Premium Market Analysis, Trader Education, Software, and Trading Strategies

Trading Strategies

PSI5 Mean-Reversion Performance Update-June 9, 2023

Photo by

This article provides an update on the performance of the PSI5 strategy applied in convergent (mean-reversion) mode to SPY, QQQ ETFs, and Dow 30 stocks. Long-term performance results are also included, from the inception of the ETFs and since 1993 for Dow 30 stocks.

For ETFs, we use all available data since inception, and for Dow 30 stocks, data starting on 01/29/1993, while taking into account delistings available by Norgate Data.

The PSI5 mean-reversion strategy is not data-mined but based on a formula from a text in probability theory that models price action. The strategy is long-only because it takes advantage of the mean-reversion dynamic in price action for timing price action reversals. The strategy can also be used in divergent (trend-following) mode. Click here for more details. For recent divergent mode results click here.

Note that the PSI5 strategy uses only one parameter for the lookback period and remains the same for all backtests and updates we provide.

Backtest details for SPY and QQQ

Time frame: Daily (adjusted data)
Strategy type: Long-only, no bear market filters.
Backtest period: Inception to 06/09/2023
Commission per share: $0.01
Position size: Fully invested
Position entry and exit: Open of next bar
Stop-loss: None

SPY ETF performance (backtest)

The strategy is up 7.8% year-to-date after losing 5.3% in 2022.

QQQ ETF performance (backtest)

Note that most backtests in blogs with tech stocks and ETFs start after 2003 to avoid showing performance during the dot-com period. We never do that because we respect our readers and followers. In general, if you see a backtest starting after 2003, unless there is a good reason, like a lack of data, as is the case with some ETFs, it is probably an attempt to hide the dot-com performance.

The strategy is up 23% year-to-date after losing 2.2% in 2022.

SPY, QQQ, and IWM ETF performance (backtest)

Strategy details

Time frame: Daily (adjusted data)
Markets: SPY, QQQ, and IWM
Strategy type: Long-only, no bear market filters.
Backtest period: Inception to 06/09/2023
Commission per share: $0.01
Position size: 1/3 of available equity
Position entry and exit: Open of next bar
Stop-loss: None

The strategy is up 8.8% year-to-date after losing 1.4% in 2022.

Dow 30 performance (backtest)

Time frame: Daily (adjusted data)
Strategy type: Mean-reversion, long-only
Bear market filter: None (all signals taken even during corrections)
Universe: Dow 30 stocks (any delistings handled in backtest)
Maximum open positions: 10
Rank metric: Based on PSI5 algo
Commission per share: $0.005
Position size per stock: Available equity/10
Trade entry: Open of next bar (no look-ahead bias)
Stop-loss: None

The strategy is up 5.9% year-to-date after losing 1.5% in 2022.


Why are you selling the PSI5 algo if it is profitable?

Answer: It is rather complicated, and only professionals may understand why.

To start with, profitability depends on execution. If I give the same algo to 20 traders, they will probably generate 20 different equity curves.

More importantly, traders will take the algo and some of them may find ingenious ways of improving it. It may end up not being the same algo.

But above all, trading efficiently requires a large amount of capital. If there is no other source of income, then some traders elect to provide training or sell their strategies. The latter is indirectly what hedge funds do without disclosing their strategies. Strick risk management requires large amounts of capital. It takes a long time to compound to the levels needed to make a living, if ever, without an additional income stream.

Finally, mean-reversion is structural, and the edge is not going to disappear any time soon. The real edge is the timing of mean reversion, and the PSI5 algo has its logic for that. In addition, the same algo structure can be used for trend-following.

Click here for a list of strategies. 

Disclaimer: No part of the analysis in this blog constitutes a trade recommendation. The past performance of any trading system or methodology is not necessarily indicative of future results. Read the full disclaimer here.

Charting and backtesting program: Amibroker. Data provider: Norgate Data

If you found this article interesting, you may follow this blog via RSS or Email, or on Twitter.