Premium Market Analysis

Trader education, Trading Strategies

The Remarkable Performance Of MR5 Mean-Reversion Strategy Continues in 2017

The MR5 mean-reversion strategy is based on a formula from a graduate text on probability theory. Despite the low volatility and sluggish performance of most mean-reversion strategies in 2017, returns were 10.3% for SPY and 33.7% for UPRO.

Note: MR5 signals for SPY and UPRO are included in our Mean-reversion signal service.  In addition, MR5 strategy is used for trading portfolios of stocks. See MRDOW and MRSP100 strategies.

Performance from January 3, 2017 to December 29, 2017

Strategy Net Return* Max. DD Win rate Trades
MR5 (in SPY) +10.3% -1.3% 94.1% 17
MR5 (in UPRO) +32.6% -3.8% 94.1% 17

*Backtest results. Performance is net of $0.01 per share commission with no leverage and it is not annualized.

Equity curves for SPY and UPRO for 2017



Performance since ETF inception

Below are results of the backtested  performance of MR5 strategy in SPY and UPRO from inception of the ETFs to 12/29/2017. All backtests are based on fully invested equity and commission of $0.01 per share. There is no leverage.



Parameter SPY UPRO B&H SPY
CAR 8.83% 32.66%  9.23%
Return 728% 1010%  773%
Max. DD -26% -39%  -55%
Sharpe 0.63 0.99  0.50
Sortino 0.98 1.53  0.80
Exposure 42% 39%  100%
Win rate 71.3% 75.6%  –
Profit factor 1.91 2.45  –
CAR/MaxDD 0.34 0.84  0.17
Total trades 506 172  1
Avg. Bars in trades 6.2 5.9  –
Avg. Profit% 1.37% 3.71%  –
Avg. Loss% -1.88% -5.10%  –
Avg. Profit/Loss% 0.44% 1.56%  –
Best year 30.2% 73.7%
Worst year -3.1% 4.4%


The MR5 strategy has been profitable every year in UPRO since the ETF inception in 2009. In SPY the strategy has had only two consecutive down years in 2001 and 2002 with a combined loss of about -3.6%. Since 2003 the strategy is on a 15-year win streak. A profit table is included below.


While some strategy and signal vendors present validation results based on highly correlated securities, we instead validate our trading strategies on comparable and anti-correlated securities. In most cases over-fitted and random systems work well with highly correlated securities and that does not amount to a proper validation method. Below is the performance of the MR5 strategy in TLT ETF since inception and the 252-day correlation of SPY and TLT.

Although the average 252-day correlation between SPY and TLT since inception of the latter is -0.34, the MR5 strategy outperforms TLT buy and hold on a risk-adjusted MAR basis.

Warning: Mean-reversion trading carries high risk since there are no stops. Most traders allocate to it only a small percentage of available capital in the range of 5% to 10% and some leverage it accordingly.  Please read the disclaimer below.

MR5 signals for SPY and UPRO are included in our Mean-reversion signal service. 

Disclaimer: The strategies are provided as educational trading tools for informational purposes only and do not constitute investment advice. We do not warrant the accuracy, completeness, fitness or timeliness for any particular purposes of any strategy. Under no circumstances the strategies should be treated as financial advice. Note that performance results are based on backtests.  Before subscribing please read our Disclaimer and Terms and Conditions.

Copyright notice: Any unauthorized copy, reproduction, distribution, publication, display, modification, or transmission of any part of this report is strictly prohibited without prior written permission.