Money Flow Index: The Volume-Weighted RSI That Is Not Chaikin Money Flow

The Money Flow Index is the Relative Strength Index with one extra term bolted into it. Same bounded 0 to 100 scale, same ratio-of-sums construction, same overbought and oversold framing. The difference is what gets summed. RSI adds up the size of price changes. MFI adds up dollars: the typical price of each bar multiplied by that bar's volume, sorted into a positive pile and a negative pile by whether the typical price rose or fell. That single substitution is why the same six periods can read 58 on RSI and 33 on MFI, and why the two lines parting company is the only thing worth running both of them for.

Quick Answer

The rest of this is the detail that paragraph skips: the arithmetic run through seven bars you can check by hand, the algebraic shortcut nobody prints, what volume-weighting actually changes against plain RSI and where plain RSI is the safer pick, failure swings, and the precise mechanical split between the three indicators with money flow in the name or the family.

What Is the Money Flow Index?

A single line in its own pane below price, running between 0 and 100, with horizontal reference lines conventionally drawn at 80 and 20. Gene Quong and Avrum Soudack are the names attached to it, and the design goal is stated in the name: the flow of money into and out of a security, rather than the movement of its price in isolation.

The premise underneath is that a price move backed by real dollars is a different event from the same price move on nothing, and that a price-only oscillator cannot tell them apart. Whether you buy that premise is a separate question from whether the arithmetic delivers on it, and the arithmetic does: the volume term is not decoration, it is the multiplier that decides how much each period contributes. A stock that grinds up for nine sessions on declining volume and then sells off twice on heavy volume will produce a rising price chart and a falling MFI, and neither of those is wrong.

It ships as a standard built-in on effectively every charting platform in common use, with a single period input and usually nothing else to configure. There is no proprietary version to hunt for and nothing to implement yourself. The reason to understand the formula is that every one of the indicator's blind spots is visible in the arithmetic and invisible in the plotted line.

What Is the Money Flow Index Formula?

Step three is the part that catches people, because the classification runs on typical price and not on the close. Those are different tests and they disagree more often than you would guess. A bar can close at exactly the same price as the one before it and still register as positive money flow, because its high and its low both shifted up and the average of the three moved with them. That is the first concrete split from a close-based volume indicator, and we come back to it further down.

Step four is where a bounded oscillator falls out of an unbounded ratio, and it is exactly RSI's trick. Feed it a ratio that can run from zero to infinity, and 100 minus 100 over one plus that ratio compresses the whole range into 0 to 100. A ratio of 1, meaning positive and negative flow are equal, gives 50 every time. That is worth committing to memory, because the 50 line is a fact about the formula rather than a convention someone chose.

That formulation matches the StockCharts ChartSchool entry on the Money Flow Index and Fidelity's technical indicator guide page for MFI, and unlike most oscillators there is very little settings argument to have about it beyond the lookback.

The shortcut the textbooks skip

Expand step four and the two-stage form collapses. MFI = 100 - 100N / ( N + P ), which is just 100P / ( P + N ). In words: MFI is positive money flow as a percentage of total money flow. Nothing is lost in the rearrangement and it is exactly equal, not an approximation.

Two reasons that matters. It makes the reading intuitive in a way the ratio form does not, because 33 now means a third of the dollar volume in the window happened on rising periods. And it handles the edge case the published form cannot: if every period in the lookback rose, negative money flow is zero and the ratio divides by zero, while 100P / ( P + 0 ) gives a clean 100. Platforms pin the line at 100 in that case for precisely this reason.

Here is the arithmetic over seven bars with a six-period lookback, shortened from 14 so every sum stays checkable by hand. The prices and volumes are made up, chosen to make the mechanics visible rather than to describe any real security. Bar 1 is the anchor: it has no previous typical price to compare against, so it is not classified into either pile.

Money Flow Ratio = 70,820,000 / 142,560,000 = 0.4968. MFI = 100 - ( 100 / 1.4968 ) = 33.19. The shortcut gets there in one step: 100 x 70,820,000 / 213,380,000 = 33.19. Same number, and you can check it on a phone.

Now look at what that window produced. Four of the six classified periods had a rising typical price. The stock finished higher than it started, 31.20 up to 31.80. And MFI came out at 33.19, nearer oversold than the midline. Nothing was rigged. The four advancing periods traded 2.2 million shares between them. The two declining ones traded 4.5 million. MFI does not count periods, it counts dollars, and the dollars went the other way.

