Australia: FIFO, the 50% CGT discount, and the holding-period gate

Australia's tax engine has a single, large, load-bearing rule: capital gains on assets held more than 12 months are included in income at half their value (the "50% CGT discount"). For an optimizer choosing which lot to realise on which day, that step function across the 12-month gate dominates everything else. Combined with FIFO lot identification (no HIFO option) and the ATO's intent-based wash-sale guidance, the Australian engine rewards patience and timing over aggressive harvest.

Under §115-25, capital gains on assets held by an individual for more than 12 months are reduced by 50% before inclusion in assessable income. So a holder at the top marginal rate of 47% pays the full rate on short-term gains and half the rate (effectively 23.5%) on long-term gains. This isn't a graduated rate change, it's a discount on the amount included. The optimizer's tax penalty has to know about this directly:

The 50% CGT discount turns 12 months into a hard step function in the optimizer's tax penalty. A name 11 months 28 days into a long-term hold is materially more expensive to sell than the same name 12 months 2 days in.

The optimizer reads the holding period for every lot every day and applies the appropriate rate. When a name approaches the 12-month boundary, the solver will frequently delay a sell-the-gain trade by a few days — paying a small TE-budget cost in exchange for the discount kicking in. This is one of the most visible behaviours in the Australian backtests: clusters of "delay" trades concentrated around the 11.5–12.5 month boundary.

ATO practice accepts FIFO as the default for parcel identification on a sale of partial holdings, with specific identification permitted if records support it. The optimizer defaults to FIFO and exposes specific identification as an opt-in. Under specific identification the engine behaves more like the US HIFO engine; under FIFO it consumes the oldest parcels first, which has the side effect of skewing the long-term proportion of realised gain higher (because the oldest parcels are most likely past the 12-month gate).

Australia doesn't have a §1091-style bright-line wash-sale rule. Instead, ATO Taxation Ruling TR 2008/1 articulates an anti-avoidance position under Part IVA of the ITAA 1936: arrangements designed to obtain a tax benefit by realising a loss while preserving substantially the same economic position may be denied as wash sales. The test is intent-based; there's no number-of-days threshold.

• Parcel history (per-acquisition-date). Tracked per ticker, with both cost and acquisition date. Holding-period calculation runs every day.
• Marginal-rate input. The full marginal rate (e.g. 47% top bracket including Medicare levy). The 50% discount is applied inside the engine, not by the input.
• Wash-sale lock vector (soft). Default 30-day lock on harvested names; configurable per-account.
• Account type flag. Individual, SMSF, or trust — selects the discount factor (50%, 33.33%, or 50%) that the engine applies.

Two patterns dominate Australian backtests on the direct-indexing strategies:

• Delay realisation past 12 months. The optimizer routinely defers sells across the 12-month gate by a small number of days. The TE-budget cost is small relative to the 50% rate reduction.
• Lower harvest velocity. Without a HIFO option (under default FIFO), the optimizer's per-trade harvest opportunity is structurally smaller than the US engine sees. Australian backtests show roughly 2/3 of the US harvest stream at the same TE budget.

For the cross-jurisdiction comparison, see the multi-jurisdiction deep-dive.

Educational illustration · not advice. Australian tax practice is intent-based; coordinate any harvest-cadence variations with an Australian tax adviser.