ApexAlgo FX

Real Account + Hedges = the real trades exactly as they happened (closed + open marked at 2026-10-23 20:13 UTC) plus every modeled option: premiums paid, sell-backs when the group closed, intrinsic at expiry when it did not. Worst case charges the full measured bid/ask spread on every option. With 30-Day Forced Close is the rule-enforced model used in the rest of this page: it also closes any doubled position still open at option expiry. Real + Rolling Hedges keeps those positions open as they really were and, at each expiry, buys a fresh 30-day 1% OTM strangle centred on that day's spot, on the lots still open that day, at that day's IV, until the position closes (table below). Options are modeled, not traded.

Identical model logic runs on Account I (live) and Account II (demo). A double is a same-pair, same-direction scale-in of exactly 2× a still-open leg. At each double a 30-day 1% OTM strangle (put 1% below / call 1% above the double-open spot) is bought lot-for-lot on the total position. IV: events with a real Saxo 1M ATM quote near the double are marked "quote" and are measured. IV logging only starts 2026-05-03, so earlier events (45 of 50) are marked "est*" and priced from an event-time estimate, not a flat constant — see the footnote below the table. Treat "est*" premiums as estimates. Execution: all published figures price entry and sell-back at mid — the midpoint of Saxo's recorded bid/ask. A worst-case variant charging the measured per-pair half-spread (median from recorded Saxo 1M tradable quotes, e.g. EURUSD 0.20, USDCAD 0.25, EURAUD 0.34, USDMXN 0.49 vol points) on entry and again on sell-back is shown in the Execution sensitivity box below. 30-day rule (enforced): any doubled position still open at option expiry is force-closed in the model at the spot price on the expiry date, with the option settled at intrinsic value. Real spot fills are never altered; the actual-vs-model divergence is reported per event below. Force-closes count as Category C events. The "single option" variant prices only the protective leg (put for longs, call for shorts) under the same rules.

Execution sensitivity: the published overlay figures assume mid-price execution — orders worked between Saxo's bid and ask. The worst case (every option bought at the full retail ask and sold back at the full bid) gives: Account I straddle -$27,267 (published mid +$13,421), single option -$11,235 (+$8,232); Account II straddle -$13,558 (+$8,738), single +$3,492 (+$13,925). Real fills land between the two bounds — and the ranking flips with execution quality: at mid the straddle's reversion leg is profitable and the straddle beats the single option; crossing the full spread, the single option wins.

IV sensitivity (β=1.0): 45 of Account I's 50 doubles predate the Saxo IV logger and are priced by the estimator described under the doubles table, using a measured β=0.37. If implied vol instead tracked realized one-for-one (β=1.0 — the upper bound, since β was fitted in a calm window), Account I straddle becomes +$3,311 (published +$13,421) and net hedge cost -$63,184 (published -$73,294); Account II straddle +$3,889 (+$8,738). The overlay's headline is materially more sensitive to this assumption than to execution on the 5 measured events, so it is stated here rather than buried in the model JSON.

Loss cap, stated precisely: loss on a doubled position is capped at ~3% of the doubled notional plus the hedge premium, enforced by the mandatory close at option expiry. It is not a flat "3%" of the account. At current sizing the worst-case single event is $61,843 (GBPCAD 13.05 lots, 34.9% of account equity at event time).

Concurrency: maximum simultaneous open doubles: 4. Overlapping doubles sharing a currency: GBPCAD + GBPCAD (2025-01-10, shared CAD/GBP); GBPCAD + GBPCAD (2025-01-10, shared CAD/GBP); GBPCAD + GBPCAD (2025-01-10, shared CAD/GBP); GBPCAD + GBPCAD (2025-01-10, shared CAD/GBP); GBPCAD + GBPCAD (2025-01-10, shared CAD/GBP); GBPCAD + GBPCAD (2025-01-10, shared CAD/GBP); EURAUD + GBPAUD (2025-03-17, shared AUD); USDJPY + USDCHF (2025-04-03, shared USD); USDJPY + USDCAD (2025-04-03, shared USD); USDCHF + USDCAD (2025-04-03, shared USD); EURAUD + EURUSD (2026-04-10, shared EUR); CHFJPY + CHFJPY (2026-04-20, shared CHF/JPY); USDCAD + USDCAD (2026-04-27, shared CAD/USD). Worst-case exposure per event = 3% of doubled notional + hedge premium, shown against account equity at event time in the table above.

These positions were (or still are) held past their option's expiry on the real account, violating the strategy's 30-day rule. All headline figures above use the model close at expiry; the actual outcomes are shown for full disclosure. Positive divergence means the real account did better than the enforced model — i.e. the account was compensated for running unhedged risk the strategy does not permit.

* est = no Saxo quote exists near that event (IV logging starts 2026-05-03), so IV is estimated as ivMean × (RVt / RVbase)β: the pair's mean recorded Saxo 1M IV, scaled by how far EWMA(λ=0.97) realized vol on the day sat above or below its level during the logging window. β = 0.37 is the measured elasticity of log implied to log realized (R² 0.19, n=2,966), not an assumption. Anchoring to the recorded mean keeps the level right in normal vol; the RV term supplies the regime response a flat mean cannot — the 2025-04-03 USDCHF double (tariff announcement) now prices at 7.4% (vs a 6.6% calm-market mean). This is still an estimate. β is fitted on a window containing no crisis, and implied vol spikes super-linearly in a real dislocation, so these premiums are more likely too low than too high. The β=1.0 row in the sensitivity box is the upper bound. Hedge P&L is settlement minus premium (negative = the hedge cost money). Account II's line-by-line hedge table is on the Account II page.