EdgeStacker
Library
Mean reversion · GLD · 1D

Does Williams %R bounce actually work on GLD? We tested it

We ran Williams %R bounce on GLD through walk-forward testing and a 100-run random-entry gauntlet. 4 of 4 unseen periods positive. Beat 82 of 100 random-entry runs.

WALK-FORWARD VERIFIED EDGESTACKER ENGINE
Return +29.1%
Win rate 72.2%
Max drawdown −12.6%
Trades 36
Sharpe 0.47
Equity curve

Equity curve over Aug 2016 to Jul 2026. Hover any point for its value. Turn on buy and hold to compare.

PeriodWindowReturnSharpeTrades
1 Aug 2018 to Jul 2020 +10.9% 1.11 5
2 Jul 2020 to Jul 2022 +2.0% 0.24 4
3 Jul 2022 to Jul 2024 +3.3% 0.38 10
4 Jul 2024 to Jul 2026 +11.6% 0.65 12
4 of 4 periods positive. Mean Sharpe 0.59. Worst period +2.0%.
Market matrix
Edge held on 12 of 28

The same rules, tested on 28 tickers. Edge held on 12. We ship it on those.

GLD SHIPPED
82/100 EDGE
META
100/100 EDGE
NFLX
98/100 EDGE
UBER
98/100 EDGE
XOM
96/100 EDGE
DIA
93/100 EDGE
JPM
93/100 EDGE
NVDA
92/100 EDGE
SPY
91/100 EDGE
QQQ
81/100 EDGE
AVGO
80/100 EDGE
MU
79/100 EDGE
Show every ticker we tested
RIOT
81/100 NONE
AMD
67/100 NONE
BA
65/100 NONE
IWM
62/100 NONE
GOOGL
58/100 NONE
MSTR
58/100 NONE
MSFT
56/100 NONE
COIN
55/100 NONE
AMZN
52/100 NONE
AAPL
47/100 NONE
SOFI
37/100 NONE
TSLA
12/100 NONE
MARA
10/100 NONE
SMCI
7/100 NONE
HOOD
too new TOO NEW
PLTR
too new TOO NEW
The rules
Entrywhen WILLR of close over 14 bars crosses above -90 AND close is above SMA of close over 200 bars
Exitwhen a signal fires (WILLR of close over 14 bars is above -30)
Sizing**Position size:** no fixed stop distance to size against. This strategy exits on signal only. Size by conviction, or add a hard stop for a size-by-risk number.
Diagnosis

This strategy made money overall, turning a 29% gain on GLD with a high win rate of 72%, meaning most trades were winners. However, the ride was lumpy, as shown by a low Sharpe ratio of 0.47 and a worst drawdown of -12.56% that lasted over two months from May to July 2026. That drawdown period was where the strategy hurt most, and it suggests the system can suffer extended losing stretches even though individual trades are often profitable. The result is not driven by just a few lucky trades, since the 36 trades spread the gains around, and the entry timing edge is strong: real entries beat 82% of random entries, so the entry signal itself is carrying the result rather than the exits. The consistency figure shows the edge appeared steadily, as the strategy was profitable in all four walk-forward test periods, meaning the profit did not come from one lucky stretch.

Run this on your own account.

Sign up free. EdgeStacker runs a fresh backtest for you, then alerts you on the Today screen when the setup fires.

Test it yourself, free
Backtested results, not live trading returns. Fees and slippage included. Not investment advice.