Forex analysis: levels, sessions, macro

This lesson shows how to analyse one currency pair from D1 context to a concrete entry: how to choose levels, how to check their reliability, and how to filter out days when trading is not worth it. It is designed for someone who already knows pairs, pips, and lots, trades one or two pairs for now, and wants to stop entering every pretty chart.

How to analyse a currency pair — three layers

Analysis works like a sieve. Each layer can reject a setup; none approves it by itself.

The order is part of the method, not a preference: context filter belongs to the setup definition, not to optional add-ons. Jumping straight to M1 without context is the fastest way to overtrading, because on a low timeframe there is always a move that looks like a signal. The same chart-reading method is covered in the lesson What is a chart; on forex, add the 24/5 session, CFD cost, and the fact that price is your broker’s quote.

Which levels and timeframes matter

A level that is supposed to work must be visible to many market participants. Therefore, you choose places that most people watch, not lines drawn to fit your thesis.

The practical rule is: a level is a zone, not a line. Instead of 1.0850, write 1.0840–1.0862 and treat the boundaries as a decision area. The width of the zone comes from the pair’s volatility on your timeframe — on a calm major, it is usually a few pips; on an exotic, significantly more. Example values are illustrative, and you check quoting conditions and spread with your broker. Recognising such zones is covered in the lesson Support and resistance, and definitions are in the glossary entries support and resistance and trend.

Protocol: how to check level reliability

Instead of judging a level by eye, count its behaviour. The procedure below takes a few minutes on a chart:

1. Mark a zone, not a line: set the lower and upper boundaries from candle bodies where reactions occurred.
2. Count touches on D1 and H4. Minimum is three reactions — an illustrative threshold that works better on a calm major than on an exotic.
3. Assess reaction quality: the close of the candle after the bounce matters, not the wick alone.
4. Measure reaction depth in volatility units, for example as a multiple of ATR on H1. A 3-pip bounce means nothing.
5. Record what happened after the false breakout. A return to the zone strengthens it; acceptance beyond it weakens it.
6. Check the distance to the nearest opposite zone. If it is smaller than the planned profit, the setup does not make sense regardless of level quality.

A level that passes steps 1–4 goes into the observation sheet. A level without three reactions is not wrong — it is simply untested, so it does not belong in the plan.

Macro filter: calendar, rates, and risk-on/risk-off

Macro does not predict the outcome of a publication. It answers two questions: whether to trade at all in a given window and whether the setup direction has support in the environment.

Interest rates and expectations affect the currency over weeks and months, not a single candle — so the central bank cycle is context, not a trigger. Daily publications work the opposite way: they do not change the trend, but they can cut a stop in seconds. How to read each is covered in the lessons on central banks and calendar in practice.

Risk-on and risk-off is a thinking shortcut, not a forecast. In one week, the dollar is a safe haven; in another, it is a high-yielding currency. Treat this classification as a direction filter checked once per session, not as an entry signal. The trading session with the tightest spread is chosen according to the lesson on sessions and liquidity.

Pair correlation — how much risk you really have

Two positions on pairs linked to the same factor are often one decision in two packages. If EUR/USD and GBP/USD move largely together, a long on both pairs means in practice one weak-dollar exposure — with double spread cost and a single reason for loss.

Instead of guessing, write down the correlation and calculate exposure together:

total exposure (R) ≈ R1 + R2 x correlation

Illustrative example: two positions of 1R each, positive correlation around +0.8, so total exposure is approximately 1.8R, not 2R. With correlation near zero, you count both risks separately. Values are illustrative because correlation changes over time — its current level is read from your own analytical tool or calculated yourself from prices, never from memory.

How correlations add up across the whole portfolio is covered in the lesson Portfolio risk and correlations, and the concept itself is explained in the correlation glossary entry.

When analysis says do not trade

Four conditions under which the best decision is no trade:

1. D1 has no clear structure — price stuck between zones, with neither a trend nor a range with clear boundaries. Without context, every H1 setup is a guess.
2. A high-impact publication falls in your session window, and the plan does not contain a rule for that situation. Lack of a rule is not neutral — it means a decision made under emotional influence.
3. Spread exceeds the limit written in the plan. The threshold spread above 20% of stop distance is illustrative; the correct value is set from your own measurements in your session window and confirmed with your broker’s quoting conditions.
4. You already have an open position strongly correlated with the setup you are considering. If total exposure exceeds the limit, the new position is not a new opportunity — it is an increase of the same bet.

Common mistakes in forex analysis

• Analysis exclusively on a low timeframe without checking D1 context — the simplest way to enter against the dominant move.
• Adding indicators after every loss instead of improving execution of the existing setup. RSI and MACD can be a filter, but they do not replace structure.
• Ignoring correlation when adding positions — three tickets in the same macro direction are not three independent opportunities, but one larger exposure.
• Opposing fundamental analysis to technical analysis instead of treating them as layers answering different questions: which direction and at what moment.
• Analysis without notes — a level assessed on Monday and reproduced from memory on Friday is two different levels.

Exercise: five sessions with an analysis journal

Choose one major pair and for five sessions record before London opens, in Polish time: D1 bias, presence of high-impact publications, allowed hourly window, and two levels chosen with the protocol from this lesson. After the session, add how many setups aligned with the bias, how many you took, and whether a second pair with high correlation was open at the same time.

The goal is measurable: five filled rows, each defensible without referring to feeling. If after five days you cannot say which filter rejected the most setups, the sheet was kept too broadly.

Check yourself

Common questions

How many levels should I have on the chart?

As many as fit in the plan of one session — usually two or three zones: the nearest resistance, the nearest support, and a structural level from a higher timeframe. Every additional level increases the number of reasons why it is worth entering.

Do I need to know macro to trade forex?

You do not have to forecast interest rates. You do need to know when publications fall and to have a written rule: I trade before them, after them, or not at all. That is one decision per session, not daily analysis.

How do I check whether a level is reliable?

Count the price reactions in the zone: at least three touches with a candle close after the bounce, plus a known outcome of a false breakout. If you cannot give those numbers, the level has not been tested yet.

Why does this lesson measure correlation and level reliability instead of judging them by eye?

Because visual judgement changes with the last trade — after a losing streak the same chart looks different. The number of reactions, the measured correlation, and the recorded outcome after a breakout give reference points that can be compared between weeks and settled in the journal. Without those numbers you cannot tell process improvement from plain luck.