Renko, Kagi and Three-Line-Break — charts without time
The answer
Renko, Kagi and Three-Line-Break draw a new mark only when price moves by an amount you chose, so the horizontal axis stops being time. They make a trend easy to see by deleting every move smaller than your threshold, and the honest description of that is not "removing noise". It is removing information, some of which was noise and some of which was not.
Why this costs you money
A trader builds a Renko system. He tests it on 5 years of data and the results are remarkable: a high win rate, few trades, long clean runs. He funds an account and trades it. After 8 months he is down.
Three separate things went wrong and all 3 were in the chart, not the market.
The prices he tested at did not exist. A Renko brick has an open and a close that are computed from the brick size. They are grid values. Nobody transacted at them. A backtest that buys at a brick open is buying at a number the market never printed.
The bricks appeared late. A brick is drawn when the move is already complete. If the brick is 20 rupees, the price has already travelled 20 rupees before you see the brick. On a reversal, the classic rule needs 2 bricks, so the price has travelled 40 rupees against the old direction before the chart admits anything changed.
The chart was redrawn behind him. He used his platform's default brick size, which was based on Average True Range, a measure of recent volatility. That value changes as new data arrives, and on many platforms the whole history is then redrawn with the new value. The chart he tested on was not the chart he traded.
There is a fourth cost and it is the slow one. A Renko chart of a stock that did nothing for 8 months and then moved 25% in 4 sessions looks the same as a stock that moved 25% steadily over a year. Time is gone, and your stop distance, your position size and your patience all depend on time.
How it works
All 3 charts share one design. You choose a number that defines a meaningful move. Anything smaller is not drawn at all.
Renko
Renko is a chart of fixed-size bricks. You choose the brick size, which this article calls B.
A new brick is added only when the price moves B beyond the top of the last brick, or B beyond its bottom. The new brick is drawn one column to the right and one brick higher or lower. Every brick is the same size.
To reverse direction, the classic version requires the price to move 2 × B against the current direction, because the reversal brick must start from the opposite edge of the last brick. That is the source of Renko's lag, and the lag is exactly 2 brick sizes, always.
There are 3 common ways to set B.
| Method | What it does | The problem |
|---|---|---|
| Fixed points or rupees | B is a constant number | Becomes tighter as the price rises |
| Percentage | B is a percentage of price | Stable across price levels |
| ATR | B is set from recent volatility | Changes over time, and the chart can be redrawn |
The ATR method is the default on several platforms and it is the one to be most careful with. A brick size derived from data after the bar it is applied to is information you did not have at the time.
Some platforms also draw a "wick" version of Renko, showing how far price went inside a brick before completing it. That version is more honest and less clean. Classic Renko has no wicks at all.
Kagi
A Kagi chart is a single continuous line that changes thickness.
The line extends upward while price makes new highs. When price falls by the reversal amount, which you choose, the line turns and moves down in a new column. It keeps going down until price rises by the reversal amount.
The thickness is the second signal. The line is drawn thick when it rises above the previous shoulder, which is the last high point in the line. It is drawn thin when it falls below the previous waist, which is the last low point. The Japanese terms are yang for the thick line and yin for the thin one.
So a Kagi chart carries 2 pieces of information at once: the current direction, from the line, and whether the last swing extreme was broken, from the thickness. That second piece is genuinely useful and it has no equivalent on a Renko chart.
Three-Line-Break
Three-Line-Break is built from closing prices only.
Draw a line up whenever the close is above the top of the previous line. Draw a line down whenever the close is below the bottom of the previous line. To reverse, the close must break beyond the extreme of the last 3 lines. That is where the name comes from.
The number 3 is a setting. A 2-line break reverses sooner and gives more signals. A 5-line break reverses later and gives fewer.
Because it uses closes only, this chart deletes the entire intraday range. A session that fell 6% and closed flat contributes nothing.
What all 3 have in common
- A threshold you chose. The chart is a function of your setting and the market together, not of the market alone.
