gold technical analysis

Reading Gold's Technical Structure: What 58 Years of Monthly Data Actually Says

Most gold technical advice is never tested against the same data as buy and hold. We re-implemented four classic rules on 706 monthly gold closes from 1968 and found that holding beat all of them, that the 50/200 rule carried no information at all, and that "sell when overbought" points exactly the wrong way on this asset.

By Auriatech Research Desk · · Last updated 2026-11-11 · 3 min read
technical analysismoving averagesRSIbacktestgoldmarket structure Contains our own backtest data
What this is. A reconstruction, not a broker screenshot. Four technical rules were re-implemented from their published definitions and run over 706 monthly gold closes (January 1968 – October 2026) taken from the World Bank's Pink Sheet commodity price data, last updated November 3, 2026. Every rule, parameter and number is specified below so you can rebuild it and disagree with it.

There is a particular kind of gold conversation that never resolves. One trader says the 200-month moving average is the signal that matters; another says RSI at 70 is a short; a third says breakouts are the only real edge. Each of them can show you a chart. None of them will put their rule on the same data as buy and hold, over the same period, with the same costs.

This article does that, and the result is not flattering to the rules.

The short version

Over 706 months:

  • Buy and hold returned 8.40% a year. All four rules returned less.
  • The 12-month Donchian breakout came closest at 8.25%, and it did one thing better than holding: it cut the maximum drawdown from 62.07% to 41.54% and produced the best Sharpe ratio of the set (0.61 against 0.56).
  • The 50/200 moving average crossover carried no information at all — every confidence interval crosses zero, in both eras tested.
  • “Sell when overbought” is backwards on gold. Months with RSI(14) above 70 were followed by an average 12-month return of +25.26%, against +6.27% for the average month. Months with RSI below 30 were followed by −5.19%, with only 20.8% of them positive.

That last point is the one worth sitting with, because it is the rule most gold advice is built on.

The data

Item Value
Source World Bank Pink Sheet, Monthly Prices, gold column (col 70)
Price basis Monthly average, USD per troy ounce
Headline sample 706 months, 1968-01 to 2026-10
Full sample 802 months, 1960-01 to 2026-10
Data updated November 3, 2026
Rules tested 4
Forward horizons 3, 6 and 12 months
Comparisons 24 headline rule × horizon tests, plus 12 RSI threshold tests

Why the sample starts in 1968

The official gold price was legally fixed at $35 per ounce from 1960-01 to 1967-12 in this series; the free London market did not open until 1968. A technical rule evaluated against an administratively frozen price is measuring the price controls, not the market. The 96 pegged months are therefore excluded from the headline statistics and reported separately as a control — and the full 802-month series is published alongside so the effect of that choice is visible rather than hidden.

Including the peg era lowers every return (buy and hold drops from 8.40% to 7.98% a year), because it adds eight years of flat $35 gold. It does not change the ranking of the rules.

The rules, defined precisely

Each rule is a long-or-flat system. You hold gold when the condition is true and hold cash when it is not. No shorting, no leverage, no position sizing, no compounding of capital across accounts.

1. The 10/50 moving average. Long when the 10-month moving average is above the 50-month moving average.

2. The 50/200 moving average. Long when the 50-month moving average is above the 200-month moving average.

3. The 12-month Donchian breakout. Long when the month’s closing price exceeds the highest close of the previous 12 months; flat when it falls below the lowest close of the previous 12 months. This is the original Turtle channel, unchanged.

4. RSI(14) tail avoidance. Long only when RSI(14) is between 30 and 70 — that is, when the indicator is not calling a top or a bottom.

RSI(14) uses Wilder’s smoothing on monthly average prices, not on daily closes, because the whole study is on monthly data.

What each rule actually did

System Return/yr Total Vol Sharpe Max DD Time in market Longest underwater
Buy and hold 8.40% 114.0x 16.86% 0.56 −62.07% 100% 315 mo
10/50 MA 7.32% 63.6x 14.91% 0.55 −59.22% 65.6% 333 mo
50/200 MA 6.04% 31.3x 14.23% 0.48 −55.70% 64.9% 333 mo
12-month Donchian 8.25% 105.6x 14.83% 0.61 −41.54% 65.6% 309 mo
RSI(14) 30–70 4.18% 11.1x 11.70% 0.41 −46.32% 69.8% 285 mo

Buy and hold’s maximum drawdown ran from January 1980 to July 1999 — 235 months from peak to trough, and 315 months before the account was back at its high. That is the number that matters most for position sizing, and it is why the 41.54% drawdown of the breakout rule is worth something even though its return was lower.

