Updated 20 November 2026 · 14 tests, newest first
This page lists every trading strategy we have tested, in plain English. Read each line as the claim, then our result, then how it did against simply holding the index, then our verdict. The lesson so far: none has cleanly beaten simply holding the index once trading costs and idle cash count.
Holding the index means buying one fund that tracks the market, such as SPY or QQQ, and doing nothing. Idle cash is money a strategy keeps out of the market while it waits for a signal.
How each test works is on our method page.
14 · Buying SPY after a 30% fall from its all-time high and holding 12 months; also tested at 25% and 20%
The claim: Buying the S&P 500 after a 30% fall loses money only 3% of the time over the next 12 months, with an average gain of 22%.
Our result: The 3% counts 1,008 days from just six crashes, and the first buy lost money in two of them. As a rule on SPY since 1993 it won 5 of 6 trades, yet $10,000 grew to about $48,000; left in SPY, it grew to about $291,000.
Against the index: Buying only after a 30%, 25% or 20% fall earned 4.84%, 5.74% and 6.76% a year; holding SPY earned 10.75%.
Verdict: All three thresholds fail.
Should you use it? No. If you hold an index fund, this test gives no reason to sell and wait for the next crash. The borrowed top-up version fails too.
Source: A Spanish-language YouTube video. Full test: What Would Actually Happen If You Bought Every 30% Crash?.
13 · Four dip-buying rules as top-ups on QQQ: Williams %R(2), KAMA(5) mean reversion, Triple RSI, Connors RSI(2)
The claim: Buying the dip with borrowed money beats borrowing steadily.
Our result: All four out-earned steady leverage: the top-ups ranged from about $166,000 (Triple RSI) to $471,000 (KAMA). But the pass rule, written before the test ran, also required no deeper fall. KAMA and Connors RSI(2) cleared it; Williams %R missed by 1.1 points and Triple RSI by 0.5.
Against the index: Steady leverage with KAMA borrowing made about $234,000, and two of the four fell deeper than it did.
Verdict: Two passed, two fell deeper.
Should you use it? Only as an add-on. Only as a top-up on an index you already hold. Williams %R needs a buy at the closing price, not the next morning.
Source: Quantified Strategies; Larry Connors and Cesar Alvarez. Full test: Four Dip-Buying Rules Out-Earned Leverage. Only Two Passed..
12 · Paper trading on delayed prices: our own IBKR paper account on Micro Nasdaq futures
The claim: Test a strategy on a paper account before you risk money.
Our result: For four weeks our paper account traded on futures prices ten minutes old, and nothing said so. On 9 of 618 decision bars the settled bar would have changed the signal.
Against the index: Not applicable: this is research about data, not a strategy.
Verdict: Check the feed before trusting paper.
Should you use it? Not a strategy (data research). The post lists three checks before you trust a paper result.
Source: Our paper lane (Saty Phase Oscillator, Issue 06). Full test: Your Paper Trades May Be Running on Delayed Prices.
11 · 8/21 EMA crossover: buy the first dip after the 8-day exponential moving average crosses above the 21-day
The claim: The strategy that "will help you retire".
Our result: Each trade earns a little, and it roughly halves the worst fall. But it is in the market only a little over half the days. $10,000 on SPY from 1993 became $49,923.
Against the index: Holding SPY turned the same $10,000 into $316,014. On QQQ: $60,501 against $167,199.
Verdict: Trails the index.
Should you use it? No.
Source: Justin Banks on X. Full test: The 8/21 Pullback, a Sixth of the Money.
10 · Triple RSI: buy the S&P 500 fund after a three-day slide in a short-term momentum gauge
The claim: Wins about 90% of its trades.
Our result: The 90% is real: 82 winners in 91 trades. But the money sits in cash 95% of the time. $10,000 became about $70,500 from 1993 to 2026.
Against the index: Simply holding the index fund turned the same $10,000 into about $298,000.
Verdict: Claim true; trails the index alone.
Should you use it? Only as an add-on. As a top-up on an index you already hold, with borrowed money; it does nothing in a crash.
Source: Quantified Strategies. Full test: Nine Wins in Ten, a Quarter of the Money.
09 · Free Dukascopy one-minute price data, used in many free backtests
The claim: Free one-minute index data, downloadable for years back with one command.
Our result: In 2013 about one minute in ten is a placeholder: a flat candle written when no price arrived. The files also show prices on 84 days the New York Stock Exchange was closed.
Against the index: Not applicable: this is research about data, not a strategy.
Verdict: Clean free data before backtesting.
Should you use it? Not a strategy (data research). The post includes a free 37-line checker.
Source: Dukascopy free historical data. Full test: One in Ten Minutes of Free 2013 Nasdaq-100 CFD Data Is a Placeholder.
08 · 5-minute opening range breakout with a gap filter (and fixes tried on UT Bot, Nadaraya-Watson and TTM Squeeze)
The claim: A popular day-trading rule: buy or sell the direction of the first five minutes.
Our result: We tried twelve fixes on four failed strategies. Only a gap filter brought one back: the average trade more than doubled. The evidence is borderline and fades when orders fill late.