- No time axis. One column can be 3 minutes or 3 months.
- Volume becomes unreadable. A volume bar under a brick covers whatever period the brick took, and that period is different for every brick.
- Gaps are handled by invention. If a stock gaps 10 brick sizes overnight, most platforms print 10 bricks at once, at 10 price levels where nothing traded.
- A fixed, known lag. 2 bricks for Renko. The reversal amount for Kagi. 3 lines for Three-Line-Break.
What it tells you, and what it does not
These charts tell you the sequence and direction of moves larger than your threshold. That is a real and clean piece of information. If you want to know how many 5% moves this stock has made in each direction over 2 years, a Renko chart answers it faster than anything else.
They also force a discipline that ordinary charts do not. You must state, in a number, what size of move you consider meaningful, before you look. Most chart readers never do this.
They do not tell you when. Not the date, not the duration, not the speed.
They do not tell you the extremes. The highest price traded may sit inside a brick that was never completed, so it is not on the chart at all.
They do not tell you anything about volume, because the time period behind each mark varies.
And they do not tell you whether your threshold was the right one. There is no correct brick size. There is only a brick size that matches the size of move you care about, which is a question about your stop and your holding period, not about the stock.
The decision rule
Set the threshold from your own risk, not from the chart.
If your stop is 3% away, a brick smaller than about 1.5% is drawing moves you would never act on, and a brick larger than 3% will not show a reversal until after you have already been stopped out.
Then test it. Halve the setting and double it. If the story of the chart changes, the story was in the setting. If it survives all 3 settings, it is probably in the stock.
And never read a price off one of these charts. Brick edges are grid values. Read the trend here; read the price from the ordinary candlestick chart.
Try this now
This takes about 5 minutes and it does 2 things: it shows you exactly what the filter deleted, and it proves that the setting is a choice.
- Open the daily candlestick chart of a stock you own or follow. Set the range to 1 year. Write down 3 things: the date of the largest single-day range, the date of the largest gap between one close and the next open, and roughly how many trading sessions are on the screen (about 250 for a year).
- Now switch the chart type to Renko. Change the brick setting from ATR or "Traditional" to a fixed percentage of 1%, or to a fixed value close to 1% of the current price. Turn wicks off if your platform offers the option.
- Find your 2 dates from step 1 on the Renko chart. Write down what happened to each one. The gap will have become a stack of identical bricks with nothing marking it as a gap. The large-range day will have become several bricks that look exactly like a slow move over several weeks.
- Count the direction changes — every place the bricks switch from up to down or down to up. Write the number.
- Change the brick size to 0.5%. Count the direction changes again. Then change it to 2% and count again.
- Last step, 20 seconds. At the 2% setting, look at the top edge of the last brick and compare it with the last traded price on your ordinary chart. Write down the difference.
What you should see. Four things, and each one is a section of this article made concrete.
The number of direction changes changes enormously. Halving the brick size usually more than doubles the count. On a typical large stock over 1 year you might see something like 30 at 0.5%, 14 at 1% and 5 at 2%. Those are 3 different descriptions of the same year, and all 3 are correct.
At one of the 3 settings, the current direction is different from the others. This is the finding worth keeping. Ask yourself which one you would have believed if you had only opened that one.
The gap disappeared as an event. It is still there as price movement, but several of those bricks sit at prices where no share changed hands.
The last brick edge is away from the last traded price, by up to 1 full brick size. That difference is what you would have been wrong by if you had placed an order from this chart.
Three real cases
1. Steve Nison, "Beyond Candlesticks", 1994 — how these charts reached the West Steve Nison introduced Western traders to Japanese candlesticks in 1991. His second book, Beyond Candlesticks, published by John Wiley & Sons in 1994, described Kagi, Renko and Three-Line-Break for what appears to be the first time in English. Almost every later description traces to that book, including the standard formulas and the yin and yang terminology.