The pattern in the table is consistent. Every rule spends a substantial share of its time in cash — between a third and a half — and in this sample the months those rules stepped aside from were mostly months gold went up. Sitting in cash is a real position with a real opportunity cost. It is the single biggest reason all four rules end up behind buy and hold.

Do the signals say anything about the future?

Return comparisons on their own can be misleading. A rule that stayed flat through the 1970s and 1980s and long afterwards could look respectable for reasons that have nothing to do with the signal.

The stronger question is: when the rule says “be in gold”, what did gold do over the next 3, 6 and 12 months — compared with what it did after the months when the rule said “be in cash”?

The 10/50 moving average

After a month where the 10-month average was above the 50-month, gold averaged +15.05% over the following 12 months. After a month where it was below, gold averaged +4.13%. A gap of +10.92 percentage points, with a 95% bootstrap interval of +1.06 to +21.04 that excludes zero.

At 3 months the gap is +2.32 points, with an interval of −0.21 to +4.89 — which crosses zero. So the signal is weak at short horizons and only appears when you look a year out.

The 12-month breakout

After a breakout month: +15.13% over 12 months, against +3.97% otherwise. A gap of +11.16 points, interval +1.95 to +21.21. At 6 months the gap is +6.42 points with an interval of +1.77 to +11.31.

The 50/200 moving average

After a signal month: +11.86% over 12 months, against +10.08% otherwise. A gap of +1.78 points with an interval of −10.01 to +14.67 — it crosses zero, and comfortably so. At 6 months the gap is +1.25 points with an interval of −5.10 to +6.95. There is no horizon at which this rule produced a reliable difference.

There is also a structural reason to distrust it here: a 200-month average needs 16 years of history before it produces its first reading, so on a 706-month sample it is only evaluated over a fraction of the data.

The RSI-tail rule

After a neutral-RSI month, gold averaged +7.28% over 12 months. After a month outside the 30–70 band, it averaged +20.62%. The gap of −13.34 points runs the wrong way for the rule: the months the RSI rule tells you to step aside from are the months gold went up most.

The overbought problem, in detail

This is the part with the widest reach, because “sell gold when RSI is overbought” is close to universal retail gold advice.

Condition at the signal month Months Avg forward 12m Average month Gap 95% interval Share positive
RSI(14) above 70 181 +25.26% +6.27% +18.99 pp +5.94 to +34.42 76.2%
RSI(14) above 60 307 +21.71% +2.91% +18.80 pp +9.15 to +30.25 74.6%
RSI(14) below 40 120 +3.58% +12.82% −9.25 pp −14.54 to −3.58 53.3%
RSI(14) below 30 24 −5.19% +11.81% −16.99 pp −19.46 to −14.54 20.8%

Read the table twice, because it says the same thing in both directions:

  • A month that closed with RSI above 70 was followed by an average 12-month return of +25.26%, and more than three quarters of those months were positive. At the 6-month horizon it was +12.02% against +2.70%.
  • A month that closed with RSI below 30 was followed by an average 12-month return of −5.19%, and only 20.8% of those months were positive. At the 6-month horizon it was −0.89%.

On gold, overbought has meant strength and oversold has meant continued weakness. That is the opposite of the mean-reversion story the indicator is usually sold with. A rule that sold the overbought months and bought the oversold ones would have been a persistent loser on this asset — not a marginal one, and not an unlucky one.

This is not a quirk of gold’s 1980s spike. The RSI-above-60 variant, with 307 monthly observations instead of 181, shows the same thing: +21.71% against +2.91%.

And it explains the RSI system’s return

Holding gold only when RSI sat between 30 and 70 returned 4.18% a year, roughly half of buy and hold. The forward test shows exactly why: the band you are avoiding is where the returns are made. There is no mystery here — the RSI rule is a systematic way of being out of the market during gold’s strongest months.

Why the obvious p-values are wrong

This study runs 24 rule-by-horizon comparisons plus 12 RSI threshold tests. Reading a 0.05 threshold across that many tests produces roughly one false positive by construction, before any of the statistical subtleties below are even considered. The Bonferroni-adjusted threshold for the headline set is 0.0021.

