- The one honest justification for technical analysis, and its limits
- What OHLC actually compresses, and what it throws away
- Why timeframe choice changes the conclusion, and how to pick one
Technical analysis has a weak defence and a strong one. The weak defence is that patterns repeat because they always have. The strong one is structural, and worth stating precisely.
Every price on a chart is a place where somebody took a position. Those positions still exist. A chart is therefore a map of where other participants are committed, and at what price they become uncomfortable — which is real information about future supply and demand.
That framing does the work all course long. A resistance level is not magic; it is a price where a cohort of buyers is sitting at breakeven and would like to get out. A breakout on heavy volume is not a shape; it is evidence that the supply which used to sit at that price has been consumed.
What a bar compresses
One daily bar is a summary of thousands of trades. Four numbers survive the compression, and each answers a different question.
- OpenWhat it records: Where the auction started, after overnight newsWhat you infer: The gap from yesterday's close — how far opinion moved while the market was shut
- HighWhat it records: The furthest buyers were willing to chaseWhat you infer: Where supply appeared
- LowWhat it records: The furthest sellers were willing to pressWhat you infer: Where demand appeared
- CloseWhat it records: Where the auction settled with everyone watchingWhat you infer: The most informative of the four — who won the session
A bar does not tell you the order in which the high and low happened. A day that opened, collapsed to the low, then recovered to close at the high looks identical to one that rallied first and dipped later. Same four numbers, opposite stories.
This is not a flaw you can fix by staring harder — it is a reason to check a lower timeframe when a single bar carries a lot of weight in your reasoning, and a reason to distrust any method that depends on the internals of one bar.
Timeframe changes the answer, not just the resolution
The same instrument can be in a downtrend on the hourly chart, a range on the daily, and an uptrend on the weekly — all simultaneously, all correctly. These are not contradictions; they are statements about different populations of participants.
- Monthly / weeklyWho it describes: Institutions, long-horizon capitalUse it for: Stage, primary trend, multi-year levels, accumulation bases
- DailyWho it describes: Swing participants, most positional activityUse it for: Structure, entries, stops, pattern completion
- Hourly / 15-minWho it describes: Intraday and short-term flowUse it for: Execution timing only — never the thesis
A stock in mid-2022 style conditions — the readings conflict because the questions differ.
- Weekly
- above a rising 30-week MAprimary uptrend
- Daily
- 12-week sideways rangeno trend
- Hourly
- lower highs, lower lowsshort-term downtrend
Log versus linear, and why it matters more than people think
On a linear scale, a move from ₹100 to ₹200 occupies the same vertical distance as ₹1,000 to ₹1,100. On a log scale, equal vertical distance means equal percentage move — which is what actually matters to your capital.
- Use log scale for anything spanning years, or any instrument that has multiplied several times. Trendlines drawn on a linear multi-year chart are close to meaningless.
- Use linear for short windows and intraday, where the percentage distortion is negligible.
- A trendline that holds beautifully on log and breaks on linear is not ambiguity — it means you asked two different questions. The log one is the one about returns.
Adjusted prices: the quiet data trap
When a company splits 1:10, the raw price divides by ten overnight. An unadjusted chart shows a 90% crash that never happened, and every level, moving average and pattern that spans that date becomes garbage.
Indian stocks see frequent bonuses and splits, so always work on adjusted data — every serious data source provides it. If a chart shows an inexplicable vertical gap, check the corporate-action history before inventing a narrative for it.
The honest limits
- ✓Where positions were built, and at what prices they are under water
- ✓Whether supply or demand has had the initiative recently
- ✓Where the market has repeatedly found buyers or sellers
- ✓How volatile this instrument is, and therefore how far a stop must sit
- ✓Whether a move is being confirmed by participation or is thin
- ✕What the company will report next quarter
- ✕Anything about news that hasn't happened yet
- ✕Certainty about direction — every read is probabilistic
- ✕Anything useful on an illiquid stock where a handful of trades set the price
- ✕Whether a business is worth owning — that's the fundamentals course
Technical analysis is a discipline of conditional probabilities and position management, not prediction. Any presentation of it as forecasting is either naive or selling something.
See it live · Every stock page here renders the daily chart alongside the engine's structural read — stage, levels, base geometry — so the picture and the measurement sit together.
See live setups →In one screen
- 1A chart maps where participants are committed and where they become uncomfortable — that's the real basis of the discipline.
- 2The close is the most informative of the four OHLC values; bar internals are lost in compression.
- 3Higher timeframe sets direction, lower timeframe sets timing.
- 4Use log scale for multi-year charts and always work on split-adjusted data.
Check yourself
Nothing is scored or stored — the explanation after each answer is the point.
A stock is above a rising 30-week moving average on the weekly, but making lower highs on the hourly chart. What's the correct reading?
Why is the close weighted more heavily than the high or low?