GBP WATCH score 3

macro weather · read 2026-09-24 17:25:10Z · ← back to the dashboard · refreshes every 15 min

Why it has this rating

Its pairs' realised daily range is 1.6x their own 60-day median (regime 1.5-2.0x; scored 1). Over the last ~5 trading days GBP moved -0.8% against the book; the haven-minus-risk tilt is +2.2% — GBP is a risk currency, and the tilt is against it (scored 1). Scheduled in the next 48 hours: BOE Gov Bailey Speaks (Fri 09:15Z) (scored 1). The bot stopped out 1 time on its pairs in the last 7 days (GBPCHF; net -391). That is shown, not scored: in the replay, after 0 / 2 / 3 stop-outs on a currency in 7 days: +0.096 / +0.118 / +0.181 R/trade — recent stops are not a reason to bench a dip-buyer.

What that means for the live bot

Keep trading — no bench

Because score 3 sits below the warning line (4). At this regime (1.5-2.0x) the replay's forward R was +0.072 per trade — lower than normal but still positive, so a bench here costs R rather than saving it.

Until: re-check at the next reading; bench the driver pairs if the score reaches 4.

GBPUSD are the pairs pushing the range

This is the monitor's rule, quoting its own replay — not a live LLM judgement. The prompt below is for the deeper look.

The pairs

pair	range / 60d median	call
GBPUSD	2.39x	driver
GBPCHF	1.49x	—
GBPAUD	1.37x	—
GBPNZD	1.07x	—

The evidence behind the rule

Two-year replay of the support-bounce backtest, 4,128 currency-days, no lookahead: the currency's pairs' R over the next 14 days by that morning's range regime. risk currencies under a >= 1% haven bid: -0.046 R/trade vs +0.120 otherwise (n=21 ccy-days). Stop-outs: after 0 / 2 / 3 stop-outs on a currency in 7 days: +0.096 / +0.118 / +0.181 R/trade.

range regime	currency-days	forward R / trade
< 1.0x	1,915	+0.090
1.0-1.5x	1,641	+0.110
1.5-2.0x	403	+0.072
2.0-2.5x	110	+0.056
> 2.5x	58	-0.092

Ask your LLM — the deeper look

Macro check on the support-bounce bot for GBP (score 3, WATCH): 1 stop-out on GBPCHF in 7 days (net -391; not scored - the replay says stops are not a reason to bench); daily range 1.6x its 60-day median across 4 pairs; next 48h: BOE Gov Bailey Speaks (Fri 09:15Z); safe-haven bid: CHF/JPY/USD +2.2% vs AUD/NZD/GBP over ~5 days. Pairs: GBPAUD, GBPCHF, GBPNZD, GBPUSD. Look at what is driving this currency right now and tell me whether its pairs should be benched, for roughly how long, and what would end the bench. Ground it in the bot's own numbers (live/files/pivct_trades.csv, STATUS.md §2) and say plainly if the evidence is too thin to act on.

To bench a pair: remove it from the universe in the milk store (piv_universe.json) — the sync writes live/files/piv_universe.csv and restarts the bot. Ask the LLM session to do it and to log the bench and its exit condition in STATUS.md §2.

Sources at this reading: {"feed_s": 11, "parquet_days": 6.7, "recorder_days": 5, "intraday": "minute recorder", "calendar": "ok", "trades_7d": 7}