σ-Normalized Currency Strength
The problem with raw percent
Suppose USD/JPY moves 0.30% in an hour and EUR/CHF moves 0.12%. Which one is the bigger event?
Raw percentage says USD/JPY, by a factor of two and a half. But USD/JPY routinely moves 0.30% in an hour on a quiet Tuesday, while EUR/CHF sitting still is its natural state — 0.12% for EUR/CHF may be the largest hourly move it has made in a week. Ranked by raw percent, the yen crosses and the Swiss crosses are simply not on the same scale, and any "strongest currency" list built that way just re-discovers which pairs are volatile.
This is the single reason most currency-strength meters feel noisy: they rank volatility, then call it strength.
What the z-score does
For each pair and each time window, the site measures the pair's own recent distribution of moves, then expresses the current move in units of that distribution:
z = (current move − mean move) ÷ standard deviation of moves
The result is dimensionless. A z of +2 means the move is two standard deviations above that pair's own average — a genuinely unusual event for that pair. A z of +2 on EUR/CHF and a z of +2 on GBP/JPY are now the same statement about surprise, even though the underlying percentages differ by several times.
| Reading | Rough interpretation |
|---|---|
| |z| < 0.5σ | Noise. This pair is doing what it always does. |
| 0.5–1.0σ | Mild directional pressure. |
| 1.0–2.0σ | A real move for this pair. Worth a look. |
| 2.0–3.0σ | Unusual. Typically news, a session open, or a stop cascade. |
| > 3.0σ | The war-feed threshold. The map fires an ⚔️ assault report. |
From pair z-scores to currency strength
A single pair tells you about a relationship, not a currency. EUR/USD rising could be euro strength or dollar weakness, and the pair alone cannot distinguish them.
So the map takes each currency and averages the z-scores of all its fronts, with the sign flipped where the currency is the quote rather than the base. USD's strength is its signed average against EUR, GBP, JPY, CHF, CAD, AUD, NZD and BTC. If the dollar is rising against every one of them, its average is strongly positive and the circle is large. If it is rising only against the yen and falling elsewhere, the average is near zero — which correctly tells you the story is yen weakness, not dollar strength.
This is the difference between the map and a single chart: a broad average cannot be faked by one loud opponent.
Reading the map
- Circle size — σ-normalized strength. Big circle, broad winner.
- Tracer fire — the animated flow along a front, moving toward the stronger side of that specific pair.
- Time window — switch it in the top bar. Short windows show session dynamics; longer windows show the trend that a swing trade would actually ride.
- Hovering a front — shows all three numbers at once: raw percentage change, the σ score, and that pair's own 1-minute σ base, so you can see exactly what the normalization did.
- The σ toggle — turn normalization off and ranking reverts to equal-weight raw percentage. Useful as a sanity check, and a fast way to see how much the normalization is changing the picture.
To practice turning a strength reading into concrete orders, the major forex pairs order generator applies the same strength-first logic across all 28 pairs and outputs a LONG / SHORT / no-entry list you can compare against your own read of the map.
Two honest pitfalls
1. Normalization dampens the hour after big news. When NFP or CPI lands, every pair's measured volatility jumps. Because σ is the denominator, the moves that follow are divided by a now-larger number and can read as oddly calm for roughly an hour. This is inherent to the method, not a bug — but do not read "quiet" as "safe" in that window.
2. Strength is descriptive, not predictive. Knowing that the dollar is +1.8σ tells you what has already happened. It does not tell you whether it continues. That question is what the reversal radar and COT positioning exist to address — and even together they produce probabilities, not certainties.