The tape doesn’t lie.

Systematic trend-following strategy across 80 futures instruments. Dynamic position sizing targeting constant annualised volatility. No discretionary overrides.

Two changes occurred at approximately the same time. First, the traded universe grew from 80 → 200 instruments. Second, account size increased from the USD 50k–150k range to the USD 150k–300k range (personal capital only — no outside investors). Position sizing scales proportionally with capital under the volatility-targeting framework, so strategy behaviour was unchanged; only the absolute size of positions increased. Strategy logic, risk targets, and all other parameters remained the same. The exact date is approximate — it was lost in an accidental git commit squash during a repo cleanup.

Traded universe grew from 265 → 269 instruments by promoting four candidates whose roll calendars were restored after a stale-instrument audit: ALUMINIUM_LME (LME aluminium), CAD2 (Canadian 2y bond), GILT5 (UK 5y bond), and MSCIEMASIA (Asia-only EM equity). Same day, seven permanently dead or duplicate instruments (SP500-GROWTH, SP500-VALUE, BRENT-LAST, BB3M, WATER-CALI, EUROFIRST100, NIKKEI_large) were silenced at the sampler layer to stop noisy CRITICAL roll alerts for markets that won't ever roll again. Strategy logic, vol target, forecast cap, and capital — all unchanged. Position sizing scales under vol targeting, so adding instruments spreads risk further; no individual position grows. Triggered by a ContractNotFound crash in the daily contract-sampling cron that surfaced 8 stale instruments simultaneously. URANIUM remains in the queue pending a 2021–2023 contract-history backfill.

The since-inception return, equity curve, Sharpe, and drawdown shown on this site are now computed directly from Interactive Brokers' daily NAV (Flex statement, EquitySummaryByReportDateInBase) as a deposit-adjusted time-weighted return — the broker's books are the canonical truth. The prior figures came from the pysystemtrade internal capital snapshot (DailyPandL), which historically ran a few points rich versus the broker's own reported TWR: at end-April 2026 the internal basis read about +30.7% against IB's reported 26.4%. The two bases have since reconverged, so this change barely moves today's headline (about +42.8% on either basis through mid-June) even though the same correction at end-May would have been roughly a six-point cut (about +38.8% down to +33.0%). This is a change of measurement basis, not of strategy or capital — strategy logic, vol target, and capital were all unchanged. Deposit days were at this point handled by zeroing the day's return so external cashflows were never counted as performance; because the whole deposit day was zeroed rather than the deposit amount alone isolated, the reconstructed series sat within roughly 0.4 percentage points of IB's own reported TWR over the periods where IB publishes a directly comparable figure. That handling was superseded on 2026-08-07; see below.

Any day carrying an external capital movement was previously recorded as a flat 0% day. That kept the movement out of performance, which was the intent, but it also discarded whatever the strategy did on that day. Dropping a day is not neutral: it can only flatter the record when the dropped day was a losing one, and at least one was. The calculation now nets the known movement out of that day's account change and keeps the trading result underneath it. This restated the since-inception return from +42.16% to +40.65% and year-to-date from +45.47% to +43.92%; the figures published on this site are the corrected ones. Three account-opening deposits remain recorded as flat days because each was larger than the account that received it — the denominator changes by up to 10x partway through those days, so no honest same-day return can be reconstructed from end-of-day values. This is a change of measurement basis, not of strategy or capital.

Strategy and risk framework unchanged from inception. Universe expanded most recently 2026-05-25; see event log for details.