Gold-silver ratio: chart since 2013, average and what moves it

The gold-silver ratio (GSR) is the gold price divided by the silver price, both in US dollars per troy ounce: how many ounces of silver buy one ounce of gold. At the 14 October 2026 UTC close it was 73.78 (4,111.62 ÷ 55.73), the highest close of 2026 but below the 77.90 median since July 2013.

That was the day the September US consumer price index came out, and silver fell more than five times as far as gold: 5.51% against 1.04%.

How to calculate the gold to silver ratio

ISO 4217 defines XAU and XAG as one troy ounce each, so dividing the two prices cancels the units. That division is all any gold to silver ratio calculator does.

Both metals trade nearly around the clock on weekdays, so the gold-silver ratio today depends on when you divide. On 5 February 2026 it was 67.35 at 22:00 UTC, 5 p.m. in New York, and 70.04 two hours later at midnight UTC, where TickerLayer’s gold and silver daily bars close.

For a live ratio, divide the mid prices of the latest quotes from GET /commodities/quote/XAUUSD and GET /commodities/quote/XAGUSD, as the gold price API guide’s script does. The gold price (XAUUSD) and silver price (XAGUSD) pages show each price with a delay.

Gold-silver ratio chart and history since 2013

Since July 2013, four daily closes in five fell between 65.17 and 89.18, around a median of 77.90: the three guides on the chart.

The high came in March 2020 and the low in January 2026. Yet the decade ended near where it began: on 14 October the ratio closed at 71.88 in 2016, 76.40 in 2021, 80.12 in 2025 and 73.78 in 2026.

What is a normal gold to silver ratio?

No level is good or bad in itself, so normal can only mean common: since July 2013, 69% of sessions closed between 70 and 90, and 73.78 sits in the 70 to 80 band.

The average gold to silver ratio depends on the window

The mean was 77.99 since 2013, 80.66 over ten years and 63.30 in 2026, so the same 73.78 is low for the decade and the top of this year.

Before 2013: 15 to 1 by law, and about 7.9 to 1 at today’s mines

TickerLayer gold and silver daily bars line up from 15 July 2013, so this page draws no 100-year chart. US law once fixed the ratio: 15 to 1 in the Coinage Act of 1792, and about 16 to 1 from 1834, when a new Act cut the gold in the $10 eagle. Mine output gives a ratio of tonnes, not prices: about 7.9 tonnes of silver per tonne of gold in 2025, by USGS estimates.

2026: from 45.88 in January to the year’s high on 14 October

The year opened at 60.53. By 28 January silver had gained 64.7% and gold 27.4%, both to their highest closes in these bars (117.70 and 5,510.73), and the ratio had hit 45.88 the day before.

Then silver lost about a quarter of its price twice in six days. On 30 January it fell 27.62% and gold 10.18%, taking the ratio from 46.24 to 57.38; on 5 February silver fell 25.07% and gold 6.16%, and the ratio rose 25.2% to 70.04, its largest one-day rise in these bars.

What moved the ratio on 14 October

Silver did, mostly after 12:00 UTC: down 0.86% at noon, it lost 4.69% more by the close, while gold lost 0.38% and then 0.67%. The September US consumer price index came out at 12:30 UTC; hourly bars cannot separate the half hour before it from the one after, so this is timing, not cause.

Why the gold-silver ratio tends to rise when metals fall

Silver amplifies gold. On daily log returns from July 2013 to October 2026, gold’s annualized volatility was 15.9% and silver’s 31.8%, twice as much, with a 0.79 correlation, so silver’s beta to gold was 1.58 (0.79 × 2.0). On average, each 1% move in gold came with a 1.58% move in silver, and the ratio moved about 0.6% the other way. Silver is also an industrial metal: by USGS estimates, electrical and electronics took 25% of US silver use in 2025 and photovoltaics 15%, against 18% for bars.

Day by day, the ratio moved the opposite way to silver on 79.4% of sessions and to gold on 59.3%, and it rose on 70.6% of the 357 sessions gold fell 1% or more. In 2026 silver’s volatility reached 73.7% against gold’s 30.8%, a beta of 2.05. To compare the 0.79 with stocks, currencies and crypto, add silver to a correlation matrix.

Copper-gold ratio, oil-gold ratio and the platinum metals

The same division works for any pair: at the 14 October close, gold/palladium stood at its highest in these bars.

The platinum price in USD (XPTUSD) closed above gold on 387 sessions, all between July 2013 and January 2015; palladium did on 951 sessions between December 2018 and December 2022.

COPPERUSD, at 6.55, sits at the level of a per-pound quote (about 14,400 dollars a tonne), so an ounce of gold bought about 628 pounds of copper, and the table multiplies the copper-gold ratio by 1,000. Here the oil-gold ratio is gold over Brent, whose daily bars cover the whole window since July 2013: 38.80 barrels an ounce, reading Brent per barrel as it is conventionally quoted. WTI vs Brent compares the two crudes.

Compute the gold-silver ratio from daily bars in Python

gold_silver_ratio.py recomputes the averages, volatilities and beta with two requests to the aggregate bars endpoint. It needs Python 3.9 or newer, nothing outside the standard library, and a TICKERLAYER_API_KEY environment variable.

