updated Sat, 17 Oct 2026 23:17:08 GMT · refreshes ≤5 min
AfterPrime spread monitor
Afterprime live half-spread vs measured floor, last 30min
leg
live hs (bp)
floor (bp)
ratio
status
stale
USDJPY
—
0.13
—
stale
EURUSD
—
0.17
—
stale
EURJPY
—
0.17
—
stale
GBPUSD
—
0.23
—
stale
GBPJPY
—
0.21
—
stale
AUDUSD
—
0.21
—
stale
AUDJPY
—
0.18
—
stale
NZDUSD
—
0.34
—
stale
NZDJPY
—
0.27
—
stale
USDCHF
—
0.19
—
stale
CHFJPY
—
0.10
—
stale
USDCAD
—
0.14
—
stale
CADJPY
—
0.18
—
stale
All legs near the calm-market floor — the cost basis behind the EW6-ZH / ZAK revival holds right now. Floor = median half-spread measured 2026-08-06 (commission-free). "wide" (>3× floor) is
the stress signal that closes the cross-cost advantage; "elevated" (>1.5×) is a watch.
AfterPrime live fills
min-lot real fills vs the measured floor (UTC)
Success = fills land at ≈ the measured AfterPrime half-spread with no last-look rejects.
slip = fill vs the touch we quoted at send (+ = worse). Kill cap −$50 cumulative; "KILLED" latches entries off.
positioning — CFTC COT CME non-commercial net, z vs 156wk
ccy
net
Δwk
z-score
%ile
as of
GBP
-106,964
-9,353
-1.98
0.6%
extreme
2026-10-13
AUD
-117,299
-18,711
-1.65
0.6%
extreme
2026-10-13
CHF
-22,078
+1,662
0.72
78.2%
2026-10-13
JPY
50,207
-12,118
0.68
71.2%
2026-10-13
NZD
-11,698
-925
-0.51
29.5%
2022-02-01
EUR
1,635
+2,857
-0.29
48.7%
2026-10-13
CAD
-94,765
-4,602
-0.03
45.5%
2026-10-13
JPY: spec net vs price vs strength net (bars) · USD/JPY + JPY-strength LOO (lines, shape-normalised)
net longnet shortUSD/JPYJPY strength (LOO)
Net = non-commercial long − short (negative = net short the currency). z / %ile are vs the trailing
156-week window; |z|≥1.5 = crowded positioning (a weak, ~3-day-lagged contrarian context — not a
timing signal). Source: CFTC COT (Legacy futures-only) + FRED DEXJPUS + G8 JPY strength.
cross-venue arb board
No overlapping quotes yet. Populates when ≥2 venues quote the same symbol
in the same window (weekend = crypto only; full board after the FX open).
Lag = seconds this venue's mid trails its region's consensus (positive = slower feed = the stale/latency-arb target). Same-region only, so network path is netted out.