Bollinger Reversion · Gold (PAXG)

Buy the lower band, sell back at the mean. A range trade with an ugly tail risk.

Total return

-4.2%

10,000 → 9,577

vs buy & hold

-122.8%

over the same period 118.6%

Max drawdown

16.6%

10,000 → 8,803

Longest underwater

2039

2039 days · 98% of the period

Settings that beat it

0 / 16

of 16 parameter settings tested

Win rate

55%

95% CI 38%–71% · n=31

This lost to simply buying and holding.

Over 2020-09-01 → 2026-09-23 it returned -4.2% while buying once on the first day and never touching it returned 118.6% — 122.8% worse. Not one of the 16 parameter settings tested beat the benchmark. The median setting returned 0.0%.

It spent 2039 days below its previous high at the longest stretch — 98% of the whole period under water.

The win rate on this page is not a readable number. With 31 closed trades the 95% confidence interval runs 38%–71% — it spans a coin flip, so the headline 55% cannot be distinguished from chance at this sample size. It is shown because leaving it out would be worse, not because it supports a conclusion.

Price & fills

Equity

account equitybuy & holddashed = starting capital

62 fills. Signals are taken on the close and filled at the next open.

A 0% risk of ruin here is a property of the test, not a safety rating. These runs are long-only spot with no borrowed money, so there is no margin call and nothing can force the position closed. The account cannot be wiped out — it gets stuck holding the asset instead. Capital deployed (1×) and max drawdown (16.6%) are the numbers that carry the risk on this page. The same rules run on margin would be a different measurement, and this figure would not stay at zero.

Outcome distribution

A single return is one chosen start date. These are 62 runs of the same rules with the start shifted forward 30 days at a time.

Worst	p10	p25	Median	p75	p90	Best	Losing starts
-8.4%	-5.4%	-3.9%	-0.7%	2.3%	6.4%	8.7%	53%

Does it survive outside its best case?

The numbers above are one market on one timeframe. A rule that only works there did not work — the market did.

Timeframe

Same rules, same parameters, different candle size on Gold (PAXG).

Timeframe	Trades	Return	Buy & hold	vs B&H	Beat it	Trades
4H	13281	-38%	119%	-157%	no	243
1D ←	2214	-4%	119%	-123%	no	31
1W	316	8%	123%	-115%	no	2

Market — 0 of 3

Same rules, same parameters, different asset — daily candles.

Market	Return	Buy & hold	vs B&H	Beat it
BTC/USDT	-46%	530%	-575%	no
ETH/USDT	-53%	276%	-329%	no
Gold (PAXG) ←	-4%	119%	-123%	no

Does it survive its own parameters?

The headline above uses one parameter setting. Here is every setting tested: 16 combinations, median 0.0%, best 30.7%, worst -11.9%. 0 of them beat buy & hold.

A strategy that only works at one setting did not work — it was fitted.

Parameter grid

Bollinger Reversion run across its own parameter range over the full period. The same rules flip from profit to loss depending only on the numbers you picked — which is why a single headline return is meaningless on its own.

	length 10	length 20	length 30	length 50
mult 1.5	-12%	-11%	3%	20%
mult 2	-11%	-4%	8%	23%
mult 2.5	-8%	-0%	17%	31%
mult 3	0%	-2%	13%	26%

Hover a cell for trade count and drawdown. “·” = invalid combination (fast ≥ slow).

Year by year

Year	2021	2022	2023	2024	2025	2026
Return	-2%	-5%	-3%	2%	7%	-5%
Max DD	9%	7%	7%	6%	1%	9%
Trades	6	5	7	3	3	3

Recent closed trades

Closed	P&L (USDT)	Held
2026-07-04	-381.04	28 bars
2026-05-07	+178.20	8 bars
2026-04-01	-325.42	13 bars
2025-08-02	+167.81	2 bars
2025-05-21	+290.06	6 bars
2025-03-05	+192.35	4 bars
2024-12-27	+157.40	8 bars
2024-11-21	-42.00	14 bars
2024-02-21	+46.73	7 bars
2024-01-16	+159.41	5 bars
2023-11-17	+112.77	8 bars
2023-10-12	-60.88	14 bars
2023-08-24	-74.97	14 bars
2023-07-10	+30.40	11 bars
2023-06-17	-187.51	29 bars
2023-05-03	+111.23	10 bars
2023-03-02	-245.24	26 bars
2022-10-03	-11.23	17 bars
2022-09-13	-144.38	24 bars
2022-07-28	-217.50	22 bars
2022-06-17	+140.05	3 bars
2022-05-22	-306.50	26 bars
2021-10-04	+15.22	17 bars
2021-08-17	+81.20	10 bars
2021-07-04	-401.45	18 bars

What it is

Reference — Glowwiki: Bollinger Bands
The definitions and formulas below follow that article. It states its own limit plainly:
“Whether this produces profit is not addressed in that article.”
That is the question this page answers.

Note from the reference: The reference carries two mutually contradictory readings of the same line — buy a touch of the lower band (mean reversion) and buy a break above the upper band (breakout) — and does not choose between them. This page implements the mean-reversion reading. It also cites Lento et al. (2007), covering ten years from 1995 across several markets, which found no evidence of consistent performance over buy and hold.

Formulas

Bands

basis = SMAₙ ; upper/lower = basis ± k·σₙ

Defaults are n = 20 and k = 2

Standard deviation

σ = √( Σ(xᵢ − x̄)² / n )

Divisor is n — the population form. Not n − 1.

Established

Bollinger Bands place a volatility envelope a multiple of the standard deviation around a moving average. [3]
The method uses the population form of standard deviation: the proper divisor is n, not n − 1. [3]
Replacing the simple moving average with an exponential one requires changing both the basis and the standard deviation calculation, not just the basis. [3]

Conventional reading

Touching the lower band is conventionally read as a stretched move likely to revert to the mean. [1, 3]
The same line also supports the opposite rule — buying a break above the upper band as a breakout. The reference presents both and does not choose. This page implements the mean-reversion reading. [3]
Lento et al. (2007), covering ten years from 1995 across several markets, reported no evidence of consistent performance over a standard buy-and-hold approach — while the reversed, contrarian version did produce positive returns. [3]
The usefulness of technical analysis as a whole is contested by the efficient-market hypothesis. The reference article states the dispute without ruling on it — and neither does this page. What is below is one measurement, not a verdict. [1, 2]

Listed as convention, not as established fact.

Why your chart may not match This site divides by n when computing the standard deviation, which is what the original method specifies. Some charting platforms divide by n − 1 instead. With a 20-period window that makes the bands about 2.6% narrower here than on a platform using the sample form — enough to change when a touch of the lower band registers, and therefore enough to change the trades below.

Sources

기술적 분석 — Glowwiki (ko)
Technical analysis — Wikipedia (en)
Bollinger Bands — Wikipedia (en)

Assumptions

Period	2020-09-01 → 2026-09-23
Parameters	{"length":20,"mult":2}
Fees	0.10% taker + 0.05% slippage, both legs
Starting capital	10,000 USDT · long-only spot, no leverage
Rolling test	365 bars × 62 start dates
Input hash	357a3833e4d19a20

Other strategies · Gold (PAXG)

Buy & Hold Martingale Grid DCA MA Cross EMA Cross RSI Reversion MACD Cross Bollinger Reversion Donchian Breakout Supertrend Stochastic Cross Parabolic SAR