Bar 3 is the detail worth pausing on. It closed at 32.10, exactly where bar 2 closed. A close-based volume indicator throws that bar away or scores it neutral. MFI counted all 16.1 million dollars of it as positive flow, because the typical price rose from 31.90 to 32.20 on a higher high and a higher low. Same bar, three different verdicts depending on which member of this indicator family you are running.

How Do You Read an MFI Chart?

Three reference lines and one honest caveat attached to each. Above 80 is the conventional overbought read, meaning more than 80 percent of the dollar volume in the window landed on rising periods. Below 20 is oversold, the mirror. The 50 line is exact balance, and it is the only level on the scale that means something mathematically rather than by convention.

Some sources narrow the extremes to 90 and 10 and describe those as the truly extreme readings, on the reasoning that 80 and 20 get tagged often enough on a normal chart that they are not rare events. Both conventions are in circulation, 80 and 20 by a wide margin, and neither is a tested parameter. What matters more than the number you pick is knowing that the bands are not symmetrical in practice: in a market that drifts up over time, an oscillator built on rising-period dollar volume spends more time in the upper half than the lower one, so 80 readings are more common than 20 readings on most equity charts.

The caveat that costs the most money is the one every reference source attaches to the thresholds and most tutorials then quietly ignore. Overbought is not a sell signal. A stock in a genuine trend can hold MFI above 80 for weeks while price keeps advancing, because a sustained one-way tape is exactly the input that produces a pinned reading. A threshold cross is a description of the last 14 periods, not a prediction about the next one, which is the same reason an RSI level on its own gets traders faked out so reliably. The conventional fix is to require something else to agree before acting, which is a question about how many independent signals you want lined up before you commit rather than a question about the oscillator.

Money Flow Index vs RSI

The two are the same machine with a different fuel line. Both take a stream of periods, sort them into a rising pile and a falling pile, sum something inside each pile, divide one sum by the other, and squash the result into 0 to 100 with the identical 100 minus 100 over one plus ratio step. Both default to 14 periods. Both get read against thresholds at the top and bottom of the range.

The difference is the thing being summed, and it is worth stating precisely because most write-ups round it off to "MFI adds volume" and stop. RSI sums the magnitude of the price changes. A period that gained 4.00 contributes eight times what a period that gained 0.50 contributes, and volume never enters the calculation at any point. MFI sums dollar volume. A period that gained 4.00 on 500,000 shares contributes less than a period that gained 0.01 on 2 million shares, because the size of the move is not a term in the formula at all. Neither indicator uses the other one's weighting. They are not the same reading with volume sprinkled on top.

Run both over the worked example above and the gap is visible. The closes were 31.20, 32.10, 32.10, 31.20, 32.40, 32.40, 31.80. Gains of 0.90 and 1.20 against losses of 0.90 and 0.60, which over six periods is an average gain of 0.35 against an average loss of 0.25, a relative strength of 1.4, and an RSI of 58.33. The same six periods scored by dollar volume give 33.19. One oscillator says the tape leaned modestly bullish. The other says two thirds of the money traded on down periods. Both are correct about what they measure.

There is a second, quieter asymmetry in the volume weighting that almost nobody mentions. Raw money flow is typical price multiplied by volume, so it is denominated in dollars, which means the price level affects the weight. In a stock that ran from 20 to 40 across the lookback, a bar near 40 contributes roughly twice as much per share as a bar near 20 did. Inside a quiet window that term barely moves and effectively cancels. Inside a window containing a large directional move it does not, and it tilts the reading toward whatever happened at the higher prices.

When plain RSI is the better pick

Volume weighting is only an upgrade when the volume figure means what the formula assumes. Two cases where it does not.

Spot forex is the clean one. Decentralized currency trading has no consolidated tape and no authoritative volume print, so what a platform labels volume on a spot pair is almost always tick volume, the number of quote updates inside the bar rather than the size traded. Multiply typical price by that and you get a number, but it is a number about update frequency, and two brokers will hand you two different MFI lines for the same pair on the same session. This is the same structural limit that applies to signing whole-bar volume into a cumulative running total and it is not fixable with a setting. On spot forex, RSI is reading a price series that genuinely exists, and MFI is reading a price series multiplied by a proxy. Exchange-traded futures do not have this problem, because the volume there is real and centrally reported.