"""Gold-silver ratio from TickerLayer daily bars. Python 3.9+, standard library only."""
import json
import math
import os
import statistics
import sys
import urllib.error
import urllib.parse
import urllib.request
from datetime import date, datetime, timezone

API = "https://api.tickerlayer.com"
KEY = os.environ.get("TICKERLAYER_API_KEY") or sys.exit("Set TICKERLAYER_API_KEY first.")


def get(path, **params):
    """GET a REST path; on an HTTP error, exit with the server's message."""
    url = f"{API}{path}?{urllib.parse.urlencode(params)}"
    req = urllib.request.Request(url, headers={"x-api-key": KEY})
    try:
        with urllib.request.urlopen(req, timeout=30) as resp:
            return json.load(resp)
    except urllib.error.HTTPError as err:  # 403: no Commodities feed on this key
        sys.exit(f"HTTP {err.code}: {err.read().decode()}")


def daily_closes(symbol, start, end):
    """UTC daily closes on weekdays, following next_offset to the end of the range."""
    path = f"/commodities/agg/{symbol}/1/day/{start}/{end}"
    params, closes = {"sort": "asc", "limit": 5000, "offset": 0}, {}
    while True:
        body = get(path, **params)
        for bar in body["results"]:
            day = datetime.fromtimestamp(bar["t"] / 1000, tz=timezone.utc).date()
            if day.weekday() < 5:  # drop Saturday and Sunday rows
                closes[day] = bar["c"]  # closes only: one stray tick can set a high or a low
        if body.get("next_offset") is None:
            return closes
        params["offset"] = body["next_offset"]


def annualized(returns):
    """Annualized volatility of daily returns, in percent."""
    return 100 * statistics.stdev(returns) * math.sqrt(252)


# Ask for one day before the first day you need: a daily range can drop its first date.
gold = daily_closes("XAUUSD", "2013-07-13", "2026-10-14")
silver = daily_closes("XAGUSD", "2013-07-13", "2026-10-14")
ratio = {d: gold[d] / silver[d] for d in sorted(gold) if d in silver}
days = list(ratio)
last_day = days[-1]
last = ratio[last_day]

print(f"{last_day}: gold {gold[last_day]:.2f} / silver {silver[last_day]:.2f} = {last:.2f}")
print("from        sessions   mean  median   p10    p90    low   high  share below")
for start in (days[0], date(2016, 10, 14), date(2021, 10, 14), date(2026, 1, 1)):
    window = [ratio[d] for d in days if d >= start]
    p = statistics.quantiles(window, n=10, method="inclusive")
    below = 100 * sum(v < last for v in window) / len(window)
    print(f"{start}  {len(window):>7}  {statistics.mean(window):5.2f}  {statistics.median(window):6.2f}"
          f"  {p[0]:5.2f}  {p[-1]:5.2f}  {min(window):5.2f}  {max(window):6.2f}  {below:9.1f}%")

low_day = min(days, key=ratio.get)
high_day = max(days, key=ratio.get)
print(f"lowest {ratio[low_day]:.2f} on {low_day}, highest {ratio[high_day]:.2f} on {high_day}")

# Daily log returns on consecutive sessions; skip pairs that straddle a gap in the data.
pairs = [(a, b) for a, b in zip(days, days[1:]) if (b - a).days <= 4]
rg = [math.log(gold[b] / gold[a]) for a, b in pairs]
rs = [math.log(silver[b] / silver[a]) for a, b in pairs]
mg, ms = statistics.mean(rg), statistics.mean(rs)
beta = sum((g - mg) * (s - ms) for g, s in zip(rg, rs)) / sum((g - mg) ** 2 for g in rg)
print(f"annualized volatility: gold {annualized(rg):.1f}%, silver {annualized(rs):.1f}%, "
      f"silver beta to gold {beta:.2f}")

# The ratio rises on any day silver's return is below gold's.
down = [(g, s) for g, s in zip(rg, rs) if g <= math.log(0.99)]  # gold down 1% or more
rose = sum(s < g for g, s in down)
print(f"gold fell 1% or more on {len(down)} sessions; the ratio rose on {100 * rose / len(down):.1f}%")

2026-10-14: gold 4111.62 / silver 55.73 = 73.78
from        sessions   mean  median   p10    p90    low   high  share below
2013-07-15     3350  77.99   77.90  65.17  89.18  45.88  122.34       34.0%
2016-10-14     2510  80.66   81.10  68.21  90.48  45.88  122.34       22.7%
2021-10-14     1299  81.03   83.43  65.46  90.48  45.88  104.02       17.9%
2026-01-01      204  63.30   63.95  56.40  69.31  45.88   73.78       99.5%
lowest 45.88 on 2026-01-27, highest 122.34 on 2020-03-18
annualized volatility: gold 15.9%, silver 31.8%, silver beta to gold 1.58
gold fell 1% or more on 357 sessions; the ratio rose on 70.6%

share below is the percentage of sessions that closed lower than the latest close. Swap XAGUSD for XPTUSD or XPDUSD to divide gold by platinum or palladium (the output still says silver), and mind their earlier daily close.