Confluence and confirmation — and the overload that fakes it
The answer
Confluence means several independent pieces of evidence pointing the same way. The word doing the work is independent.
Almost every indicator on a chart is computed from the same closing prices. When 5 of them agree, that is usually not 5 opinions. It is 1 opinion, restated 5 times, and the agreement contains no extra information at all.
Why this costs you money
The loss here does not come from a bad indicator. It comes from false confidence, and false confidence changes position size.
Here is exactly how it happens. You are considering a trade. RSI is above 50. MACD has crossed up. The Stochastic is rising. Price is above its 20 day moving average. CCI has crossed +100. Five indicators agree.
You feel certain, so you take a position 3 times your normal size.
All 5 of those indicators are functions of the closing price over roughly the same window. RSI compares average up closes to average down closes. Stochastic compares the close to the recent range. MACD subtracts one average of closes from another. CCI measures the distance of the typical price from its average. The moving average is the average of closes.
If the last 15 closes rose, all 5 will be positive. They cannot do anything else. Their agreement is arithmetic, not evidence. You have taken triple size on the strength of a fact you already knew when you looked at the price.
The second cost is slower and it removes traders from the market permanently. As you add indicators, the number of days on which all of them agree collapses. A trader with 7 conditions may find 4 valid setups a year. They then start allowing "6 out of 7", which means they now have a rule that they overrule, which is the same as having no rule.
The third cost is the one nobody admits to. With enough indicators on the chart, you can always find one that supports the trade you already wanted to take. Indicator overload does not improve decisions. It supplies justifications.
How it works
Every indicator has inputs. That is the only thing that matters for independence, and it is a fact you can check in one minute.
Here is what is actually available to a retail trader, grouped by input:
| Input family | What it is | Examples |
|---|---|---|
| Closing price | The single number most indicators use | RSI, MACD, Stochastic, moving averages, CCI |
| Range (high and low) | How far it moved within the bar | ATR, Bollinger Band width, Supertrend, Ichimoku lines |
| Volume | How many shares changed hands | OBV, VWAP, volume bars |
| Other stocks | What the rest of the market is doing | Advance-decline, new highs versus new lows |
| Option prices | What people pay for protection | India VIX, CBOE VIX, open interest |
| The company | Earnings, debt, cash flow | Valuation, results, filings |
| The calendar | What is scheduled | Results dates, policy meetings, budget |
Notice how lopsided that table is. The first row alone contains most of the indicators on a typical retail chart. Rows 4, 5, 6 and 7 contain almost none of them, and those are the rows with genuinely different information.
The independence rule is simple. Two indicators are independent if they read from different rows. Two indicators from the same row are one indicator with 2 names, no matter how different their formulas look or how different their charts appear.
There is a second dimension, and it is worth 1 sentence: timeframe. RSI on a daily chart and RSI on a weekly chart use the same input but different windows, so they are partially independent. That is a weaker kind of independence than a different input, but it is real and it is cheap.
What real confluence looks like
A defensible setup uses at most 4 checks, from 4 different rows:
- A price level. Support, resistance, a prior high, a round number that matters. This is from the chart, not from an indicator.
- One trend or momentum reading. Choose 1. Not 3. They are the same row.
- One volume or breadth reading. This is the row most people skip, and it is the one that adds the most.
- A calendar check. Is there a results announcement, a policy decision, or a budget inside your holding period?
That is it. Four checks, from 4 different inputs. It is a shorter list than most traders use and it carries more information.
What it tells you, and what it does not
Real confluence tells you that more than 1 kind of evidence supports the same conclusion. That genuinely raises the probability, because for both to be wrong, 2 different things have to be wrong.
It tells you when to size up, modestly. If your normal position is 1 unit, real confluence might justify 1.5. It does not justify 3.
It does not create certainty. Even 4 independent checks can all point the wrong way, because they are all backward-looking and the future arrives with new information.
It does not fix a bad idea. Confirmation applied to a trade with no edge produces a well-documented losing trade.
And agreement between correlated indicators tells you nothing. This is the central claim of the article and it is worth stating as plainly as possible: if 2 measurements are computed from the same numbers, their agreement is a property of arithmetic and it carries zero additional information about the future.
The decision rule
Before adding any indicator to your chart, ask 1 question: what input does this read that nothing already on my chart reads?
If the answer is "closing price", and you already have something reading closing price, do not add it. You are adding a second opinion from the same source.
