Home › Explore Investing Strategies › Technical analysis
Reading indicators: moving averages, MACD, RSI and volume
Formulas for moving averages, MACD, RSI and volume. SPY and QQQ since 2000: returns after RSI 30 and 70 and MACD crosses, and how closely indicators track.
In one line
After days when SPY's RSI was below 30, the next 60 trading days averaged 7.2%, above the 2.2% for all days, but in 2008 five of six first-day signals lost money. Returns after MACD crosses were almost the same as the overall average, and RSI correlated 0.82 with the distance from the 20-day line.
Article
You can find everywhere that an RSI below 30 means oversold and that the MACD line crossing above its signal line is a buy signal. But the same RSI gives different values depending on how it is calculated, and the same signal worked in some years and failed in others. Knowing what an indicator summarizes, and when it was reliable, keeps you from simply following the signal on the screen.
An indicator is what you get when you run price and volume through a formula. Once you know the formula, you can see what the indicator summarizes and what it misses. This article writes out the exact definitions of four indicators, then calculates what followed the standard RSI and MACD signals on SPY and QQQ since 2000 and how much the indicators overlap. Figures are as of September 2026 and were calculated by the operator with this site's indicator engine. Trading results of moving-average rules are in Moving-average trading.
Formulas
- Moving averages: a simple moving average (SMA) is the average of the last n closes. An exponential moving average (EMA) is today's value = α × today's close + (1 − α) × yesterday's value, with α = 2 ÷ (n + 1). The first value starts from the simple average of the first n days (the TradingView method). Some programs start from the first close instead, so early values differ.
- MACD: created by Gerald Appel in the late 1970s. MACD line = 12-day EMA − 26-day EMA, signal line = 9-day EMA of the MACD line, histogram = MACD line − signal line. The histogram is known to have been added by Thomas Aspray in 1986.
- RSI: introduced by J. Welles Wilder in his 1978 book New Concepts in Technical Trading Systems. Daily gains and daily losses are averaged separately. The first averages are 14-day simple averages, and after that each is updated as (previous average × 13 + today's value) ÷ 14. RS = average gain ÷ average loss, and RSI = 100 − 100 ÷ (1 + RS). This smoothing is exponential with α = 1/14, so it remembers the past with the same weights as a 27-day EMA.
- Volume ratio: the day's volume divided by the average of the prior 50 trading days. This article reads volume through this ratio.
Same name, different values
Three versions of RSI(14) are in common use, depending on how the averages are taken: Wilder's method, Cutler's method with 14-day simple averages, and a version with a standard EMA (α = 2/15). On SPY's close of September 30, 2026 they gave 47.8, 54.2 and 43.7. Since 2000, SPY closed with RSI below 30 on 114 days by Wilder's method, 378 days by Cutler's and 478 days by the EMA version. Wilder's method remembers the past longer, so it reaches extremes less often. The same "RSI below 30" condition shows up more than three times as often depending on the formula. This site follows the TradingView standard (EMA seeded with an SMA, RSI with Wilder smoothing), as the methodology page states.
After RSI 30 and 70 (as of September 2026)
RSI below 30 was rare. Since 2000, SPY was below 30 on 114 days (1.7% of trading days), or 60 separate stretches, and QQQ on 108 days (55 stretches). On 99 of SPY's 114 days the close was below the 200-day line, and the median volume ratio on those days was 1.87. RSI above 70 came on 422 days for SPY (6.3%) and 620 for QQQ (9.2%), and once above 70 it stayed for 3.2 days on average (4.3 for QQQ).
| Case | Days | Average after 5 trading days | Average after 20 trading days (share up) | Average after 60 trading days (share up) |
|---|---|---|---|---|
| SPY, RSI below 30 | 114 | 1.74% | 4.16% (74.6%) | 7.23% (76.3%) |
| SPY, RSI above 70 | 422 | 0.00% | 0.61% (69.0%) | 1.50% (74.6%) |
| SPY, All days | 6,726 | 0.19% | 0.75% (64.4%) | 2.24% (69.4%) |
| QQQ, RSI below 30 | 108 | 2.24% | 2.92% (57.4%) | 7.79% (63.0%) |
| QQQ, RSI above 70 | 620 | 0.06% | 0.62% (60.8%) | 1.49% (62.4%) |
| QQQ, All days | 6,726 | 0.23% | 0.89% (62.4%) | 2.58% (67.5%) |
Days below 30 were followed by higher averages than all days, and days above 70 by about the same as all days. Above 70 did not mean a decline was coming. Results after RSI below 30 were widely spread, though: the standard deviation of SPY's 60-day return was 11.4% after those days, against 7.5% for all days.
Same signal, different years: 2008 and 2022
Taking the first day RSI fell below 30 as the signal, 12 of SPY's 60 signals ended in a loss after 60 trading days. The losses clustered in two periods: five of six signals lost in 2008 (average −6.6%), and four of ten in 2000 to 2002. From 2009 on, only 3 of 37 lost (two in October 2018 and one on February 25, 2020), and the average was 11.6%. QQQ did worse. Its four signals between September 9 and October 2, 2008 were followed by −21% to −33% over 60 trading days. In 2022, SPY's three signals all gained (1.9%, 3.8% and 7.3%), while QQQ's three all lost a little (−1.1% to −1.9%).
The custom rules in the backtest tool include an RSI recovery rule: buy when RSI comes back above 30, sell when it reaches 70. We ran it with its default settings from January 2000 to September 2026 (whole account in or out, 0.15% fee per trade, fills at the next trading day's open). On SPY, 77% of its 22 round trips ended in a gain, yet it returned 4.1% a year, half of buy and hold (8.2%). On QQQ 83% of trades gained, and it returned 0.6% a year (holding 8.6%). Because the rule does not sell until RSI reaches 70, it held through long bear markets. On SPY it bought in July 2007 and sold in July 2009 at a 31.3% loss; on QQQ it bought in April 2000 and sold in November 2001 at a 55.2% loss.
