Smart Money Flow Index and Smart Money Index: Formula, Origins, and What the Data Shows
The difference between the Smart Money Index and the Smart Money Flow Index, the SMI formula with a worked example, what its promoters claim, and how overnight and daytime returns split on the S&P 500 ETF since 2000.

On this page
The Smart Money Index (SMI) is a sentiment indicator built from the time of day at which the Dow Jones Industrial Average moves. It subtracts the move in the first half-hour of trading, assumed to come from emotional investors, and adds the move in the last hour, assumed to come from professionals, then keeps a running total. The Smart Money Flow Index (SMFI) is a related indicator with its own formula. This page covers what each one is, the formula with a worked example, the claims made about it, and what the split between overnight and daytime returns shows on our data.
What the Smart Money Index counts in a session

Smart Money Index and Smart Money Flow Index
The two names are easy to confuse, but they are not the same indicator, and what is known about each comes from thin sources. Wikipedia says the SMI was invented and popularized by money manager Don Hays, and that the SMFI is based on it but uses a slightly different, proprietary formula and is featured on Bloomberg Professional, so the two can give different signals. Wikipedia gives no source for any of this: its only reference is the archived site of the Hertler Market Signal newsletter (Wayback Machine capture of October 15, 2011), which mentions neither Hays, the SMFI nor Bloomberg. That newsletter page says the index was described by Lynn Elgert in the February 22, 1988 issue of Barron's, and Wikipedia repeats that sentence word for word. So the authorship rests on one uncited claim for Hays and one newsletter's account for Elgert. The research firm SentimenTrader adds to the confusion: its 2020 review treats the two names as one indicator, “also known as the Smart Money Flow Index”, popularized by “Don Hayes” in the 1990s, and computes its own version on the S&P 500, from 9:30 to 10 a.m. and from 3 to 4 p.m. Because the SMFI formula is not published, everything below is about the SMI.
How the Smart Money Index is calculated
The Hertler page gives the idea behind it: emotional, news-driven investors tend to trade in the first half-hour, while rational, professional investors tend to trade in the final hour after evaluating the day's action. The index therefore counts the first move against the market and the last move for it.
The worked example: two Dow moves and the SMI's change

The level of the SMI means nothing on its own, since it depends on the start date. What matters is its path compared with the Dow's. Neither source we found gives a precise trading rule; SentimenTrader reads the index through a divergence with the stock indexes, and the archived Hertler page only compares its path with them after 2000.
What is claimed for it, and what a review found
The archived Hertler page credits the index with signaling in advance the 1982 market bottom and the 1987 top, then the July 1998 top, the October 1998 bottom and the January 2000 Dow top, and says it was featured in an article by Conrad Aenlle in The New York Times on February 13, 2005. The same page also says that since the 2000 market top the Smart Money Index has failed to keep pace with the major stock indexes, and that the newsletter turned to its two components separately, which it calls the early and late market components. Those are the newsletter's own statements, and we could not check them on intraday data.
The one review we found that looks at failures as well as successes is SentimenTrader's, written by Troy Bombardia in January 2020. It notes that a divergence is hard to test because “no 2 traders staring at a chart will define a ‘divergence’ the exact same way”, then goes through major tops on its S&P 500 version. The timeline puts both accounts side by side.
Calls credited to the Smart Money Index, and the misses
SentimenTrader: a bullish divergence from January to October 1973 failed; stocks crashed from October 1973 to October 1974. SentimenTrader
Hertler: credited with signaling the 1982 market bottom in advance. Hertler Market Signal (archived)
Hertler: credited with signaling the 1987 market top in advance. Hertler Market Signal (archived)
Hertler: the index is described by Lynn Elgert in Barron's. Hertler Market Signal (archived)
Hertler: advance warning of the July 1998 market top. Hertler Market Signal (archived)
Hertler: advance warning of the October 1998 market bottom. Hertler Market Signal (archived)
SentimenTrader: the index peaks in August 1999 and has collapsed by March 2000, ahead of the 2000 top. SentimenTrader
Hertler: advance warning of the January 2000 Dow top; afterwards, the index fails to keep pace with the stock indexes. Hertler Market Signal (archived)
SentimenTrader: the index peaks more than 3.5 years before the October 2007 top. SentimenTrader
Hertler: the index is featured in a New York Times article by Conrad Aenlle. Hertler Market Signal (archived)
SentimenTrader: the index falls from the first to the third quarter while stocks rise, before the fourth-quarter crash. SentimenTrader
SentimenTrader publishes its review: a mixed record, not clearly better than a coin toss. SentimenTrader
SentimenTrader's conclusion is that the record “is mixed, and it isn't clear that its track record is much different from a 50/50 coin toss”. It adds a reason to doubt the premise: the last hour, supposedly the smart money's, is when exchange-traded funds rebalance and when volume spikes, which, in our reading, makes it crowded with every kind of trader. Neither the newsletter nor the review counts every signal by a fixed rule, so neither is a test; together they show that the famous calls come with failures that the promotional account leaves out.
What daily data can show: overnight against session
We do not have intraday data for this test, so we cannot compute the SMI. What daily bars do allow is a related split: the move from the previous close to the open, which happens overnight, against the move from the open to the close, during the session. This is not the SMI; it only tests whether the time of day at which returns arrive is stable enough to build on. The chart covers the last 250 trading days of the S&P 500 ETF, each part compounded on its own.
Overnight and session returns on the S&P 500 ETF