Cap the chart at 3 indicators. One from the price and momentum row. One from the range or volume row. One optional third from a genuinely different row.
Require agreement across rows, not within a row. RSI and MACD agreeing is not confluence. RSI and volume agreeing is.
If your checks are so numerous that valid setups appear fewer than 12 times a year, remove checks. A rule you cannot follow often enough to learn from is not a rule.
There is one more instruction, and it is uncomfortable. Decide your indicator set before you look at the trade. Choosing which indicators to consult after you have formed an opinion is not analysis. It is shopping for agreement, and it is the most common way experienced traders fool themselves.
Try this now
Two parts. The first takes 2 minutes, the second takes 3.
Part 1. The input audit
- Open the chart you actually use, with all the indicators you actually have on it. Write down every indicator by name.
- Beside each, write down its inputs — closing price, high and low, volume, or something else. Use the table above.
- Count how many distinct inputs appear in your list.
What you should see. Most readers will list 4 to 7 indicators and find that they read 1 or 2 distinct inputs. If everything on your chart reads closing price, you have 1 opinion and a lot of decoration, and you can delete most of the chart today with no loss of information.
Part 2. The agreement test
- Set your chart to 6 months, daily.
- Go through it and mark every day on which all your indicators agreed on the same direction. Count those days.
- Then measure what price did over the next 10 trading days after each of those agreement days.
- Compare that to what price did over 10 days starting from a random selection of other days in the same 6 months. Pick 10 dates without looking at the chart — for example, the 5th and 20th of each month.
What you should see. One of 2 results, and both are useful.
If the agreement days number more than about 40 out of 120 trading days, your indicators are highly correlated. They agree constantly because they are the same measurement. In that case, the comparison in step 4 will show almost no difference, and you have just proved that your confirmation process is not selecting anything.
If the agreement days number fewer than 5, your checks are so restrictive that you have no usable sample, and any impression you have of them working is based on 3 or 4 examples.
The result you want is somewhere in between, with a visible difference in step 4. If you get it, you have found something worth keeping. Most people do not get it the first time, and finding that out in 5 minutes is worth more than another year of believing.
Three real cases
1. NIFTY 50 and SENSEX — confirming an index with itself Indian traders routinely check whether the NIFTY 50 and the SENSEX are moving together, and treat the agreement as confirmation of market direction. The SENSEX holds 30 companies and the NIFTY 50 holds 50, and the large majority of the SENSEX's weight is also in the NIFTY. The 2 indices are largely the same companies weighted slightly differently. When they agree, nothing has been confirmed. On the rare days they disagree meaningfully, that disagreement is information — but the agreement, which is nearly every day, is not. This is the cleanest everyday example of 2 measurements that look independent and are not.
2. Long-Term Capital Management, September 1998 — independence assumed, not verified LTCM was a US hedge fund founded by John Meriwether, with Myron Scholes and Robert Merton among its partners, both of whom had received the Nobel Memorial Prize in Economic Sciences in 1997. The fund held a large number of positions across many markets that were believed to be independent, so that losses in one would be offset elsewhere. In 1998, following Russia's default, those positions moved against the fund at the same time. The fund lost most of its capital within weeks and was recapitalised in an arrangement organised by the Federal Reserve Bank of New York in September 1998. The lesson translates directly and painfully to a chart: diversification you have not verified is concentration you cannot see. Five indicators believed to be independent behave exactly like a portfolio of positions believed to be independent.
3. S&P 500 and Nasdaq-100, 2023 — two indices, the same seven companies Through 2023, US equity index gains were unusually concentrated in a small group of very large technology companies. Both the S&P 500 and the Nasdaq-100 rose strongly, and both were carried substantially by that same small group. A trader checking "are both major US indices confirming this move?" was checking the same 7 companies twice. The equal-weighted version of the S&P 500 tells a different story from the standard version over that period, and comparing the 2 is a genuine independence check available to anyone. Article 19 shows how.
The question that resolves it
A novice adds an indicator and asks: does this one agree?
An expert adds an indicator and asks: what does this one read that the others cannot?
The first question can always be answered, and the answer is usually yes, which is why it produces confidence rather than information. The second question is answerable in one minute, is answerable before you place any trade, and eliminates most of what is on a typical chart.
What would make this wrong
If correlated indicators added real information, then a chart with 6 price-derived indicators would produce better outcomes than the same chart with 2. Anyone with a trade history can look for this. Compare your results from periods when you used a crowded chart against periods when you used a simple one, on the same instruments.
