RSI(2) Strategy on QQQ: Does Connors' Rule Work on the Nasdaq-100?

00The verdict in three lines

• Does the RSI(2) strategy work on QQQ? It made money, +0.492% a trade after costs on 213 trades, with 69.0% winners and a profit in every period, but it passed only three of our six gates.
• Weaker than on SPY. On SPY the same rule made +0.616% a trade and beat every random run; on QQQ it beat 93.2% of them, short of our 95%, and the first period had only 36 trades.
• Not an independent test. 93 of the 213 QQQ trades started on the same day as a SPY trade. The SPY result stands; new data is still the real check.

01Why we ran it on QQQ

Our RSI(2) test on SPY passed all six gates. A result on one market can be luck, so we applied the same rules to QQQ, which tracks the Nasdaq-100, without changing anything: same thresholds, same exit, same costs, same periods, gates and control.

• Buy at the close when QQQ closes above its 200-day average and its 2-day RSI is below 10 (RQ2-10) or below 5 (RQ2-5).
• Sell at the first close above the 5-day average. No stop, no target, no time limit.

The rules are explained step by step in the SPY article. Two things were known before the run and written into the specification: QQQ fell about 80% from 2000 to 2002, which keeps a rule that needs an uptrend out of the market for years, and QQQ and SPY often dip on the same days, so a pass here would strengthen the SPY result only partly.

02How we tested it

QQQ daily prices and dividends from its first day, 10 March 1999, to 20 October 2026, the last day of the SPY test (Yahoo Finance; dividends start in 2003; the 2:1 split of March 2000 is adjusted; our total return with dividends matches Yahoo's adjusted close). Signals start in March 2000, once the 200-day average exists. Costs: 0.04% per round trip, with 0.10% shown. Periods: 2000 to 2005, 2006 to 2015, and a holdout from 2016 that we had not used on QQQ, but it is only partly fresh: we tested NQ futures, the same index, from 2020, and we knew the SPY result. First-period bar t ≥ 2.5. Control: random buys on days when QQQ is above its 200-day average and below its 5-day average, with the same exit and the same number of trades, 500 times. Engine: the SPY engine unchanged, with 15 checks of the QQQ data and of the engine against an independent re-implementation (rqq_selftest_result.txt).

03Results: does the RSI(2) strategy work on QQQ?

* Partly used before: never on QQQ, but NQ futures (the same index) were tested from 2020, and the SPY result was known.

Both versions made money in every period. The problem is the first one: from 2000 to 2005 RQ2-10 traded only 36 times, because QQQ spent most of 2000 to 2003 below its 200-day average, and with so few trades its t-statistic (2.32) stayed under our bar. RQ2-5 was very strong in that period (t 3.23) but had only 17 trades.

04RSI(2), or any dip?

Random buys on any dip in an uptrend made +0.346% a trade in the median run. RQ2-10 made more, but it was better than 93.2% of the random runs, not the 95% we require; the stricter RQ2-5 passed this gate. On SPY both versions beat every random run.

05QQQ next to SPY

Same direction, weaker numbers. 93 of the 213 QQQ trades began on the same day as a SPY trade, so the two tests overlap a lot; QQQ adds some evidence, not a second independent proof.

06RSI(2) on QQQ against buy and hold

Growth of $1 from March 2000, log scale. The strategy earns 3-month T-bill interest while out and pays the cost on every trade; buy and hold reinvests dividends.

Starting at the top of the dot-com bubble, holding QQQ fell -83.0% before it recovered; the strategy's largest fall was -15.3%. Over the whole period holding still grew more (×7.68 against ×4.34), as on SPY: a strategy in the market 11.8% of the time gives up most of the long rise.

07How solid is it?

Costs and next-open fills change little, and removing the best trades leaves a profit. The worst open loss (open positions included) came in April 2000, in the first weeks of the dot-com crash: 12.6% down at a close before the exit.

