#79 — Core-pricing arithmetic — re-pricing the gross-positive family at OANDA core tier

2026-07-07

Verdict: negative · ⛔ core tier is WORSE, not better — the real cost unlock is a different venue entirely (~0.7p ECN)

OANDA US core (commission+razor-spread) tier is +0.3p worse all-in than spread-only on USD-quote majors and +1.0p worse on JPY crosses, plus a $10,000 minimum balance incompatible with the project's $8-$100 test accounts, so structural fade (+1.089p gross) and BB re-entry fade both stay net-negative or worse — the only real cost unlock found is an IC-Markets-style ECN account (~0.7p all-in, $200 min, 0.01 lot).

Hypothesis — what it tests

Whether re-pricing the project's gross-positive-but-net-spread-blocked strategies (structural fade, BB re-entry fade, regime-entry contrarian tilt) at OANDA US's core commission+raw-spread pricing tier flips any of their net verdicts from negative to positive.

Description

Pure measurement/arithmetic study, not a hypothesis test with gates: takes OANDA US's currently-published core-tier terms (commission $0.70 per 10,000 units per leg = $14/100k round-turn, core raw spread 'as low as 0.0', $10,000 minimum balance) and re-prices the project's family of strategies that were gross-positive but net-spread-blocked under the standard spread-only pricing assumption — structural fade (+1.089p gross, -0.890p net) and BB re-entry fade (backtest +0.5-1.1p net margin, later retracted — the whole edge was a phantom fill; corrected fill model = -1.11 p/trade) — using recorded per-trade figures only, with no backtests re-run. Finding: the flat USD commission converts to ~2.06 pips on JPY-quote pairs, more than eating back the ~1.1p spread markup OANDA removes, so all-in core-tier cost is higher than measured spread-only medians on every traded pair (+0.3p USD-quote majors, +1.0p JPY crosses); structural fade worsens from -0.89 to about -1.49p/trade and BB re-entry fade's thin backtest margin shrinks and flips negative on JPY legs (a margin since retracted outright: the backtest edge was a phantom fill, corrected to -1.11 p/trade live). No verdict flips from negative to positive under OANDA's own core tier, and the tier isn't even accessible at the project's live account sizes ($8-$100 vs the $10k minimum). The one identified real cost unlock is a different broker/venue entirely — an IC-Markets-style ECN account (~0.7p all-in, $200 minimum, 0.01 lot size) — not OANDA's own core tier.

Key result

Commission $0.70/10k units/leg = $14/100k round-turn current (legacy $5/side = $10 RT); core raw spread 'as low as 0.0', measured EUR/USD 0.4p; all-in RT current vs spread-only median: EUR_USD 1.80 vs 1.5 (+0.30), AUD_USD 1.60 vs 1.3 (+0.30), GBP_USD 2.10 vs 1.8 (+0.30), EUR_GBP 1.34 vs 1.4 (~flat), USD_JPY 2.56 vs 1.6 (+0.96), EUR_JPY 2.96 vs 2.0 (+0.96), GBP_JPY 3.96 vs 3.0 (+0.96), CHF_JPY 3.66 vs 2.7 (+0.96); structural fade gross +1.089p / net -0.890p (spread-only, implied RT cost 1.979p) -> net -1.49p at core current / -1.00p at legacy $5; BB re-entry fade backtest net +0.5-1.1p (later retracted — phantom fill; corrected fill model = -1.11 p/trade) -> would have been ~+0.1-0.2p EUR_USD at core current, JPY legs flip negative; regime-entry contrarian tilt gross +0.01-0.20p -- cost floor 7-130x the signal, hopeless at any tier; eligibility requires $10,000 maintained balance vs live accounts $8-$100 (010~$100, 001~$8.34); ECN alternative ~0.7p all-in, $200 minimum, 0.01 lot -- the only identified cost unlock.

Indicators

bid-ask spread (ask_c-bid_c)/pipcommission per 100,000 units (USD)pip value by quote currency

Algorithms

pip-equivalent commission conversionall-in round-trip cost arithmetic (raw spread + commission)net-margin re-pricing of recorded trade results

Code

research/experiments/core_pricing/RESULTS.md research/experiments/spread_clock/RESULTS.md research/experiments/spread_clock/spread_clock.py

Interactive version (search, filters, figures): the experiment explorer.

← #78 Spread clock — hour-of-week toll profiles + guard cost-benefit #80 Meta-allocation — equity-MA strategy switching vs equal-weight →

This page is one experiment from the audited record behind the book The Edge That Wasn't — get the book · about the project.