What research is for
Research is a filter, not a forecast
The purpose of research is not to predict where a currency will be next month. Nobody does that reliably. The purpose is to reduce the number of trades you take from “every chart that looks interesting” to “the small set where I can articulate why, what would prove me wrong, and how much that costs.”
Framed that way, good research produces fewer ideas, not more. If your process generates a conviction trade on every instrument every morning, it is not filtering. It is manufacturing justifications for the trading you were going to do anyway.
This page describes a repeatable process: where the information comes from, how to read the two documents that matter most (the calendar and the central-bank statement), how to check what else in the market agrees with you, and how to structure an idea so it can be judged rather than merely felt.
Three properties of a usable research process
- Repeatable. The same steps in the same order, whether you are up or down on the week. Consistency is what makes the results interpretable later.
- Falsifiable. Every conclusion comes with the specific observation that would overturn it. A view with no invalidation is a belief, not a thesis.
- Bounded. It ends. A process that keeps going until you find a reason to trade will always find one.
Stage one
What data actually drives markets
There is an unlimited supply of market commentary and a fairly short list of things that genuinely move prices. Learning to distinguish the two is most of the discipline.
Primary
Official statistics
Inflation, employment, growth and trade figures published by national statistical agencies on a published schedule. These are the raw inputs. Go to the original release where you can, a headline summary frequently omits the revision or the component that explains the market’s reaction.
Primary
Central bank communication
Rate decisions, statements, minutes, projections and speeches. The single most direct influence on a currency, because policy determines the interest-rate differential that underpins its relative value.
Primary
Market prices themselves
Bond yields, equity indices, commodities and volatility measures are continuously updated aggregations of what everyone believes. They are frequently a better guide to consensus than any written commentary, and they lead rather than follow.
Secondary
Consensus forecasts
The expected figure for each upcoming release. Essential context, because markets move on the gap between actual and expected. A strong number below expectations still sells the currency.
Secondary
Positioning data
Aggregate measures of how speculative participants are leaning. Published with a lag and best used to judge how crowded a trade already is rather than as a timing signal.
Treat with care
Opinion and commentary
Analyst notes, financial media and social feeds. Occasionally useful for surfacing what you have missed, but they are interpretations rather than data, are frequently written after the move to explain it, and are the easiest way to substitute someone else’s conviction for your own process.
Stage two
How to read an economic calendar
The calendar is the most useful research document in currency trading and the most commonly misread. Each entry has four fields that matter, and beginners typically look at only one of them.
- Time. Convert to your own zone once and work from that. Note that publication times shift with regional daylight-saving changes.
- Previous. Last period’s figure, and critically, whether it has been revised. A large revision changes the story even when the new headline looks fine.
- Forecast. The consensus expectation. This is what is already in the price.
- Actual. The released number. It matters only relative to forecast.
The four rules
- Trade the surprise, not the level. An inflation print of 3.1% is neither good nor bad in isolation. Against a 2.8% forecast it is hawkish; against a 3.4% forecast it is dovish. The same number, opposite implications.
- Read the revision. A jobs number beating forecast by 30,000 alongside a 60,000 downward revision to the previous month is a net deterioration, and markets usually treat it that way after the initial spike.
- Look past the headline. Core inflation excluding food and energy often matters more than headline CPI. Wage growth often matters more than the jobs count. The internals frequently drive the sustained move once the headline reaction fades.
- Ask what it does to the policy path. A release only matters to a currency insofar as it changes what the central bank is likely to do. Data that is surprising but policy-irrelevant produces a spike and a fade.
Practical calendar discipline
- Check the coming 24 hours before opening a chart, not after finding a setup you like.
- Mark high-impact events for both currencies in any pair you are considering.
- Decide in advance whether you trade through an event, reduce into it, or stand aside, and write it down.
- Expect wider spreads and slippage through releases; a stop is an instruction, not a guaranteed price.
An economic calendar is in development for the NOVEDOX terminal, see tools for current status.
