G10 Institutional Activity

01 — The scorecardTen currencies, one number each, and a band that decides whether the number is tradeable

This monitor ranks the ten G10 currencies on a single 0–100 composite, then converts that number into a tradeable bias. It is a relative-value instrument, not a forecast: the output is an ordering, and the useful information lives in the gaps between adjacent rows rather than in any one score.

Every score is a weighted average of five inputs read out of the currency’s own narrative — the central bank’s policy path, inflation against target, growth and the labour market, carry, and speculative positioning. The weights were recovered by fitting those inputs to the desk’s published scorecard, so this reconstruction reproduces the published numbers rather than approximating them: mean absolute error 0.46 points, and all ten bias bands reproduced exactly.

A currency earns a long bias by clearing 65 on the composite, which in practice requires three of the five pillars above neutral and at least one of them well above it. Four clear that bar this week and only one sits below the short line, so the table is top-heavy: the tradeable structure is a single pair at the extremes, not a portfolio of them.

02 — Composite scoresFour currencies clear the long threshold; one clears the short line

The distribution is unusually compressed in the middle. Nine of ten currencies sit between 54.2 and 72.1 — a 17.9-point range — while the franc sits 15 points below the next weakest. That shape matters more than the level: a tight middle means the ranking between, say, the krona and the loonie is inside the model’s own error bar and should not be traded, while the gap between either of them and the franc is not.

Three of the four long-bias currencies are there on policy and carry together: the RBA, Norges Bank and the ECB have all hiked into above-target inflation, and all three pay a positive rate. The yen is the exception and the most interesting row in the table. It scores 65.2 on the strength of policy normalisation and inflation, while its carry pillar sits at 40 — the second-lowest reading in the matrix — and its positioning pillar at 35. It is a long-bias currency that costs money to hold.

03 — Bias historyThe same side of the line has held for four consecutive prints

Bias in this framework is sticky by construction, because four of the five pillars move only when a central bank meets or a major data series prints. The history log shows the consequence: nine of ten currencies have been on the same side of the 50 line for all four prints in the window.

The practical reading is that a band flip is an event worth acting on, not noise to be waited out. Because the underlying pillars change slowly, a currency that crosses 65 or 44 has usually had a genuine change in policy or inflation behind it. The corollary is equally important: a currency drifting inside the neutral band is telling you nothing, and the honest response is to do nothing.

04 — How the score is builtFive pillars, and weights that were fitted rather than assumed

The five pillars are not equally weighted, and the weights are the part of this model most worth understanding, because they determine what actually moves a ranking.

Growth carries 29.7%, the largest single weight, which is initially counter-intuitive for a currency monitor. It is defensible: growth and the labour market are what give a central bank the option to hold rates high, so they gate the policy pillar rather than competing with it. Policy carries 25.7% and is the most immediate input — it is the only one that can reprice the whole curve in a single afternoon.

Positioning at 17.6% is the one input read contrarian at extremes, and it is deliberately the third-largest weight rather than a footnote. Carry at 16.4% is what separates a currency you can hold from one you can only rent. Inflation at 10.7% is the smallest weight, which surprises most readers and is worth stating plainly: on this model, an inflation print matters mainly through what it does to the policy path, not on its own.

05 — The pillar matrixWhere the ranking is actually decided

The matrix is the most useful single view in the monitor, because it shows the shape of a currency rather than its rank.

Two readings stand out. First, the carry column is where the bottom of the table is decided: the yen at 40 and the franc at 25 are the only sub-neutral carry readings in the G10, and they are precisely the two currencies the model struggles to rank on policy grounds. The BoJ is normalising and the SNB is not easing, yet neither can be held cheaply, and the composite penalises that.

Second, the positioning column is the only one that is not monotone with rank. The euro scores 68 there, the highest in the table, while the yen scores 35, the lowest. Both are long-bias currencies on policy. The difference is entirely about who already owns them, and it is the reason the euro is the cleaner expression of the same policy view.

06 — Positioning as a modifierWhy the desk reads CFTC extremes contrarian, and where that changes the trade

Positioning is the only pillar the desk reads in reverse. A large net short is not evidence that a currency is weak; at a one-year extreme it is evidence that the marginal seller has already sold, which is why it scores positively in this model.

The counter-example is the yen. It carries the weakest positioning score in the matrix at 35 because the speculative community is already crowded long, which caps the immediate upside even as the policy case improves. A currency can be right on policy and still be a poor entry — that is the single most useful thing the positioning pillar does.

07 — Constructing the tradeLong the top of the table against the bottom, sized by the divergence gap

The construction follows mechanically from the table. Go long the highest-scoring currency, short the lowest, and size the position to the divergence gap — the difference between the two scores. This week that is long AUD against short CHF, a gap of 33.0 points.

The veto rule is doing real work this week. Both legs of the AUD/CHF pair are positioned favourably — the Australian dollar is a one-year-extreme short and the franc is not crowded — which is what makes this the pair the model wants rather than merely the pair it produces. Had the top of the table been the yen, the same 33-point gap would have been a materially worse trade.

08 — Currency notesThe evidence behind each row, in the desk’s own words

Each row below is the evidence the score was built from, in the order the pillars are weighted. Read the policy sentence first, then the inflation print, then check the positioning line — that sequence reproduces the ranking.