The honest limits are 4.
First, correlation is not identity. RSI and MACD are highly related but not identical, and there are configurations where one turns and the other does not. This article argues that the extra information is small, not that it is zero. The question is whether it is large enough to justify the confidence people take from it, and it is not.
Second, the input table is a simplification. Bollinger Bands use closing prices for the middle line and standard deviation of closes for the bands, so they sit partly in the price row and partly in the range row. Ichimoku uses high and low midpoints, which is a different input from a moving average of closes, even though both are trend tools. Do the audit honestly rather than mechanically.
Third, timeframe independence is real. A weekly reading and a daily reading of the same indicator are not fully redundant. Using RSI on 2 timeframes is a weaker form of confluence than using RSI and volume, but it is not nothing.
Fourth, and against the article's own argument: a checklist has value beyond its information content. Requiring 4 conditions slows you down and prevents impulsive entries. If your indicators are correlated but the ritual of checking them stops you trading at 10:05 am on a whim, the ritual is earning its place. Just be clear about which benefit you are receiving. Do not size a position as though you have 4 pieces of evidence when what you actually have is 1 piece of evidence and a good habit.
In India
Indian retail charting culture leans heavily towards stacked indicator setups, and templates with 5 or more indicators are widely shared and taught.
Three Indian features matter.
The available independent inputs are fewer. Volume data is published and reliable on NSE and BSE. Breadth data exists. But long, easily charted histories of breadth series are less accessible to Indian retail traders than in the US, and option-implied data beyond India VIX is not widely used at retail level. That means the rows in the table that add the most information are the rows Indian traders have the least access to, which partly explains why so many charts are crowded with price-derived indicators instead.
Index concentration limits what index breadth can tell you. NIFTY 50 has a small number of very heavily weighted constituents. The index can rise while most of its members fall. That is a fact about index construction, not a warning signal, which makes breadth a less clean confirmation source here than in the US.
Weekly index option expiries add a calendar input that is genuinely independent. The pattern of behaviour around Indian index option expiry is driven by positioning rather than by price history, so knowing where you are in the expiry cycle is real information that no chart indicator contains.
In the United States
US retail charting is equally prone to overload, but the set of genuinely independent inputs available is wider.
Three US features matter.
Breadth data is deep and published. The NYSE advance-decline line, new highs versus new lows, and the percentage of stocks above their 200 day moving average are all long-running series, freely charted. These read a different input — the number of individual stocks doing something — and are therefore real confirmation for an index signal.
Equal-weight index versions exist and are easily compared. There is a widely followed equal-weighted version of the S&P 500. Comparing it to the standard capitalisation-weighted index is a 1 minute independence check that tells you whether an index move was broad or narrow.
Options data is accessible at retail level. VIX, put-call ratios and open interest are published and widely charted. These read what people are paying for protection, which is an input no price indicator contains.
Where they differ, and what that tells you
The formulas are identical in both markets. What differs is how many genuinely different rows of the input table a trader can actually reach.
A US retail trader can, in 5 minutes and for free, consult price, range, volume, breadth across thousands of stocks, an equal-weight version of the index, and options-implied volatility. That is 6 distinct inputs. Real confluence is available to them.
An Indian retail trader has easy access to price, range, volume, India VIX and same-day advance-decline figures. That is fewer rows, and 1 of them — index breadth — is compromised by the index's own concentration.
What that tells you is where an Indian trader should look for independence instead, and the answer is not more indicators. It is:
- The individual stock's own volume, which is reliable and published, and which most Indian chart templates under-use relative to their 5 momentum indicators.
- The calendar, which is free and genuinely independent: results dates, RBI policy dates, the Union Budget on 1 February, and index option expiry.
- The sector, checked directly. If your stock is breaking out, look at 3 peers. If they are not doing anything similar, you have a single-stock event. If they are, you have a sector move. That is a real cross-sectional check available without any breadth series at all, and it takes 90 seconds.
That third item is the practical answer to India's breadth problem. You cannot rely on index breadth, so build your own tiny version of it by looking at 3 or 4 peers directly. It is the highest-value habit in this article for an Indian reader, and it requires no new tool.
Carry this
- Independence comes from different inputs, not from different formulas.
- Most chart indicators read the same closing prices. Their agreement is arithmetic.
- Cap the chart at 3, from different rows: price, volume or range, and one other.
- Choose your indicators before you form an opinion, not after.