Previous Week High Breakout: Tested on the S&P 500 and 7 Stocks

Why we ran the test again

In our first test, we ran a popular bias rule on 1,403 weeks of Nasdaq-100 data. The rule is simple: a close above the prior period's high is bullish, a close below the prior low is bearish, and traders then expect price to come back to that level. It's everywhere among followers of ICT and of the trader-educator TTrades, and TradingView user joshuuu published it as a public indicator in August 2023. On the Nasdaq-100, price reached the level way more often than usual, but the direction was no better than average.

Fair pushback: maybe the Nasdaq-100 is a special case. It's concentrated, volatile and dominated by tech. So we ran the exact same rule, unchanged, on:

• SPY, the S&P 500 ETF and the usual stand-in for ES futures;
• Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta and Tesla, the seven giants a lot of traders watch every day.

Same bars, same definitions, same measurements. Nothing was tuned market by market.

Result 1: the level part held everywhere

The flip side held too. After a bearish close, the next week traded below that week's low 63% to 71% of the time, against 41% to 45% for any week.

This part of the rule is solid, but it's also pretty mechanical. A week that closes above its prior high usually closes near its own high, so the next week starts within reach of it.

• S&P 500 (SPY)80.0%
• Apple77.6%
• Microsoft74.1%
• Nvidia77.6%
• Amazon74.4%
• Alphabet77.3%
• Meta72.8%
• Tesla77.0%

Exact values from table 2 in this article. Bar lengths share a zero baseline; no new calculation or simulated price path. Read the table and its stated sample and limits in this article.

Result 2: the direction part didn't hold anywhere

Here's the average return of the next week, by type of close:

In seven of the eight markets, the week after a "bullish" close was no better than an average week, and usually a bit worse. Tesla is the exception, and it's also the market where luck plays the biggest role: its average week gains 0.95%, with annualized volatility above 55%.

On the S&P 500, the Nasdaq-100 pattern showed up almost exactly. The week after a close below the prior low averaged +0.51%, more than double a normal week, and fewer than 1% of 10,000 random samples of the same size did as well. On single stocks, the same tendency showed up in four of the seven names, but in none of them was it strong enough to stand out from chance.

Why the S&P 500 acts like the Nasdaq-100, and stocks look noisier

An index averages hundreds of companies. When it breaks below last week's low, it's usually a broad shock: a data release, a rates scare, a geopolitical event. Broad shocks have tended to get partly reversed. A single stock breaking its low might be reacting to its own news, like a missed earnings estimate or a lawsuit, and that kind of move can keep going.

Single stocks are also way more volatile. A signal worth a few tenths of a percent on the index gets drowned out fast when a stock moves 5% to 10% in a normal week. The more volatile the market, the more weeks it takes before any edge, real or imagined, shows up.

A word on survivorship

We picked these seven stocks because they're the biggest companies today. That's a choice made with hindsight: in 2000, nobody knew Nvidia and Tesla would become giants. So their average weekly returns are much higher than a typical stock's. That doesn't distort the comparison between types of close, which is what we're testing. But it does mean the absolute returns in the table aren't what someone picking stocks in real time could have expected.

What this means if you trade ES or stocks

• Prior highs and lows are useful reference levels in any market. The tendency for a nearby extreme to get revisited was consistent across all eight.
• A close outside the range isn't a directional call. In no market did it reliably improve the next week.
• Think twice before selling breakdowns on the index. On both the S&P 500 and the Nasdaq-100, the week after a breakdown was, on average, the best of the three types.
• Expect more noise on single stocks. Even a real tendency would need a lot more data to show, and company news can override it.

That's why a daily bias built only on price structure is incomplete. Levels tell you where the market may react. They don't tell you which side wins.

How to run the test yourself

1. Export weekly or daily bars for the market you trade, adjusted for splits.
2. Label each bar bullish (close above the prior high), bearish (close below the prior low) or inside.
3. For the next bar, record whether it traded past the extreme and whether it closed higher.
4. Compare each group with all bars, then draw random samples of the same size to see how often luck does as well.

If you trade the daily version of the rule, measure the session you actually trade, for example 9:30 to 12:00 ET, not the full day.

Limits

• Weekly bars. People usually apply the rule to daily bars, and a daily test could give a different result.
• Survivor stocks. The seven names are today's biggest companies, picked with hindsight.
• Split weeks removed. We left out weeks around a stock split to avoid fake highs and lows.
• No costs. The test measures what happened next, not the profit of a strategy.

How we did it

• Weekly adjusted prices for SPY, AAPL, MSFT, NVDA, AMZN, GOOGL, META and TSLA from Alpha Vantage (TIME_SERIES_WEEKLY_ADJUSTED), through October 9, 2026. Highs and lows are scaled by the adjustment factor, and returns include dividends.
• Histories start in November 1999, except Alphabet (2004), Tesla (2010) and Meta (2012), from their first full week of trading.
• Random comparison: 10,000 draws per market, with a fixed seed.

Sources

• joshuuu. (2023, August 9). daily bias @ttrades x joshuuu [TradingView indicator]. TradingView. tradingview.com/script/ezS0ydMQ
• Alpha Vantage. TIME_SERIES_WEEKLY_ADJUSTED. Data retrieved October 2026.

FAQ

Does the previous-high bias rule work on the S&P 500? It predicts that the high gets revisited (80% of the time, against 62% normally), but not a higher close: the next week averaged +0.09%, against +0.19% for all weeks.

Does it work on individual stocks? The level part held on all seven mega-caps we tested. The direction part didn't reliably hold on any of them.

Why did bearish closes do better on the S&P 500? Broad index drops have often been partly reversed the next week. That's a historical tendency, not a rule.

Is ES different from NQ for this rule? On weekly data, the two indexes behaved almost the same.

Can I use this rule on Tesla or Nvidia? You can use the levels as references. But their weekly swings are so big that any directional edge from the rule is hard to separate from luck.

---

Research and education only. Not financial advice. Past results do not predict future returns. A backtest is not real performance.

How this article was made: drafted with AI assistance from public data and the sources listed above; facts and figures checked by Gabriel Décarie and Karel Panneton.