Reference
Release types and typical market impact
A rough triage of what deserves your attention. Impact ratings describe typical historical behaviour, not any particular release.
| Release type | Typical impact | Frequency | What to look at |
|---|---|---|---|
| Central bank rate decision | Very high | Every 4–8 weeks | The forward guidance and vote split usually matter more than the decision, which is often anticipated. The press conference frequently moves markets more than the announcement. |
| Inflation (CPI) | Very high | Monthly | Core versus headline, and the month-on-month rate as well as the annual figure. Determines how constrained the central bank is. |
| US non-farm payrolls | Very high | Monthly | Headline jobs, unemployment rate, average hourly earnings and revisions to the prior two months. All released together, sometimes pointing in different directions. |
| Central bank minutes & speeches | High | Ongoing | Changes in language, the range of views expressed, and any hint that the committee’s balance is shifting. |
| GDP (preliminary) | High | Quarterly | The first estimate, which moves markets far more than later revisions. Check the composition, consumption versus inventories. |
| PMI surveys | Moderate to high | Monthly | The 50 threshold separating expansion from contraction, and whether services or manufacturing is driving the change. |
| Retail sales | Moderate to high | Monthly | The core measure excluding autos, which strips out a volatile component and gives a cleaner read on consumption. |
| Employment claims | Moderate | Weekly | The four-week moving average rather than any single week, which is noisy. Useful as an early warning between monthly reports. |
| Producer prices (PPI) | Moderate | Monthly | Treated as a leading signal for consumer inflation. Can shift expectations for the CPI release that follows. |
| Consumer confidence | Low to moderate | Monthly | Directional trend over several months rather than a single reading. A leading indicator of future spending. |
| Trade balance | Low to moderate | Monthly | Matters most for export-driven and commodity economies. Persistent trends matter more than any single month. |
| Housing data | Low | Monthly | Rate-sensitive and slow-moving. Rarely a currency driver on its own but contributes to the growth picture. |
| Unscheduled events | Unpredictable | Irregular | Geopolitical developments, interventions and political shocks. Cannot be planned for, which is the argument for position sizing that survives surprises. |
Stage two, continued
Interpreting central-bank language
Central banks communicate deliberately and conservatively. Statements change slowly by design, which is exactly why small changes carry so much information: a phrase that was in the last statement and is missing from this one was removed on purpose.
Read the diff, not the document
The most productive way to read a policy statement is against the previous one, looking specifically at what was added, removed or softened. Analysts do this systematically, and the market reprices on the differences rather than on the overall tone.
The vocabulary
- Hawkish, leaning toward tighter policy: concerned about inflation, inclined to raise rates or hold them higher for longer. Typically supportive of the currency.
- Dovish, leaning toward looser policy: concerned about growth or employment, inclined to cut. Typically negative for the currency.
- Data-dependent, explicitly declining to commit to a path, which raises the market impact of each subsequent data release.
- Forward guidance, explicit signalling about the likely future path, intended to shape expectations without acting immediately.
- Neutral rate. The policy rate thought to be neither stimulating nor restraining the economy. Where officials place it tells you how restrictive they think current policy is.
What to look for
- Removed qualifiers. Dropping a word like “gradual” or “patient” is a signal in itself.
- The vote split. A unanimous hold and a narrowly divided hold are very different pieces of information about the next meeting.
- Projections. Where published, the committee’s own forecasts for growth, inflation and rates are the clearest statement of intent available.
- The press conference. Unscripted answers regularly move markets more than the prepared statement, and often in the opposite direction to the initial reaction.
- Dissent and speeches. Individual officials speaking between meetings map the range of views on the committee.
Cross-asset checks
Corroborating a view
A currency view that nothing else in the market supports is usually wrong or early. Before committing, check whether related instruments tell the same story.
- Short-dated bond yields. The 2-year yield tracks policy expectations closely and often moves before the exchange rate does.
- Equity indices. A broad sell-off usually means risk-off currency behaviour, which can override your fundamental story entirely.
- Commodities. Oil for CAD, industrial metals for AUD. If your commodity-currency view contradicts the underlying commodity, examine why.
- Related pairs. If you are bullish the euro, EUR/GBP and EUR/JPY should show something consistent, not just EUR/USD.
- Volatility. Rising volatility argues for smaller size and wider stops regardless of how good the idea looks.
Stage four
Evaluating a trade idea
Four components. An idea missing any one of them is not yet a trade, however convincing the chart looks.
Thesis
One sentence stating what you believe and why. “The policy gap is widening in favour of currency A because its central bank has signalled it will hold while the other has signalled cuts.”