The second case is long-horizon charts. On weekly and monthly bars, volume carries a lot of structure that has nothing to do with conviction: index rebalance weeks, quarterly expiration, holiday sessions, secondary offerings. A 14-week MFI is weighting its periods with all of that baked in, and the noise it introduces can outweigh the information the volume term was supposed to add. RSI over the same window is at least noisy about one thing instead of two. Intraday and daily equity charts, where volume is real and roughly comparable session to session, are where the volume term earns its place, which is the same conclusion treating volume as the participation check on an intraday move arrives at from the other direction.

One more honest note on running both. On most liquid charts MFI and RSI track each other closely for long stretches, which makes sense given they share a formula and a price series. If you are going to keep both on screen, the reason is the disagreements. Everything else is duplicated pixels. That is a different question from stacking a second derivation on top of RSI itself, which is what running a stochastic over the RSI line does, and the two are not substitutes for each other.

What Does an MFI Divergence Tell You?

Price and MFI making opposite statements about the same swing. There are two and they are mirror images.

Bullish divergence: price makes a lower low while MFI makes a higher low, or fails to confirm the new low at all. The read is that the second flush happened with less dollar volume behind the selling than the first one did, so the pressure driving price down is thinning. Bearish divergence: price makes a higher high while MFI makes a lower high, or fails to confirm. The second push was funded with fewer dollars than the first.

The caveat is the same one that applies to every divergence on every oscillator, and it is the sentence most tutorials put in a footnote. A divergence does not guarantee a reversal. It is a description of a disagreement that already happened, with nothing in the construction to say how long the disagreement is allowed to run or which side resolves it. A meaningful share of them close by the oscillator catching up to price rather than by price rolling over to meet the oscillator. Treat it as the reason to start watching, and require a separate trigger, a break of structure or a trendline break or a level reclaim, as the reason to act.

Duration does not strengthen it either. A divergence three bars old and one three months old look identical on the chart, and the long one has already been wrong once.

Failure swings

A second named signal, mechanically checkable, and unlike divergence it does not require you to look at price at all. It lives entirely inside the oscillator pane.

A bullish failure swing runs in four steps. MFI drops below 20 into oversold. It rallies back above 20. It pulls back but holds above 20, refusing to make a new low in oversold territory. Then it breaks above the high it made on that first rally, the prior reaction high. The bearish version is the exact mirror: MFI pushes above 80, drops back below it, bounces but fails to reclaim 80, then breaks below the low it made on the first drop. The signal is the break in step four, not the extreme in step one.

What makes the pattern more interesting than a bare threshold cross is the hold in step three. An oscillator that goes to an extreme and then refuses to go back is describing a change in who is willing to trade at those prices, and the structure gives you a defined level to trigger off rather than an arbitrary line. It is also the one MFI signal with a natural invalidation built in, because a break back through 20 on the bullish version tells you immediately that you were wrong.

Money Flow Index vs Chaikin Money Flow

Two words in common and almost nothing else. The names collide badly enough that the search results for one are full of the other, and plenty of guides treat them as the same indicator under two labels. They are not close.

MFI, as covered above, is a bounded 0 to 100 oscillator built from typical price times volume, classified by whether the typical price rose or fell against the previous period, summed over 14 periods, read for overbought and oversold against 80 and 20. Chaikin Money Flow scores each bar by where its close landed inside that bar's own high-low range, multiplies that score by volume, and divides the rolling sum by total volume over a window conventionally set at 20 or 21 days. The output runs from -1 to +1, centred on zero, and is read as accumulation against distribution rather than as overbought against oversold.

Every layer differs. Different inputs, since CMF needs the position of the close within the range and MFI needs only the average of high, low and close. Different comparison, since CMF never looks at the previous bar and MFI looks at nothing else to decide the sign. Different scale, different default period, different question. A CMF of +0.30 and an MFI of 78 are not two readings of the same thing at different resolutions.

The practical consequence is a specific one. Because CMF has no term for the open and no term for the previous close, it cannot see a gap at all, and a stock that gaps down hard but closes near the high of its own range scores as accumulation. MFI has the opposite behaviour, because the gap moves the typical price and the typical price is what gets compared. On a gap-heavy chart the two will disagree loudly and consistently, and neither is broken when they do.

Money Flow Index vs On-Balance Volume

The third member of the family, and the furthest from MFI of the three. On-Balance Volume is a cumulative running total with no lookback window at all: add the whole bar's volume on a close above the prior close, subtract the whole bar's volume on a close below it, change nothing on a flat close. No weighting term exists anywhere in it. Every bar counts fully, in one direction or the other, and a bar that closed one cent higher adds exactly as much as a bar that closed four dollars higher.