Three tests are reported for every comparison, and they disagree on purpose:

The Welch t-test assumes the observations are independent. Forward windows overlap — twelve consecutive 12-month windows share eleven months of price history — so this test’s p-values are systematically too small. For the 10/50 rule at 12 months it reports p = 3.3e-10, which is not a number to act on.

The moving-block bootstrap resamples contiguous blocks of months, preserving the serial correlation that a month-by-month resample would destroy. This is the interval to read for the size of the effect.

The non-overlapping subsample keeps only every h-th starting month, so the observations genuinely do not share price history. It is the most conservative test and usually the most instructive. For the 10/50 rule at 12 months it gives p = 0.104 on 38 signal and 20 no-signal observations. For the breakout rule, p = 0.207. For the 50/200 rule, p = 0.774.

Read the interval and the non-overlapping p-value. The t-test is printed so you can see how much it overstates.

The era split is where the rules get tested

Full-sample results can be carried by a single decade. The same 12-month forward test was re-run on 2000–2026 only:

Rule Era Gap 95% interval Crosses zero?
10/50 MA 1968–2026 +10.92 pp +1.06 to +21.04 No
10/50 MA 2000–2026 +12.35 pp +3.38 to +20.99 No
Donchian 12m 1968–2026 +11.16 pp +1.95 to +21.21 No
Donchian 12m 2000–2026 +5.06 pp −5.12 to +15.43 Yes
50/200 MA 1968–2026 +1.78 pp −10.01 to +14.67 Yes
50/200 MA 2000–2026 +1.42 pp −6.85 to +9.96 Yes
RSI 30–70 1968–2026 −13.34 pp −28.09 to −0.57 No
RSI 30–70 2000–2026 −7.07 pp −17.96 to +2.82 Yes

Now the picture is much more honest:

  • The 10/50 moving average holds up. Same direction, similar size, no overlap in time. That is the strongest evidence in the study.
  • The breakout rule does not survive the split. Its full-sample interval excludes zero, but its post-2000 interval does not. The full-sample result is carried by the earlier decades.
  • The 50/200 rule never worked. No horizon, no era.
  • The RSI-tail rule is a consistent loss in the full sample, and becomes merely uncertain in the later era.

One result surviving out of four is the realistic outcome, and it is worth saying plainly what the survivor is: the 10/50 moving-average state carries information in the direction of “gold keeps going up in a trend”. That is a description of gold’s long bull market as much as it is a tradable signal, it is not significant once overlapping observations are removed, and it certainly is not a reason to expect a better return than holding.

Where gold stands against these rules right now

At the end of October 2026:

Indicator Value
Gold monthly average $4,105 per ounce
10-month MA $4,507
50-month MA $2,904
200-month MA $1,811
RSI(14) 57.8

The price is below its own 10-month average, which is a pullback inside the trend state: the 10-month remains well above the 50-month, and the 50-month is far above the 200-month. On this study’s evidence, that configuration — the 10/50 state — is the one associated with better forward returns, and the pullback has not yet broken it.

RSI(14) at 57.8 sits inside the 30–70 band that this study finds has the weakest forward returns. Over the full sample only 24 months printed RSI below 30 and 270 printed above 70 — on monthly data the extremes are rare, which is worth remembering before treating either as a signal.

What this study is not

It measures monthly average prices, so every rule is evaluated as though it could be entered exactly at a monthly average and held to the next one. A real order fills at some intraday price and pays a spread, and a rule that trades often — the breakout rule especially — pays more of that than a rule that does not. No transaction cost, financing, storage or tax is included anywhere, so all four rules are if anything flattered.

It does not test whether these rules work on intraday gold, on other instruments, or on today’s price series. A rule that works on monthly averages may be unusable on a tick chart, and the reverse is also possible.

And it does not test whether a readable version of the framework is worth using. “Above the 10-month” is a description of a market state that you can act on. What this data shows is that acting on it did not make you money relative to holding — it made you miss the best months.

The only claim being made is narrow and specific: on 706 monthly gold closes from 1968-01 to 2026-10, holding beat all four rules on return, the 10/50 moving-average state carried forward information in both eras tested, the breakout rule’s result did not survive the era split, the 50/200 rule did nothing, and the RSI overbought rule pointed the wrong way.