If the sentence needs three clauses and a chart to make sense, the idea is not clear enough to size.
Catalyst
What causes the market to agree with you, and roughly when? A scheduled release, a policy meeting, a technical level giving way.
Without a catalyst you may be right and still bleed cost and attention for months while nothing happens.
Invalidation
The specific price or event that proves you wrong. A broken level, a data release contradicting the thesis, a policy shift.
This defines the stop, and therefore the size. Deciding it in advance is what stops a losing trade from becoming an argument with the market.
Risk-reward
The distance to a realistic target against the distance to invalidation. Many traders require at least 2:1 before taking an idea at all.
The target must be a level the market plausibly reaches, not the number that makes the ratio acceptable.
The research checklist
Run through this before committing capital. It takes a few minutes and eliminates most of the trades you would later regret.
- Can I state the thesis in one sentence without using the words “looks like” or “feels”?
- What is the catalyst, and when is it expected?
- What exactly would prove me wrong, at what price or on what event?
- How much money do I lose if that happens, in cash and as a percentage of equity?
- Is the reward at least twice the risk to a target the market can realistically reach?
- Does anything else in the market corroborate this, yields, related pairs, the commodity, the risk tape?
- What is the strongest argument against it, and why am I not persuaded?
- Is this idea already crowded? Has the move largely happened before I arrived?
- Does this duplicate risk I already hold in another open position or currency?
- Is there a scheduled release before the target is plausibly reached, and have I decided how to handle it?
- Have I calculated the position size from the stop distance rather than estimating it?
- Would I take this trade if my last three had been losers, or if they had been winners?
Size it with the position size calculator once every answer is satisfactory.
After the trade
Reviewing honestly
Research does not end at execution. The review is where the process improves, and it has one important rule: judge the decision, not the outcome. A well-researched trade that lost was probably still a good decision; an unresearched trade that won was still a bad one, and will cost you eventually.
- Record the thesis and invalidation as written before entry, not as remembered afterwards.
- Note whether you followed the plan, separately from whether the trade made money.
- Tag each trade by setup type so you can review by category over a meaningful sample.
- Record your state of mind, rushed, bored, recovering a loss. The pattern usually shows up before the losses do.
- Review monthly, not daily. Small samples produce noise, and reacting to noise is how a working process gets abandoned.
The NOVEDOX terminal keeps a complete trade history and account ledger, which gives you the factual record to review against your written notes.
Journalling in the academyCommon questions
About research
Does NOVEDOX publish its own research reports?
We publish educational material about research method, which is what this page is. We do not publish market calls, price targets, forecasts or recommendations, and we do not employ analysts issuing views on instruments. If a page on this site ever tells you what a currency will do, treat it as a mistake.
How much time should research take each day?
Less than most people assume, once the process is settled. Twenty to thirty minutes before your session covers the calendar, the macro backdrop and the levels that matter, with a longer weekly review of the policy picture and your own journal.
Spending four hours a day reading commentary is usually a substitute for having a process rather than evidence of one.
Where should I get economic data from?
Prefer primary sources where practical: national statistical agencies for economic releases and central-bank websites for policy statements, minutes and projections. Both are free and published on a schedule.
A reputable economic calendar is a convenient aggregator for timings and consensus forecasts, but check the original release when a number matters to a position you hold.
Can I rely on analyst forecasts?
Use them as information about consensus rather than as predictions. The value of knowing the forecast is that it tells you what is already priced in, which is precisely what determines the market’s reaction when the actual figure arrives. Forecasts are frequently wrong; the consensus they represent is still useful.
What if my research is right but the trade loses?
That will happen regularly and is not evidence of a broken process. Markets are probabilistic: a sound thesis with a genuine edge still loses a substantial proportion of the time, and unscheduled events override good analysis without warning.
This is why position sizing matters more than accuracy. Judge the process over a meaningful sample of trades, not over the last one.
Is technical analysis part of research?
Yes. It supplies the invalidation level and the entry, which are two of the four components of a tradable idea. Fundamental research tells you what should happen; technical work tells you where you are wrong and what it costs. The methods are covered on market analysis.
Test the process before you fund an account.
Open a NOVEDOX demo with $10,000 in simulated funds and run your research on live streaming prices, no deposit required.
Trading involves risk. 74.3% of retail accounts lose money.