- Price, daily candles (hollow up, filled down)
- Overnight only: prior close to open, compounded, % (solid line, second color)
- Session only: open to close, compounded, % (solid line)
- Close to close, % (dashed line)
From October 30, 2025 to October 28, 2026, the overnight part alone returned +12.75% and the session part +1.85%, for a total of +14.83%. Sessions closed above their open on 126 of 249 days and nights on 141. Over the full history:
| S&P 500 ETF, from | Overnight only | Session only | Close to close |
|---|---|---|---|
| January 3, 2000 | +536.7% | +32.1% | +741.2% |
| January 4, 2010 | +315.4% | +124.1% | +830.9% |
| October 31, 2016 | +182.7% | +50.5% | +325.5% |
Since 2000, 53.0% of sessions and 55.2% of nights were up, over 6,746 days. On the Nasdaq 100 ETF from 2000 the split is starker: overnight +1,739.8%, session −48.7%.
The pattern is not a rule year by year. Over the 26 full years from 2000 to 2025, the session beat the overnight move in 9: 2001, 2003, 2009, 2012, 2013, 2016, 2022, 2023 and 2025. The overnight part lost money in 2001, 2002, 2008, 2016 and 2022.
Overnight and session returns by year, S&P 500 ETF, 2000 to 2012

- Overnight, close to open
- Session, open to close
Overnight and session returns by year, S&P 500 ETF, 2013 to 2025

- Overnight, close to open
- Session, open to close
Two biases in the data matter: dividend-adjusted prices place each dividend in the overnight part on its ex-date, which flatters it, and in our S&P 500 index series the open equaled the previous close on 94.8% to 98.0% of days in each year from 2000 to 2005, against 0.0% to 0.4% each year since 2014, which is why we used the ETF. On the ETF, rounded to three decimals, the open equaled the previous close on at most 5.2% of days in any year since 2000 (in 2017).
Does it work?
We cannot say for the SMI: no long intraday test is available to us, and the main source behind it says it stopped keeping pace with the market after 2000. The daily split does show that, on this ETF, most of the gain since 2000 came between the close and the open, but with years when the opposite held, and without telling anything about who traded when. Our bench measures chart patterns detected by rule against random entries, not indicators (see our methodology); the gap page covers the opening moves that make up the overnight part.
For other volume and sentiment tools, see the Money Flow Index, the Chaikin Money Flow and the page on divergence, or the full list of indicators.
Frequently asked questions
What is the Smart Money Flow Index?
A sentiment indicator based on the Smart Money Index but with a proprietary formula, featured on Bloomberg Professional according to Wikipedia, which cites no source for it. Its exact calculation is not published.
How is the Smart Money Index calculated?
Each day, take yesterday's value, subtract the Dow's change in the first 30 minutes and add its change in the final hour.
Who created the Smart Money Index?
It is not settled. Wikipedia says it was invented and popularized by Don Hays but gives no source for that. The Hertler Market Signal newsletter says it was described by Lynn Elgert in Barron's on February 22, 1988, and Wikipedia repeats that sentence.
What is the difference between the SMI and the Money Flow Index?
The Money Flow Index is an oscillator built from price and volume on each bar. The SMI uses no volume, only the time of day at which the Dow moves.
Can the Smart Money Index be calculated with daily data?
No, it needs the first half-hour and the final hour of each session. Daily bars only allow a split between the overnight move and the session move, which is a different measure.