MACD crosses
Crosses of the MACD line and the signal line were frequent. Since 2000, SPY had 288 upward and 289 downward crosses, 21.6 a year on average, and the median gap between crosses was 10 trading days. Of the 288 upward crosses, 155 were followed by a downward cross within 10 trading days. Returns after crosses had almost nothing to do with their direction. SPY's 20-day average was 0.72% after upward crosses, 0.44% after downward crosses and 0.75% for all days, and its 60-day averages were 2.02%, 2.05% and 2.24%. QQQ's 20-day averages, 1.08% and 1.07% (all days 0.89%), showed no difference either. Zero-line crosses were less frequent, 7.7 a year (7.9 for QQQ), with the same result. Splitting SPY's upward crosses by whether volume was above average left the 60-day averages equal at 1.99% and 2.03%.
How much the indicators overlap
Indicators built from the same closes mostly moved together. On SPY's daily values since 2000, RSI correlated 0.86 with the stochastic and 0.82 with the close's distance from the 20-day line, and the MACD line (relative to price) correlated 0.96 with the distance from the 50-day line. Moving averages (position), MACD (strength) and RSI (speed) have different names but largely show the same information again. Volume was different. RSI and the volume ratio correlated at −0.49, because volume piled up on sharp drops when RSI was low. Volume does add information the price indicators do not, but it did not confirm the upside (Technical analysis basics).
In 2014, Neely, Rapach, Tu and Zhou reported that moving-average, momentum and volume indicators forecast the monthly US equity risk premium about as well as macroeconomic variables, and were especially good at catching declines near business-cycle peaks. Unlike the daily signals in this article, their results look at the whole market month by month.
Related concepts and search terms
- Overbought and oversold: names for RSI above 70 and below 30. They are Wilder's reference levels, but as the numbers above show, above 70 did not mean a decline was coming.
- Divergence: a mismatch where price makes a new high while the indicator falls. Results depend heavily on how the highs and lows are picked, so this article does not calculate it.
- Stochastic oscillator: the position of the close within the last n days' high-low range, on a scale of 0 to 100. In the calculation above it had the highest correlation with RSI.
- RSI(2): the two-day RSI used by Larry Connors and Cesar Alvarez in a 2008 book. It is known for a short-term rule that buys when RSI(2) falls below 10 while price is above the 200-day line.
- ADX and Parabolic SAR: a trend-strength indicator and a trailing-exit indicator that Wilder introduced in the same 1978 book as RSI.
Search terms: RSI Wilder smoothing, Cutler RSI, MACD 12 26 9, MACD histogram, overbought oversold, divergence, RSI backtest, MACD crossover backtest.
Limits
- These are results for two US index ETFs. Individual stocks may fail to recover from RSI below 30 more often than an index.
- Counting every day makes the samples overlap. Of SPY's 114 days with RSI below 30, 45 fell in 2000 to 2002, 2008 and 2020.
- Wilder's RSI depends on the start date in its early values. These calculations run on data from the 1990s, so values after 2000 are barely affected.
- Signals that need a person to pick highs and lows, such as divergence, were not calculated.
- Averages after a signal differ from the results of a trading rule. Rule results change a lot with exit conditions and costs, as the RSI recovery rule above shows.
Operator's assessment
Indicators summarize the current state precisely, but how much they said about what came next varied by period. After SPY's RSI fell below 30, the next 60 trading days averaged 7.2%, above the 2.2% for all days, but the gap came from the short declines after 2009, and in 2008 five of six first-day signals lost money. Run as the backtest tool's RSI recovery rule, the same idea won 77% of its trades yet returned 4.1% a year, half of buy and hold. MACD crosses came 21.6 times a year on SPY, and returns after them matched the overall average. The indicators were also highly correlated: 0.82 between RSI and the 20-day distance, and 0.96 between the MACD line and the 50-day distance. The confluence score is likewise a weighted average of nine such technical indicators. When reading it, these numbers favor not treating agreement among nine indicators as nine independent confirmations, and checking the regime call first.
What this article does not cover
This article does not provide trading signals based on indicators. The figures are averages and backtests calculated from past daily prices of the US ETFs SPY and QQQ. Other tools may calculate differently, so we do not guarantee matching values. Everything here is reference information based on past, public data and is not investment advice.
Articles and questions on the same topic
Moving averages and trend following
- Strategy articles: Turtle trading: the 20-day and 55-day breakout rules, tested on ETFs, Dual momentum (GEM): switching between stocks and bonds on 12-month returns
- Backtest questions: Does selling below the 200-day moving average reduce drawdowns in monthly investing?, Holding TQQQ only while QQQ is above its 200-day line: did it beat buy and hold?, QLD (2x Nasdaq-100) with a 200-day line rule: how different were the drawdown and return from buy and hold?
- Investor profiles: Trend Breakout Strategy (Jesse Livermore), New-High Growth Momentum Strategy (William O'Neil)
Related
Moving-average trading · Technical analysis basics · Support, resistance and breakouts · Golden cross backtest · Glossary · A real stock analyzed
Basics and trading strategies
← Plan before entry: reason, exit condition, entry price and stop · Contents · Leveraged ETFs at a glance: monthly investing and five rules →
See these metrics on a real stock →
About & methodology · Explore Investing Strategies · Glossary · Questions and answers · Rule status · Sample analysis · Pricing · Terms · Investment disclosure · Privacy policy · Open the analyzer · 한국어 · 日本語 · 繁體中文 · 简体中文