Technical analysis: foundations

What this cluster is for

A chart is a record of what already happened. Nothing more than that.

Every chart type is a decision about what to throw away. A line chart throws away the day's range. A candlestick keeps the range and throws away the path inside it. Heikin Ashi throws away the real open, the real close and every gap. Renko throws away time. Point and Figure throws away time and every move smaller than a number you picked. A volume profile throws away the order of events and keeps something no other chart has: how much actually traded at each price.

None of those choices is wrong. Not knowing which one you made is.

That is what these 9 lessons are for. Most chart teaching gives you a vocabulary of shapes and a promise. These give you 2 things instead: what the evidence actually supports, which is less than you were told and more than the sceptics admit, and a precise account of what each chart type deleted from your screen.

Each article ends with something you do on your own charts, with your own holdings, in about 5 minutes.

By the end of this cluster you will be able to:

• Separate the parts of technical analysis with real evidence behind them from the parts with almost none
• Read a candle, and then count whether the pattern you like works on your own stock
• Name exactly what each chart type removed before you make a decision from it
• Recognise when a chart's "obvious trend" is a property of a setting you chose
• Find the price at which the market actually did its business, and see where your own entry sits relative to it
• Ask the 3 questions that make you very hard to fool with a backtest

Nothing here names a stock to buy. Nothing here is a signal. Everything here should still work in 2036, on a company that does not exist yet.

The reading order

The first 3 articles decide whether any of this is worth your time. Read them first, in order.

Is any of this real?

1. What technical analysis is, and what it is for — the 3 assumptions underneath every chart, and where each one breaks.
2. Does technical analysis actually work? — it is not 1 question. Trend and momentum have real support. Named patterns have very little.
3. The history of technical analysis, and Dow Theory — Charles Dow never wrote a theory. Other people built one from his editorials, and parts of it still hold.

The chart itself

1. Chart types — line, bar and candlestick — the same prices drawn 3 ways, and what each one hides.
2. Reading candlesticks, and counting your own patterns — what a body and a wick mean, and the count that tells you whether a pattern works for you.

Charts that filter

Each of these removes something on purpose. The lesson in all 3 is the same: a filter cannot tell signal from noise, because it has no way to know which is which.

1. Heikin Ashi, and what averaging removes — the open and close it shows you are numbers nobody traded at, and it cannot draw a gap.
2. Renko, Kagi and Three-Line-Break — charts without time — halve the brick size and the trend changes. That proves the trend was in the setting.
3. Point and Figure — the oldest chart, and its false targets — precise signals, and price objectives that are width multiplied by 2 numbers you chose.

The chart that adds something

1. Market Profile and Volume Profile — where volume actually traded, rather than where price went. The only chart in this cluster that gives you a fact the others cannot.

The checklist

Every article ends in an action. Collected here they are one long evening's work, and they are the most valuable thing in this cluster. Tick them off as you go.

Test the claims before you trust them

• For 3 stocks, find 1 dated announcement each and check the 10 sessions before it. Mark yes, no or unclear. (1)
• Cover the right side of a chart and forecast 20 dates, with a confidence score of 1 to 5. Then score what "always up" would have got. (2)
• Count non-confirmations between 2 averages over 2 years, then find the base rate for the same decline without any signal. (3)
• Write 1 candlestick pattern as a numeric definition, find every instance in 2 years, count wins and losses, then find the base rate. (5)

Find out what your chart threw away

• Draw the same stock and the same 6 months as line, then bar, then candlestick. Write down what each version added. Then switch to log scale and to weekly. (4)
• Find your 3 largest gaps on a candlestick chart, switch to Heikin Ashi, and watch them disappear. Then compare 1 Heikin Ashi close with the real close. (6)
• Put a Renko chart of 1 year beside the candlestick chart of the same year. Write down what happened to your largest gap and your largest range day. (7)

Prove that the setting is a choice

• On the Renko chart, count direction changes at 0.5%, 1% and 2% brick sizes. At least one setting will disagree with the others about the current direction. (7)
• On a Point and Figure chart, count columns and read the current signal at 3 box sizes. Then compute the horizontal count objective at 2 of them and compare. (8)

Find out where you actually bought

• Add a volume profile over the last 3 months of a stock you own. Write down the point of control. (9)
• Write down your own average buying price and work out (your price − POC) ÷ POC × 100. (9)
• Check whether your average price sits inside or outside the value area, and find the thinnest part of the profile. (9)

Habits that cost nothing

• Before any decision from a filtered chart, say out loud what it deleted.
• Never read an order price off Heikin Ashi, Renko or Point and Figure. Those values are computed, not traded.
• Before believing any backtest, ask 3 things: was the rule written down first, how many rules were tried, does it survive costs and a later period.
• Before calling a level support, check how much volume actually traded there.

India and the United States, equally

Every article covers both markets at the same depth, then does the thing most chart teaching skips: it says where the 2 differ and what that difference tells you.

On this subject the differences are unusually concrete, because they change numbers rather than opinions. Indian price bands cap a session's range, so they change how many bricks and boxes a chart can draw and they create volume nodes that are records of a limit rather than of an agreement. Indian turnover concentrates in a small number of names, so a volume profile outside the liquid stocks is built from very few trades. In the United States, most default profiles exclude the pre-market and after-hours sessions, which is exactly where earnings are repriced, so the point of control is computed from a chart that is missing the quarter's largest event.

A chart method developed on US futures does not arrive in an Indian single stock unchanged. Those sections are the translation.

A note on what this is not

This cluster is not a pattern catalogue and it will not give you a signal.

It is the part of technical analysis that survives being asked for evidence, plus an honest account of the part that does not. If that leaves you with fewer tools than you expected, that is the intended result. A smaller number of instruments you understand precisely will cost you far less than a large number you half believe.

Related

• ← All 22 topics
• How the market works — the order book, liquidity and market makers
• The tools