Two facts about that lineage matter. The Japanese origins are usually dated to the 1870s, when Japan opened a stock exchange, and the Kagi chart is often said to have been used by rice traders before that. And there is almost no independent academic literature testing any of the 3 chart types. They were adopted because they look clear, not because evidence supported them.
2. Sidney Alexander, 1961, and Fama and Blume, 1966 — the same idea, tested 40 years earlier This is the case that Renko tutorials never mention, and it is the most relevant evidence that exists.
Sidney Alexander published "Price Movements in Speculative Markets: Trends or Random Walks" in the Industrial Management Review in 1961. He proposed the filter rule: buy when a price rises x% from a low, sell when it falls x% from a high, and ignore every move smaller than x%. Read that sentence again. It is a Renko chart written as an instruction. Alexander reported that filter rules appeared to beat buying and holding.
In 1966 Eugene Fama and Marshall Blume published "Filter Rules and Stock Market Trading" in the Journal of Business. They tested filter sizes from 0.5% up to 50% on individual Dow stocks. Small filters produced very large numbers of trades. Once realistic transaction costs were subtracted, the rules did not beat buying and holding.
The lesson is not that Renko is worthless. It is that the core mechanism of a Renko chart was defined, tested and found unprofitable after costs, in a mainstream journal, before most Renko traders were born. Anybody selling a Renko course should be able to explain why their version escapes that result.
3. The Nifty 50 and the S&P 500, February and March 2020 — the cost of deleting time Both indices fell heavily from late February 2020 and turned upward in the second half of March 2020. On 13 March 2020 the Indian market-wide circuit breaker was triggered and trading halted. The Nifty 50 fell by a very large percentage in single sessions during that period, and the S&P 500 had several daily moves above 9% in both directions .
Draw that period as a fixed-brick Renko chart. You get a tidy stack of down bricks and then a tidy stack of up bricks. It looks like an orderly decline followed by an orderly recovery.
Now compare it with the candlestick chart. The decline took about 5 weeks. Some of those bricks were produced in a single session, and some took a week. A position sized for a brick that normally takes 3 sessions was suddenly holding through 6 bricks in a day.
Time is not decoration on a chart. Time is how fast you can be wrong. These charts remove it, and they remove it most completely in exactly the conditions where it matters most.
The question that resolves it
A novice asks: which of these charts shows the trend most clearly?
An expert asks: what size of move does this chart call meaningless, and is that the size that would hurt me?
The second question has an answer and the first does not. If your stop is 3% away and your brick is 5%, the chart is treating a move that would end your position as too small to draw. That is not a clean chart. That is a chart that has deleted your risk.
What would make this wrong
The claim in this article is that these charts are filters with a threshold you choose, and that the filtering removes signal along with noise.
It would be wrong if the same trend rule, run on Renko bricks and on ordinary closes, over the same instruments and the same period, with entries and exits at real traded prices in both versions, produced better results on the Renko version out of sample. That is a fair test and it is not hard to run. I am not aware of a published result showing it.
Three honest limits on the sceptical position.
First, the discipline is real even if the chart is not. Forcing a trader to state a threshold in advance genuinely reduces the amount of improvised decision-making. That benefit is about the person, not about the market, and it is still a benefit.
Second, the Kagi thickness rule contains more than a filter. It marks the breaking of the previous swing high or swing low, which is a real event on a real price series. That part is not arbitrary, and Kagi deserves separate treatment from Renko for that reason.
Third, these charts suit some instruments much better than others. An instrument that trades continuously loses far less to this kind of filtering than a stock that jumps. That is an argument about instruments, not about the method.
In India
Price bands cap how many bricks a session can produce. Indian stocks carry a daily price band, commonly 2%, 5%, 10% or 20% depending on the stock, and stocks in the derivatives segment have a wider dynamic band. If your brick is 5% of the price and the band is 5%, then no session can ever produce more than 1 brick, no matter how large the imbalance between buyers and sellers is. The chart cannot express pressure beyond the band, so the pressure is simply invisible.
