Live composite indexes — currencies, equities, commodities, bonds & risk sentimentUpdated 3m ago
G10 currencies ranked by cross-pair performance
Each G10 currency's strength is calculated by averaging its contribution across all 7 pairs where it appears. The key rule: when a currency is the base (first in the pair), the pair's % change applies directly. When it's the quote (second in the pair), the change is inverted.
If AUD/NZD drops −0.75%, this means AUD weakened against NZD. For AUD (base): the contribution is −0.75% (pair fell = AUD got weaker). For NZD (quote): the contribution is +0.75% (pair fell = NZD got stronger). This is correct because AUD/NZD falling means each AUD buys fewer NZD — AUD lost value, NZD gained.
When you expand a currency bar, each pair row shows:
The overall index score is the simple average of all 7 contributions. A currency that is weak across all its pairs will have a strongly negative score; one that is strong across all pairs will be strongly positive.
Traditional risk-on/risk-off classifications rely on textbook assumptions (e.g. "Gold is always a safe haven"). We reject this approach. Instead, we measure how each asset actually co-moves with VIX (the CBOE Volatility Index) over a rolling 90-day window using Pearson correlation of daily returnssourced from our own internal CorrelationCache database. VIX is the market's real-time fear gauge — if an asset truly acts as a risk-off instrument, it must measurably rise when VIX rises. The baskets auto-rebalance every time our correlation matrix is refreshed.
For each trading day t, we calculate the simple daily return:
Where Pt is the closing price on day t and Pt-1 is the closing price on the previous trading day. Weekend days (Saturday/Sunday) are excluded for cross-asset calendar alignment.
Given n overlapping trading days between VIX and Asset X, with return vectors x = [RVIX,1, ..., RVIX,n] and y = [RX,1, ..., RX,n]:
Result ranges from −1.0 (perfect inverse) to +1.0 (perfect co-movement). A value near 0 indicates no linear relationship. Minimum 5 overlapping data points required.
Using 5 hypothetical trading days to illustrate the math:
| Day | VIX Return (x) | S&P Return (y) | x·y | x² | y² |
|---|---|---|---|---|---|
| 1 | +0.050 | −0.020 | −0.001000 | 0.002500 | 0.000400 |
| 2 | −0.030 | +0.015 | −0.000450 | 0.000900 | 0.000225 |
| 3 | +0.080 | −0.035 | −0.002800 | 0.006400 | 0.001225 |
| 4 | −0.040 | +0.025 | −0.001000 | 0.001600 | 0.000625 |
| 5 | +0.020 | −0.010 | −0.000200 | 0.000400 | 0.000100 |
| Σ | +0.080 | −0.025 | −0.005450 | 0.011800 | 0.002575 |
This near-perfect −1.0 confirms the strong inverse relationship: when VIX rises (fear increases), S&P 500 falls. Our actual 90-day measured ρ reflects real-world noise and non-linear dynamics that reduce perfect correlation.
Once baskets are established, the Risk Sentiment Score for any timeframe is:
DailyPriceHistoryCorrelationCache