Bollinger Band Mean Reversion on SPY
$10,000 invested on 2016-11-18 would be worth
$15,309Just holding SPY: $39,252
- Profit $5,309
- Holding made $23,943 more
- Total return
- +53.1%
- holding +293%
- Win rate
- 80%
- 36 won · 9 lost
- Trades
- 45
- in 10 years
- Max drawdown
- -29%
- -$3,423
- Profit factor
- 2.71
- won ÷ lost
- Test period
- +13%
- holding +69%
The verdict
Does Bollinger Band mean reversion work on SPY? Over 2016-11-18 to 2026-11-17 buying on a close below the lower Bollinger Band (20-day, 2 standard deviations) and selling on a close back above the middle band returned +53%, against +293% for simply holding SPY…
Right now: In a trade as of the 2026-11-17 close.
The rules
- 1Market: SPY (S&P 500 ETF), daily bars, long only
- 2Enter when the close is below the lower Bollinger Band (20-day, 2 std dev)
- 3Exit when the close is back above the middle band (20-day average)
- 4Signals on the close, filled at the next open
Growth over 10 years
The strategy against simply holding SPY, drawdown underneath. Everything after “Test period” was held back when the rules were judged.
Where the profit and loss came from
$10,000 from 2016-11-18 to 2026-11-17, after fees and slippage. Trailed holding SPY by $23,943.
- Net profit
- +$5,309
- +53%
- Total won (winning trades)
- +$8,664
- 36 trades
- Total lost (losing trades)
- −$3,356
- 9 trades
- Win rate
- 80%
- 36 won · 9 lost
- Average win
- +2.0%
- per winning trade
- Average loss
- -2.9%
- per losing trade
- Best trade
- +$674
- +6.6%
- Worst trade
- −$1,440
- -12.4%
- Biggest fall from a high
- −$3,423
- the worst point you would have sat through
- Average time in a trade
- 14 days
Hypothetical: the whole amount goes into each trade and profits are reinvested. Taxes and brokerage minimums are not included. Not advice.
Did it hold up on unseen data?
- Return
- +13.1%
- Buy & hold
- +69.3%
- Max drawdown
- -8.8%
- Win rate
- 73%
- Profit factor
- 2.48
- Sharpe
- 0.44
- Trades
- 15
Costs: 5 bps fee and 5 bps slippage per side. Signals on the close, filled at the next open, long only. Run on 2026-11-18T12:46:10.077Z with delayed daily prices. How backtests are run.
Strategy checks
How this idea holds up against the checks that separate skill from luck.
- 4passes
- 2watch
- 1fails
- 4your call
Coin toss
Passes80% wins vs 60% needed
Clearly above the break-even win rate for its reward-to-risk.
Outlier dependence
PassesWithout best 3: 28.7%
The result does not hinge on a few lucky trades.
Big-move days
Passes40% of gains on the top 5% days
Gains are not concentrated on the market’s wildest days. (A proxy: we do not yet use an events calendar.)
Time in market
Your call18%
Rarely invested: the rest of the time your money earns nothing, so compare with holding.
Sample size
Passes45 trades
Enough trades for the statistics to carry some weight.
Versus buy and hold
Watch53% vs 293%
Holding the same instrument did better, with no work.
Biggest drawdown
Watch-29.2%
Could you watch 29% of your account vanish and still follow the rules? Measured on closing prices, so intraday dips are not included.
Recovery time
Fails2.8 years
Longest stretch below a previous high, and it is still below one now.
Winning vs losing days
Your call55% of active days up
265 up days and 221 down days while invested.
Losing streaks
Your call1 trades, 6 days
The longest run of consecutive losers. Expect at least this again, and ask if you could keep going.
Day of week
Your callBest Thu 2.3%, worst Wed -1.4%
Average trade return by the weekday you entered. Small samples mislead, so treat differences as a question, not an answer.
Session timing, passive-index comparison and intraday drawdown are not measured yet.
Common questions
What traders usually ask about this strategy, answered from this backtest by AlphaEdgeCheck.
Does the Bollinger Band Mean Reversion on SPY make money?
Over 10 years of daily prices, $10,000 became $15,309 (+53%) after 0.05% fees and 0.05% slippage per side. So it made money, but simply holding SPY would have turned the same amount into $39,252.
Is the Bollinger Band Mean Reversion on SPY better than buy and hold?
No. It ended $23,943 behind simply holding SPY. It was invested only 18% of the time, so the money sat idle for much of the market's rise.
How many trades did it make, and is that enough?
45 trades in 10 years, 36 winners and 9 losers. That is enough for the statistics to carry some weight.
What was the worst loss?
The worst single trade lost $1,440 (−12%). The biggest fall from a previous high was $3,423 (−29%), and the longest losing streak was 1 trades in a row.
How much money does testing it first save?
With $10,000, trading it instead of holding would have left you $23,943 behind. The backtest showed that for free.
How did it do on data it was not fitted to?
We judge rules on the first 70% of the history and hold the last 30% back. In that held-out test period it returned +13% against +69% for holding. It passed 4 of 11 checks overall.
Can I use the Bollinger Band Mean Reversion on SPY with my broker?
Pick your broker in the section above and sign up for the beta, and we will email you the script for it when it is ready. It is research and education only: we do not connect to brokers or place orders, so check your broker's rules before automating anything.
Beta members
Get this strategy for your broker
Pick your broker or platform and sign up for the beta. We will email you the Bollinger Band Mean Reversion on SPY script for it. Research and education only: check your broker's rules before automating anything.
India
US and global
Platforms
Research, not advice. Results are hypothetical, use simplified assumptions and delayed daily prices. Past performance does not indicate future results. Risk disclosure.