If you want to use technical structure anyway

None of this says stop looking at the chart. It says know what you are looking at:

  • Treat the moving-average state as a description of the regime, not a trigger. “Gold is in an uptrend” is a useful thing to know. “Therefore I should buy” is a claim this data does not support, because the flat months that rule skips are the months gold went up.
  • Use the drawdown numbers for sizing, not for entries. The breakout rule’s 41.54% drawdown against holding’s 62.07% is the part of these results that changes what you do. If your account can fund one drawdown but not the other, that is the trade being made, and it is a rational one.
  • Be especially careful with RSI on this asset. If you are going to trade against an overbought reading on gold, you should know that you are trading against the historical tendency of the strongest months to follow the strongest momentum readings.
  • Check the sample your rule was tested on. A rule tuned on 2020–2026 data on gold is tuned on a monotonic uptrend. The 50/200 rule is a perfect example of a signal that looks compelling on a chart and carries no information over a real cycle.

Educational content only. Not investment advice. Past performance is not indicative of future results. Trading forex and CFDs carries substantial risk of loss.

Frequently asked questions

Do technical rules work on gold?

Not if the benchmark is holding. Over 706 monthly closes from 1968 to 2026, buy and hold returned 8.40% a year; the best of four classic rules returned 8.25%. The rules did one thing better than holding: the 12-month Donchian breakout cut the maximum drawdown from 62.07% to 41.54% and produced the best Sharpe ratio of the set. That is a trade of return for drawdown depth, not an edge.

Which technical rule is the most reliable on gold?

The 10-month-above-50-month moving average. When it was in that state, the following 12 months averaged 15.05% against 4.13% when it was not, with a bootstrap interval that excludes zero and the same gap in the 2000-2026 era. It is still not statistically significant on the non-overlapping subsample, and it describes gold's long bull market more than it describes an exploitable signal.

Is the 50/200 moving average crossover worth using on gold?

The data here says no. It showed no reliable difference in forward returns at any horizon in either era, with every confidence interval crossing zero. The 200-month average needs roughly 16 years of history before it produces a first reading, so on a 706-month sample it is evaluated over a small fraction of the data.

Should I sell gold when RSI is overbought?

On this sample that instruction loses money. Months with RSI(14) above 70 were followed by an average 12-month return of 25.26%, against 6.27% for the average month, and 76.2% of them were positive. Months with RSI below 30 were followed by an average 12-month return of -5.19%, with only 20.8% positive. Overbought has meant strength on gold, and oversold has meant continued weakness.

How do I know if a technical rule is real?

Three tests, all published with the numbers here: a Welch test for the naive p-value, a moving-block bootstrap confidence interval that preserves the serial correlation between adjacent months, and a non-overlapping subsample test that keeps only every h-th starting month. Read the interval and the non-overlapping p-value, not the t-test. With 24 rule-by-horizon comparisons, the correct threshold is the Bonferroni-adjusted 0.0021, not 0.05.

Where is gold relative to its indicators right now?

At the end of October 2026 the monthly average was $4,105 per ounce, below the 10-month average of $4,507 but far above the 50-month at $2,904 and the 200-month at $1,811. RSI(14) was 57.8, inside the 30-70 band that this study finds has the weakest forward returns.

Sources

  1. World Bank — Commodity Price Data (The Pink Sheet), monthly, updated November 3, 2026 — accessed 2026-11-11
  2. World Bank — CMO Historical Data Monthly (direct download used for this analysis) — accessed 2026-11-11
  3. Wilder, J. (1978) — New Concepts in Technical Trading Systems — accessed 2026-11-11
  4. Brock, Lakonishok & LeBaron (1992) — Simple Technical Trading Rules and the Stochastic Properties of Stock Returns, Journal of Finance 47(5), 1731-1764 — accessed 2026-11-11
  5. Lo, A. W. (2004) — The Adaptive Markets Hypothesis: Market Efficiency from an Evolutionary Perspective, Journal of Portfolio Management 30(4), 75-94 — accessed 2026-11-11

Use the free tools

Open-source calculators — no signup required.

Use the free tools
Risk disclosure — read before acting on anything above

All content on this site is provided for educational and informational purposes only. Nothing published here constitutes investment advice, a recommendation, or an offer to buy or sell any financial instrument. Trading foreign exchange, CFDs, and precious metals involves a high level of risk and may result in the loss of your entire capital. Past performance of any strategy, robot, or expert advisor (EA) — whether simulated, backtested, or live — is not indicative of future results. You are solely responsible for your trading decisions. If you are unsure whether a strategy is suitable for you, seek independent advice from a licensed financial professional in your jurisdiction.

Educational content only. Not investment advice. Trading forex and CFDs carries substantial risk of loss.