MFI is weighted twice over. Once by volume, like OBV. And once more by the price level, because raw money flow is typical price times volume rather than volume alone, so a period trading at 40 carries twice the weight per share of a period trading at 20. OBV is equal-weight-per-bar and unbounded. MFI is price-level-weighted, windowed at 14, and hard-bounded at 0 and 100. One produces a line whose absolute value is meaningless and whose slope is the whole signal. The other produces a number you read directly against fixed thresholds.

Bar 3 from the worked example is the cleanest illustration of the split. Close unchanged at 32.10, typical price up from 31.90 to 32.20, 500,000 shares. OBV discards the bar entirely under the flat-close rule, all 500,000 shares of it. CMF counts it, weighted by where 32.10 sat inside a 31.80 to 32.70 range. MFI counts all 16.1 million dollars of it as positive money flow. Three indicators, one bar, three different answers, and none of them is a bug.

The row worth arguing over is the last one. All three are blind to something, and the thing each one discards is roughly the thing another one is built around. Running all three is not triangulation, it is three partial views that will contradict each other by design, and a trader who takes the majority vote has invented a system nobody tested. Pick the one whose blind spot you can live with on the charts you actually trade, which is the same filtering logic behind the question of which indicators genuinely measure different things.

Where the Money Flow Index Breaks

None of these are faults in the formula. They are what a bounded rolling-window ratio does when its inputs stop behaving like the inputs it assumed.

Start with the structural two, because they apply always. It lags: every value is a sum over periods that have already closed, so MFI describes what dollar volume already did and contains no predictive term anywhere. And it is not a system. A number saying 82 percent of recent dollar volume landed on rising periods has no view on trend, level, structure or risk, which is why every reasonable write-up pairs it with something that does. It is also not a settings problem you can optimise your way out of. The TradingView support page for Money Flow (MFI) documents what one major platform exposes for the indicator, which is a length input and very little else.

The trend row is the expensive one and it deserves a sentence of its own. A strong uptrend produces exactly the input that pins MFI high: most periods rising, and the heavy volume arriving on the rising ones. The oscillator is not malfunctioning when it sits at 88 for three weeks, it is reporting accurately, and the trader fading it is betting against the only thing the reading is sure about. Every authoritative source on the indicator states some version of this and it is still the most common way people lose money with it.

The ageing-out row is the one that catches everybody eventually. MFI is a rolling-window statistic, so the line can step hard on a session where price barely moved, purely because a large period dropped off the back of the lookback. Nothing about the market changed. The window did.

Where MFI Fits on a Real Chart

It answers one question well: over the last 14 periods, what share of the dollar volume traded on periods where the typical price was rising. That is a genuinely useful thing to know and it is not readable off the price chart at a glance. What it cannot contribute is where price is, what it is approaching, whether the structure is clean, or how much you should risk. The sensible ordering is regime and structure first, then MFI as a participation check on what the first two already suggested, which is the same layering that runs through the wider technical analysis overview and through an intraday momentum playbook where volume confirms rather than triggers.

What a screenshot read can and cannot see here

Worth being straight about, since this site sells a tool. SnapPChart has no Money Flow Index field. It does not compute a typical price, it does not multiply anything by volume, it holds no 14-period lookback, and it carries no state for an MFI value the way it does for market structure and trend, breaks of structure and changes of character, liquidity sweeps, fair value gaps and order blocks, the moving average stack, the VWAP relationship, MACD, volume participation, support and resistance, and candlestick patterns. There is no RSI field either, which is worth saying in the same breath given how much of this post is about the two of them. What the engine does is read a chart screenshot you upload. So if you plot MFI in a lower pane on your own platform before taking that screenshot, that line is part of what the analysis sees, as a shape drawn beneath the price chart. That is a picture of an oscillator, not a verified computed value: it cannot confirm your period is 14, it cannot tell you the reading is being held up by one earnings-day volume bar from two weeks ago, and it cannot know whether the volume underneath it was shares or tick count. Which chart states a screenshot-based read genuinely carries and which it only infers from shape is the subject of the wider guide to how AI reads a chart, and a neutral description of what a single chart read covers sits on the AI chart analysis page. If MFI is the reason you are taking the trade, verifying MFI stays your job.