pathological case. It's what mechanically happens when every realization gets taxed on its own instead of waiting for the end.

Four bars: E1 and E2 on gold with positive gross return (+1.1% and +1.0%) and negative net return (-0.6% and -0.7%), with a note on the 158% and 174% tax drag on total gross gain
Original diagram built from the experiment's own verified CSV output: the tax drag flips the sign of the result for both strategies.

Two cases you can check by hand

Two buy-and-hold examples, to see the mechanism without trusting an aggregate figure. Bitcoin in validation generated a gain of 4,394 on a 10,000 position, entirely within the first bracket: an exact tax bill of 834.87, a friction of precisely 19.00%. Ether generated a gain of 21,258, which does cross into the next bracket: 1,140 in the first bracket (19% of 6,000) plus 3,204 in the second (21% of the rest), 4,344 total. Neither of these rotated even once. This is just taxation applied to a single sale at the end of the period, and it's a useful reference point for what happens to a strategy that does rotate.

The median: gold is the extreme, not the norm

The gold case makes the headline because it flips the sign of the result, but it's not representative of the typical cost. Across every champion-asset combination with valid data, the median tax drag runs roughly 19-22% of gross return: a real slice of the gain, with no extreme case required, that any rotating strategy has to account for before it can claim to beat buy-and-hold.

Deferral as a structural edge

Buy-and-hold and DCA don't avoid taxes. They defer them. That difference, paying now, year after year, versus paying once at the end, isn't a matter of form. It's a structural edge any strategy that rotates has to overcome before it produces real value, exactly the same way real perpetual funding is a cost the long/short package carries on its own long legs, as covered in the previous piece of this series. The honest caveat: at the simulated scale (10,000, with USDT/EUR parity assumed as a documented simplification), almost everything falls inside the first 19% bracket. What this result measures is the effect of deferring versus rotating, not the effect of tax progressivity itself.

There's a second effect worth naming, separate from progressivity, that the experiment doesn't isolate but that pushes in the same direction. A dollar of tax paid five years from now is worth less today than the same dollar paid next year, even before inflation, because that money keeps compounding in the meantime instead of already sitting with the tax authority. Rotating doesn't just generate more taxable events. It moves each one of them earlier relative to buy-and-hold, and that acceleration carries a real opportunity cost even if the marginal rate were identical in both cases. A strategy that ties, gross, still loses on this basis alone, and a strategy that only edges buy-and-hold out slightly on a gross basis can end up behind on a net, present-value basis once that timing cost is added on top of the realization-event cost already measured in the tournament's own numbers. Neither cost shows up in a gross backtest at all, which is part of why gross comparisons keep circulating unchallenged.

Nothing in this article is tax or investment advice, and nothing on NodeWitness is. For your own situation, consult a professional. You can read the tournament's full verdict, why the short side loses even while collecting funding, or NodeWitness's own guide to Bitcoin taxes in Spain.

Last updated: August 31, 2026