18 popular trading strategies from X, tested as posted

The claims

On 30 October 2026 we searched X for trading strategies with concrete rules and wide reach, mostly 150 likes or more. Most came from specific posts; the 7/21 EMA cross and the turn of the month are common setups we also saw shared. Each claim is quoted in the results table with the account that posted it.

Results

What the results say

The one survivor is small, and fading

Triple RSI buys an index ETF after a short pullback while it trades above its 200-day average. It won 79% of its trades before 2020 and 70% in the sealed holdout, but it is in the market only 10% to 25% of days, so on its own it earned about 3% a year against about 10% for holding. Its edge per decade: t 1.82 in the 1990s, 2.83 in the 2000s, 0.74 in 2010–2020. Similar dip-buying on single S&P 500 stocks fails. The RSI adds little: "three lower closes above the 200-day average" does about the same. This is short-term mean reversion in index prices, an effect researchers have described for decades.

Win rates mislead

Triple RSI won 79% of its trades; the trend rules won 23% to 36%. Neither number says whether a rule beats holding the index. Only the return after costs against the market does.

94 years of history is not a trade

Industry trend following cleared the bar with t 9.6 over 1927–2020. Its edge shrank decade by decade (t 9.1 before 1974, 4.4 after, 1.8 since 2000) and turned negative in the holdout. On sector ETFs that can actually be bought it never had one. Industry indexes cannot be traded, and old daily data has stale prices that make trends look smoother than they were.

Single stocks: every rule loses to the market

Pullbacks, crosses, breakouts and pin bars on S&P 500 stocks all trailed the market once costs were paid. With a euro account, the currency conversion on every trade makes it worse. Without it, none comes close to the bar either: the best t is +0.14.

The exact rules we tested

As posted. Where a post was vague, our reading is in brackets, written down before any result was seen.

1. Triple RSI: buy when RSI(5) is below 30, has fallen 3 days in a row, was below 60 three days ago, and the close is above the 200-day SMA. Sell when RSI(5) rises above 50.
2. Connors R3: close above the 200-day SMA, RSI(2) fell 3 days in a row from below 60 and is now below 10. Exit when RSI(2) rises above 70.
3. Industry trend following: 48 industries. Enter when the close is above the 20-day high or the upper Keltner band (EMA20 + 1.4 ATR). Trailing stop at the higher of the 40-day low and the lower Keltner band, never lowered. Each position sized to equal volatility, total exposure at most 200%, T-bills otherwise. [ATR from daily closes; held from the next day.]
4. RSI(2): buy when RSI(2) is below 10. Exit when the close is above the previous day's high. No trend filter, as posted.
5. Turn of the month: hold the index for the last 4 trading days of each month and the first 3 of the next; cash otherwise.
6. Moving-average crosses: enter on a bullish 5/20 SMA, 8/21 EMA or 20/50 SMA cross. Exit on a bearish 5/20 SMA cross or after 10 trading days.
7. Weekly RSI: buy when the weekly RSI(14) is 30 or lower. Sell 3 months (63 trading days) later.
8. Hidden Markov regimes: fit a 3-state model to daily S&P 500 returns; hold SPY except in the highest-volatility state, then T-bills; switch only after a new state has led for several days [3]. [Fitted only on data before each year, refitted yearly, read without smoothing so it never sees the future.]
9. S&P 500 or gold: at each month end, hold the S&P 500 next month if S&P 500 divided by gold is above its 20-month average, otherwise gold.
10. 8/21 EMA pullback: trend up while EMA8 is above EMA21. Buy the pullback into the 8 [the day's low touches EMA8 and the close stays above EMA21]. Exit when the 21 breaks [a close below EMA21]. Long only.
11. MACD cross: buy on a bullish MACD(12,26,9) cross, sell on the bearish cross.
12. 21 EMA pin bar: a stock in an uptrend [closes above EMA21 for 5 days, EMA21 rising] dips to the 21 EMA [low at or below it, close above], RSI(14) near 50 [40 to 60] and a bullish pin bar [lower wick at least twice the body, close in the top third]. [Stop at the pin bar's low, target twice the risk, 20-day time stop: the post gives no exit.]
13. Momentum base breakout: a stock up at least 50% in 3 months [63 trading days] forms a tight base [10-day range under 10%]. Buy the breakout above it, stop below the base, exit on a close below the 21 EMA.
14. RSI 60 on month, week and day: buy when daily RSI(14) crosses above 60 while the weekly and monthly RSI(14) are 60 or more. Exit when daily RSI(14) falls below 40.
15. Golden cross: buy when the 50-day SMA crosses above the 200-day SMA with the close above both. Sell when it crosses back below.
16. Pre-earnings run-up: buy 15 trading days before an earnings release. [Exit at the close the day before the report; the release date is the company's SEC earnings filing.]
17. 7/21 EMA: long while EMA7 is above EMA21 and the close is above EMA200; flat otherwise.
18. 10/20 SMA: long while the 10-day SMA is above the 20-day SMA, short while it is below.

Unless a rule says otherwise, signals are taken at the close and traded at the next day's open. Stops and targets fill inside the bar, at the open if price gaps through them.

Data, costs and the pass bar

Index ETFs

SPY, QQQ, IWM and DIA, daily, dividends included, 1993 to 2026. 0.02% per side. Judged against holding the same ETFs in equal parts.

Single stocks

The stocks in the S&P 500 at each date, daily, 2005 to 2020. Half the bid-ask spread per side, plus 0.15% currency conversion per side for a euro account (the second t leaves the conversion out). Judged against the market with monthly calendar-time alpha.

Industries

The 48 US industry portfolios of the Kenneth French data library, daily, 1927 to 2025. 0.05% per unit traded. Judged against the US market.

Crypto

Bitcoin, Ether and 9 other large coins as perpetual futures, daily, October 2020 to October 2026. 0.065% per side plus the funding actually paid. Judged against holding the same coins in equal parts.

Pass bar

Fixed before running: alpha t of at least 2.58 for the first ten tests and 2.50 for the next eight (the bar rises with the number of tests), 2.45 to 2.50 for the S&P 500 re-runs, and positive alpha in both halves of the development period. Development ended in 2020 for stocks, ETFs and industries; the crypto tests used 2020 to 2026.

Holdout

January 2021 to June 2025 stayed sealed and was opened only for rules that passed development. A rule passes it with alpha above zero.

What would change the verdict

• For a rule that failed: a rule we misread. If you posted one of these, send us the exact rules and we will run them as written and publish the result, pass or fail.
• For Triple RSI and R3: positive alpha after costs in live trading. Our rule is six months of forward tracking before calling any signal tradable, and Triple RSI's alpha since mid-2025 is below zero.
• A different market or period is a new test with its own holdout, not a rescue for an old one.

Also read, but not testable

Selling puts on red days (needs option prices), paid indicator set-ups, one-minute "liquidity sweep" day trading and AI-bot posts without rules. One post said altcoins at the lower Bollinger band "pumped 2-3x": an equal-weight index of 10 large altcoins touched its weekly lower band 5 times since 2021, and the next 8 weeks returned +4.7%, −32.8%, +19.9%, +2.5% and −7.0%.

Stock and ETF rules like these can be audited in the product: export each signal's ticker and date to a CSV and choose a holding period. The crypto rows are Autrady research, not yet in the self-serve product. For the scripts, email [email protected].