A band-locked sequence looks like a calm trend. During February 2023 several Adani group stocks hit their lower price bands on consecutive sessions. Drawn as Renko, that is a neat column of identical down bricks. It reads as an orderly decline. What actually happened is that on several of those days there were sellers and almost no buyers at the band price, which is the opposite of orderly.
Market-wide circuit breakers stop the chart entirely. The Indian index circuit breakers halt trading at 10%, 15% and 20% moves in the Nifty 50 or the Sensex, with the halt length depending on the level and the time of day. A halted market produces no bricks. The chart shows a pause that looks like calm.
Corporate actions produce fake bricks. Indian companies split, issue bonus shares and pay large dividends more often than US companies do. If your data series is not adjusted, a 1-for-1 bonus creates an instant 50% fall, which on a 1% Renko chart is about 50 consecutive down bricks. Check that your platform is using an adjusted series before you trust any of these charts.
Liquidity is invisible here. A Renko chart of a stock trading 500 shares a day looks structurally identical to a Renko chart of the Nifty. Ordinary charts at least show you thin, ragged candles. These charts hide thinness completely, and India has a very long tail of thinly traded stocks.
In the United States
Futures suit these charts better, and the reason is structural. US index futures trade for close to 23 hours a day. Gaps between sessions are small and rare, so the number of invented bricks across gaps is small. The E-mini S&P 500 contract also has a fixed minimum price movement of 0.25 index points, so a sensible brick size is a whole multiple of that tick.
The session setting changes the chart. A CME futures chart can be drawn using the full electronic session or using regular trading hours only. Those 2 settings produce different highs, different lows and therefore different bricks from the same day. Two traders comparing Renko charts of the same contract very often have different session settings and do not know it.
US equities have real trades inside the gaps. Pre-market trading runs from 04:00 and after-hours trading runs to 20:00 Eastern time. When a stock gaps after an earnings release, the prices inside that gap were mostly traded, in the extended session. So the bricks a platform prints across a US equity gap are approximately honest. They correspond to prices at which business was done, even though the regular-hours chart does not show it.
Decimalisation changed what a small brick means. Until 2001 US stocks were quoted in fractions, and the smallest increment was often 1/16 of a dollar, which is 6.25 cents. Any brick smaller than that was meaningless. After decimalisation, very small bricks became possible, and mostly became a way to generate trades rather than information.
Where they differ, and what that tells you
The same brick size produces a different chart in the 2 markets, for reasons that have nothing to do with the companies.
In India, the range of a session is truncated by design. The band sets a hard maximum on the day's move, so the maximum number of bricks per session is fixed by regulation. In the United States there is no equivalent limit on a single stock's daily move. A US stock can produce 40 bricks in a session; an Indian stock with a 5% band and a 1% brick can produce at most 5.
What that tells you is that a brick count is not comparable across the 2 markets. If you tested a rule such as "enter after 6 consecutive up bricks" on US data, that condition means something different in India. In India it is partly a statement about how many sessions passed, because each session can contribute only so much. The rule has quietly changed from a statement about price into a statement about time, which is the one thing these charts were supposed to remove.
The second difference is what a gap contains. An Indian single-stock gap is empty. There was no trading between 15:30 and the next pre-open. When your platform prints 8 bricks across that gap, all 8 are fiction. A US equity gap usually contains real extended-hours trades, so the same 8 bricks are approximately real, and you can go and confirm them by switching extended hours on.
The rule that follows applies to every filtered chart in this cluster. The more continuously an instrument trades, the less a filter costs you. The more it jumps, the more the jumps were the information. India's single stocks jump more and are stopped more often. That makes these 3 chart types a better fit for US futures than for Indian equities, and the method did not change at all. The instrument did.
Carry this
- Renko, Kagi and Three-Line-Break are filters. You choose the threshold, and the chart is a function of your choice.
- Brick edges are grid values. Nobody traded at them.
- The lag is fixed and knowable: 2 bricks, or the reversal amount, or 3 lines.
- Halve the setting and double it. If the trend changes, the trend was in the setting.