08By year

RQ2-10 lost in 6 of its 25 years with trades (2000, 2008, 2011, 2016, 2018 and 2022) and had no trade in 2001 and 2002, when QQQ stayed below its 200-day average.

09Verdict by variant

No proven edge: positive, but not convincing by our bars. RQ2-5 failed only the trade count, as it did on SPY.

TWhat this shows, and what to take into your own tests

What the results show

• The same RSI(2) rule that passed every gate on SPY made +0.492% a trade after costs on QQQ, on 213 trades from 2000 to 2026, with 69.0% winners, and was positive in all three periods.
• It was not convincing by our rules: only 36 trades from 2000 to 2005, when QQQ spent years below its 200-day average after the dot-com crash, and better than random dips in an uptrend in 93.2% of runs, short of the 95% we require.
• The stricter threshold (RSI(2) below 5) made +0.675% a trade and beat the random dips at the 97.4th percentile, but with 17, 46, 48 trades it fell short of 50 in each period.

What remains unknown

• Whether the rule keeps working on either ETF: the data after 20 October 2026 has not been looked at for SPY or QQQ.
• How much of the QQQ result is the same signal as on SPY: the two share many of their dips, so this is not an independent test.
• Other Nasdaq instruments, thresholds or exits; we changed nothing from the SPY test.

Why it was buried

Three gates failed. The first period had too few trades (36) and was not clear enough (t 2.32), and the rule beat random dips in an uptrend only at the 93.2th percentile. It made money everywhere it traded, so the label is no proven edge: the direction is the same as on SPY, the evidence is weaker.

What you can apply to your own tests

• Test a rule that worked on one market on a second one with the rules unchanged. Here the second market gave the same direction, less clearly.
• Check how independent the second market is: about half of the QQQ trades were on the same days as the SPY trades.
• Count the trades per period before trusting a result: a crash can take a rule out of the market for years.

A note on the numbers: drawdown figures are measured on closed trades unless a table says it includes open losses. Open losses inside a trade are not counted in a closed-trade drawdown, so real account risk is higher, especially for systems with no stop. Backtests are not live trading.

FAQCommon questions

Does the RSI 2 strategy work on QQQ?

Not convincingly in our test: 213 trades from 2000 to 2026 at +0.49% each after costs, with 69% winners, against +0.62% on SPY with the same rules. It was positive in every period but passed three of our six gates.

Why did the RSI 2 strategy do worse on QQQ than on SPY?

QQQ fell about 80% after 2000 and stayed below its 200-day average for years, so the rule traded only 36 times from 2000 to 2005; and its edge over random buys on dips was smaller (the 93.2th percentile, against the 100th on SPY).

Is QQQ a good market for mean reversion?

In our test, random buys on any dip in an uptrend made +0.35% a trade on QQQ, so dips did tend to recover; the RSI(2) extreme added less on top of that than on SPY.

How was the RSI 2 strategy tested on QQQ?

With the SPY rules unchanged, on QQQ daily closes with dividends from 2000 to 2026, 0.04% of costs per round trip, a control of random buys on dips in an uptrend, and rules written and dated before the data was run, costs included on every trade, results split into development, validation and holdout periods, a random-entry control as the benchmark, and every variant reported. Read the protocol.

10Check our arithmetic

Code provenance: our own code; the engine is the one of the SPY test, unchanged.

11Data notes

Source. Yahoo Finance daily prices of QQQ and its dividends; FRED for the 3-month T-bill rate. Indicators use the closes adjusted for the 2000 split; returns include dividends. Fills at the close (and, as a check, at the next open); a fixed 0.04% cost. Not independent of SPY, and the holdout is only partly fresh.

12Corrections

7 Nov 2026, wording: the scoreline and the note under the period table now say that the holdout was partly used before. No result changed.

+More tests and guides

• Rsi 2 strategy
• RSI 2 strategy backtest
• Internal bar strength strategy
• Noise area breakout nq
• Free prop firm challenge calculator
• What all the backtests found so far
• The backtesting protocol
• How to backtest a trading strategy without fooling yourself
• Backtesting glossary