A day can be up while the hour is breaking.

A daily candle compresses a whole news cycle into four numbers. That is useful. It is also how a market that spent six hours at 40% and six hours at 70% becomes a moderate body and a wide range, and then becomes, in memory, “a 55% kind of day.”

The hourly series of that same day is two different markets stacked on top of each other. If you only trade the daily close you will be right about the destination of the day and wrong about every entry you took inside it. If you only trade the five-minute you will be right about the last argument and wrong about the week.

Crypto traders already know this as multi-timeframe analysis. Prediction markets need it more, not less, because the destination is binary and the path is where all the money is. OHLC on a chance is the single-interval version. This page is what happens when you put ten intervals on one row.

Predm’s markets table carries three columns of dots across 10 timeframes: 1M · 1W · 1D · 8h · 4h · 2h · 1h · 30m · 15m · 5m. The columns are trend, trend projection and timing. The words are load-bearing and they are the only words this page will use. Nothing here names how a dot is computed. The marketing site does not publish the method, and a guide that leaked it to rank for a keyword would be a worse page.

Trend is direction on that timeframe: up, down, or not enough history to say. A market can be up on the day and down on the hour. That is not a bug in the column. That is the information.

Trend projection is whether that direction is gaining or fading. An uptrend that is fading is a different row from an uptrend that is still building. Projection is never shown where trend itself is missing: we do not project a direction we do not have.

Timing is the character of the interval, calm to chaotic. A calm uptrend and a chaotic uptrend can print the same direction and be opposite trades. Timing is a number when there is enough history, and blank when there is not.

The homepage decoder is the visual version of these three sentences. This page is the reading guide for when you are on the board itself.

Alignment is the boring, valuable case. Trend agrees across the higher timeframes and the lower ones are not fighting it. The market has a direction that has survived more than one compression. That does not mean it continues. It means you are not inventing a story from a single interval.

Divergence is the usual case, and the useful one. Daily up, four-hour fading, hour breaking: the higher-timeframe estimate is still intact and the lower-timeframe book is doing something else. That something else is either a pullback inside the day or the start of the day being wrong. You cannot know which from one row. You can know that the daily close is no longer a settled fact.

Beginners treat divergence as a signal to do something clever. It is usually a signal to do less. A market that does not agree with itself is a market whose last price is a poor summary. That is the same lesson as the case against the line, applied across time instead of inside one candle.

Read left to right: long timeframes first. The month and the week tell you the weather. The day tells you today’s argument. The hours tell you whether that argument is still on. The minutes tell you whether you are late.

Then read the three columns as a sentence, not as a dashboard. “Up, fading, chaotic” is a sentence. “Up, gaining, calm” is a different sentence. A row of twelve green dots is not a sentence. It is a colour, and colour is not a method.

Skip rows that are mostly blank. Blank means insufficient history. A market that listed yesterday cannot have a monthly trend. Drawing one would be the same offence as drawing a candle on a silent book.

Finally, open the chart. The board is a filter, not a verdict. The candles are where you see whether the dots are describing a real range or a thin one. Technical analysis for prediction markets is what to do once you are on that chart.

Thin markets produce theatrical lower-timeframe dots. One fill can flip an hourly trend. Believe the higher timeframes first, or do not trade the market at all.

New markets have no right side of the board. That is correct. A product that filled those cells with a guess would look more complete and be more wrong. Insufficient is a reading.

Resolving markets start to agree across timeframes because the destination is eating the path. Alignment near 97% the night before settlement is not a brilliant multi-timeframe read. It is a book that has already decided. The interesting rows at that point are the ones that still disagree — a daily settled and an hourly that will not die is a book that still has an argument in it.

Not a black-box signal service. The dots are readings, labelled in words, on a board you can sort. They are not entries, and they are not sized for you.

Not a published formula. This page will not name the calculation. If a screenshot on X claims to have reverse-engineered it, that is their claim.

Not live on every venue. The board Predm ships today reads Polymarket and Hyperliquid. The Hyperliquid rows are perps, so “resolving” does not apply to them; the alignment-versus-divergence reading still does.

What timeframes does Predm show?

1M · 1W · 1D · 8h · 4h · 2h · 1h · 30m · 15m · 5m. Long to short, left to right — the same order as the terminal. A market that is too new or too quiet will be blank on the left, the right, or both.

Should I trade only when every timeframe agrees?

Alignment is a filter, not a trigger. Full agreement near resolution is often just the destination. Full agreement three months out on a liquid event is closer to a real consensus. Still open the chart.

Why is a cell blank?

Not enough completed candles on that timeframe to say anything that is not a guess. Blank is the honest output. A grey guess would look nicer and be a worse product.

Is this the same as classical multi-timeframe analysis?

Same idea: do not let one compression of the data be the whole story. Different wrapper: three labelled readings per timeframe, on a board you can scan, instead of twelve chart layouts tiled on a desktop.