Against the index: Alone, it turned $10,000 into a fraction of what QQQ made. Kept on top of a QQQ holding, it beat QQQ by several points a year, before real-world delays.
Verdict: Borderline; helps only beside QQQ.
Should you use it? Only as an add-on. And only if its edge survives your real fill speed.
Source: Carlo Zarattini and Andrew Aziz, Can Day Trading Really Be Profitable? (2023). Full test: Not a Winner. Maybe a Sidekick..
07 · Three earlier strategies re-priced at the speed you can really trade
The claim: Every backtest fills your order the instant a price bar closes.
Our result: A 47-second delay, the speed of our own automated orders, cost 9 to 16% of profit. At two minutes late, the breakout's edge stopped looking real.
Against the index: Not an index test: this issue measures what waiting costs.
Verdict: Delay is a cost; test it.
Should you use it? Not a strategy (data research).
Source: Strategies from Issues 01 and 06. Full test: The 47-Second Tax.
06 · Saty Phase Oscillator, a Nasdaq day-trading signal (the strategy in our own paper account)
The claim: Over the full history: less money than the index, much less risk.
Our result: Strong on the years it was tuned on: 12.06% a year. On the years it never saw: 4.27% a year, on too few trades to call.
Against the index: The Nasdaq 100 made 20.52% a year in the tuned years and 17.11% in the unseen years.
Verdict: Unproven.
Should you use it? No. Not until it beats the index on years it never saw.
Source: Saty Phase Oscillator, as run in the Lab's paper account. Full test: We Broke Our Own Strategy.
05 · Connors RSI(2): buy the S&P 500 after a sharp two- or three-day drop, sell on the bounce
The claim: One of the most reposted dip-buying rules, and it wins 73% of its trades.
Our result: The edge is real, but the money is in the market only about a tenth of the time. An account running it made 4.47% a year.
Against the index: Holding the index made 12.71% a year. One dollar became $4.95 holding the index and $1.79 with the strategy.
Verdict: Real edge, loses to the index.
Should you use it? Only as an add-on. Kept on top of an index holding it added return, with borrowed money and more risk.
Source: Larry Connors and Cesar Alvarez, Short Term Trading Strategies That Work (2008). Full test: The Winner That Doesn't Win.
04 · Industry trend following, tested over 100 years
The claim: 18.2% a year for a hundred years.
Our result: It replicates: we got 18.27%. But 63% of the growth came before 1980. The version you can actually buy, with sector funds, made 7.97% a year, and 5.85% once retail borrowing costs and trading costs count.
Against the index: Over those same fund years, 2006 to 2026, the market made 11.37% a year.
Verdict: Replicates; loses to the index today.
Should you use it? No.
Source: Carlo Zarattini and Gary Antonacci, A Century of Profitable Industry Trends. Full test: The Century That Ended in 1990.
03 · Weekly SPX put credit spread: an options trade that collects a small payment each week unless the S&P 500 falls far
The claim: Wins 95.4% of its weeks, with a worst fall of $4,605 on one contract since May 2020.
Our result: The win rate is real. Run the same rule from 2013 and the worst fall grows five to ten times, to between $22,837 and $46,136. Three of the four worst weeks came just before the backtest starts.
Against the index: The post does not compare it with holding the index. Its point is the hidden risk.
Verdict: Win rate real, risk understated.
Should you use it? Maybe. Only with far more capital than the published $4,605 worst case suggests.
Source: The Milk Man (@MrMilkTrading). Full test: The Missing Ten Weeks.
02 · Day-trading strategies opened up by the end of the $25,000 pattern-day-trader rule
The claim: The $25,000 minimum for day trading is gone, so small accounts can now day-trade.
Our result: The rule opened a door, not an edge. One trailing-stop strategy took 54,545 trades in thirteen years and lost about $940,000 per contract, mostly to costs. A published opening-range test fell from $138,639 to $4,860 once realistic slippage, the gap between the price you see and the price you get, was charged.
Against the index: The post does not compare these strategies with holding the index. They fail on costs before that question comes up.
Verdict: Fails under realistic slippage.
Should you use it? No.
Source: FINRA rule change. Full test: The $25,000 Door.
01 · Five popular strategies: Connors RSI(2), the 5-minute opening range breakout, UT Bot, TTM Squeeze and the Nadaraya-Watson envelope
The claim: Five popular strategies, each shared online with its full rules.
Our result: Four failed once trading costs and a fair comparison were counted. Only Connors RSI(2) survived, trade by trade.
Against the index: The survivor, Connors RSI(2), made $115,925 per contract. Simply holding the index made $291,484 over the same years, more than double.
Verdict: Four fail, one survives per trade.
Should you use it? No. Issue 05 replaced the Connors verdict: read that first.
Source: Their published rules (one author per strategy, named in the post). Full test: Five Strategies, One Survivor.
If you wrote one of these rules and think we got it wrong, write to [email protected]. We print replies in full.
Your turn
Send us the strategy you are about to risk money on. Write to [email protected].
Full method: How We Test a Trading Strategy.
Past performance is no guarantee of future results. We check anyway.
This is research on historical data, not investment advice. Full disclaimer.