AUD · Australian Dollar 72.1 Long Bias

The RBA's unanimous 29 September hike to 4.60% and its explicit willingness to tighten further strengthen the Australian dollar's hawkish carry support. Bullock said financial conditions must be tight enough to curb a persistent underlying-inflation pattern and still sees the labour market as a little tight, though 4.6% unemployment and weaker housing keep the pace data-dependent. The 29 September CFTC report shows a one-year-extreme 63,239-contract net short, preserving upside squeeze risk.

NOK · Norwegian Krone 70.6 Long Bias

Norges Bank's 24 September hike to 4.50% strengthens the krone's carry support and confirms that above-target inflation is still driving policy. The bank may raise rates again if inflation proves persistent, although its forecast now points to a period of holding rates high. The bias remains bullish.

EUR · Euro 67.7 Long Bias

The ECB remains restrictive after its September hike, and the 2 October flash estimate strengthens the near-term inflation case: headline HICP accelerated to 3.8%, core to 2.5% and services to 3.2%. Much of the surge still comes from energy, while meeting-by-meeting guidance limits conviction on the next move. The 29 September CFTC report shows euro speculative positioning at a 63,256-contract net short, near the cleanest point of the past year, reinforcing the bullish lean.

JPY · Japanese Yen 65.2 Long Bias

The BoJ's 18 September hike and 1 October Summary of Opinions establish further normalisation as the base case, while the 2 October Tokyo CPI surge to 2.7% core and 3.0% core-core materially strengthens the near-term inflation case. A still-tight labour market reinforces that policy support, although crowded long positioning and weak carry temper the immediate upside. The scorecard bias is now bullish.

USD · US Dollar 63.1 Neutral

The Fed's September hike and still-hawkish guidance keep the dollar's policy backdrop restrictive, but the 2 October jobs report materially weakens the follow-through case: payrolls rose only 29k, unemployment reached 4.2% and wages slowed to 3.0% y/y. With core PCE also softer than expected, the dollar's relative policy advantage has narrowed sharply even though inflation remains above target. The bias is now neutral.

GBP · British Pound 62.8 Neutral

The BoE kept rates unchanged, but the split vote and its warning about persistent inflation show that further tightening has not been ruled out. Stronger consumer spending provides some reassurance, though weakness in employment and wages keeps the outlook uneven. The 29 September CFTC report shows sterling positioning at a 91,075-contract net short, leaving a clean squeeze setup and a modest upside lean.

NZD · New Zealand Dollar 57.5 Neutral

Growth has held up better than feared, reducing the immediate recession risk and supporting the RBNZ's willingness to keep policy restrictive. High unemployment still tempers that support, while the 29 September CFTC report shows the speculative net short extending to 17,315 contracts. That is cleaner than a fortnight ago but still relatively crowded over a one-year horizon, so the bias remains neutral.

SEK · Swedish Krona 54.4 Neutral

The Riksbank's 24 September forecast now points to hikes beginning this year as growth improves and underlying price pressure persists, giving the krona a firmer policy story. August employment also increased, although unemployment remains elevated and measured inflation is still low. The bias remains neutral pending a clearer data follow-through.

CAD · Canadian Dollar 54.2 Neutral

Governor Macklem's 21 September remarks highlighted a stronger recent recovery and rising risk that energy costs broaden into persistent inflation, keeping a future Bank of Canada hike in play. Renewed US trade uncertainty, excess supply and weak August employment still cap the policy advantage, while the risk-on backdrop offers some support. The bias remains neutral.

CHF · Swiss Franc 39.1 Short Bias

The SNB's 24 September decision leaves rates at zero and calls current policy appropriate, while its inflation outlook remains subdued despite higher energy costs. That removes some easing pressure, but the franc still lacks carry support and the broader relative setup remains bearish.

09 — Method, limits and sourcesWhat this monitor can and cannot tell you

Four limitations are worth stating plainly, because a ranking instrument invites more confidence than it deserves.

• It is a relative-value tool. A score of 72.1 says nothing about where the Australian dollar goes against the dollar; it says it is the most attractive of ten currencies relative to each other.
• Pillar scores are judgements. They are read out of prose by an analyst, not computed from a data feed. Two careful readers can differ by several points on any single pillar, and the composite inherits that.
• The weights are fitted, not given. They are stable enough to be useful and are not the desk’s actual coefficients. Treat the ranking as robust and the exact point values as indicative.
• It lags by construction. Four of five pillars move only on scheduled events, so this monitor is a weekly instrument. Do not run it intraday and expect it to mean anything.

Central bank decisions, statements and meeting minutes from the RBA, Norges Bank, ECB, Bank of Japan, Federal Reserve, Bank of England, Reserve Bank of New Zealand, Riksbank, Bank of Canada and Swiss National Bank. Inflation prints from the relevant national statistics agencies, including the euro-area flash HICP estimate and the Tokyo CPI. Labour market and wage data from national statistical releases. Speculative positioning from the Commodity Futures Trading Commission Commitments of Traders report for 29 September 2026. All figures are as at 2 October 2026 unless stated otherwise.

10 — DisclosureConflicts, methodology status and terms