Structure and Symmetry: A Framework for Thinking About FX

Most people come to the currency market looking for a forecast. They want to know where EURUSD will be in three months, whether the dollar is finished, what the next central bank meeting will deliver. It is a reasonable instinct and it produces reliably poor results, because it asks the market a question it is not obliged to answer.

After nine years of running a live FX and precious metals strategy, I have come to think the more useful question is a different one. Not "where is this going", but "what kind of market am I actually looking at, and does its structure currently offer anything worth risking capital on". That shift, from prediction to structure, is the foundation of how we think at Vadantia.

Markets have shape before they have direction

Price is the most visible thing in a market and among the least informative on its own. What carries information is the shape price makes: where it accelerates, where it stalls, where it returns, where it refuses to return. Those shapes are not random. They are produced by the same forces every time, because the participants producing them are working under constraints that do not change.

Currency markets are enormous, continuous and highly liquid, and they are populated by participants with genuinely different objectives. Corporates hedge because they must, not because they have a view. Central banks act on policy mandates. Reserve managers rebalance. Speculative capital chases and then unwinds. Leverage forces exits at inconvenient moments. Stops cluster in obvious places because most people place them in obvious places.

None of that is a market anomaly waiting to be arbitraged away. It is the permanent architecture of how the market functions. And architecture leaves structural signatures.

What symmetry actually means

Symmetry is a word that attracts mysticism, so let me be precise about how we use it. We do not mean that markets are geometrically perfect, or that there is a hidden numerical code beneath price. We mean two specific, observable and measurable things.

The first is proportionality. Moves within a structure tend to relate to one another in reasonably consistent ratios. Impulse and retracement, expansion and consolidation, the depth of a pullback relative to the leg that preceded it. These relationships are not exact and they are not laws. They are distributions, and distributions can be measured, tested and used.

The second is recurrence across scale. The structural grammar you see on a monthly chart appears again on the weekly, the daily and the four hour. A compression before a directional release looks recognisably similar whether it takes six months or six hours to build. This is not a mystical observation. It is consistent with five decades of empirical work on the self-affine character of financial price series, from Mandelbrot’s early scaling results through to modern volatility research. If price structure is approximately self-similar, structural recurrence across timeframes is the expected outcome rather than a coincidence.

Our timeframe stack is not four separate analyses that we then reconcile. It is one geometry read at four magnifications.

The practical consequence is important. Monthly and weekly for structure, daily and four hour for the active opportunity, one hour for execution timing. When those magnifications agree, the evidence is stronger. When they disagree, that disagreement is itself information, and it usually argues for patience.

States, not signals

The second half of the framework is that markets are not in one condition permanently. They move through states.

A market can be trending, ranging, compressing or expanding. It can also be disordered, which is our term for conditions where structure is genuinely unreadable: overlapping ranges, conflicting timeframes, no clear level at which we would know we were wrong. Each state behaves differently, rewards different behaviour and punishes different mistakes. The same pattern, in a trending market and in a disordered one, is not the same piece of evidence.

This is why we treat patterns as evidence rather than as instructions. A double top is not a sell. It is an observation that may or may not indicate that a prior directional state is losing integrity, and its weight depends entirely on the state it appears in, the liquidity around it, the volatility conditions at the time, and whether the higher timeframes agree.

The returns in this business are not made by predicting the next candle. They are made at the transitions: range to trend, compression to expansion, trend losing structural integrity. Those transitions are driven by behaviour that is durable rather than fashionable. Positioning unwinds. Liquidity gets sought out around well known levels. Herding and anchoring do what they have always done. These features are not going to be competed away, but capturing them requires a discipline that most participants find difficult to sustain.

The information in broken symmetry

Some of the most useful evidence in our process comes from structures that fail.

A market that should have continued and does not is telling you something. A level that gets swept and immediately reclaimed is telling you something different, and often more urgent. A pattern that completes and then reverses through its own invalidation point is not a failure of the framework. It is the framework working, because the invalidation was defined in advance and the market has just answered the question.

If you cannot state clearly what would prove you wrong, you do not have a trade. You have an opinion with money attached to it.

Risk is the only symmetry you control

Everything above concerns reading the market. The part you actually control is the shape of your own exposure.

Market structure is uncertain. Position structure does not have to be. Risk defined at entry, position size calibrated to prevailing volatility rather than fixed, exposure managed at portfolio level and not only trade by trade, and a genuine willingness to stand aside when conditions are disordered. Deciding not to trade is a decision, and on many days it is the correct one.

This is where symmetry becomes deliberately asymmetric. The point of defining risk precisely is to build a payoff profile in which losses are bounded by design and gains are allowed to develop. You do not need a high strike rate to run that profile profitably. You need consistency in how the profile is applied, which is a behavioural problem far more than an analytical one.

Where the technology fits

We describe our process as machine learning enhanced discretionary. The order of those words matters.

Statistical and machine learning tools help us classify market states, measure how patterns have behaved historically in comparable conditions, assess liquidity and volatility context, and check whether our own conviction is properly calibrated. They make a discretionary process more structured, more repeatable and more measurable. Every model candidate, accepted and declined, is logged and timestamped, which means the process can be reviewed rather than merely described.

What the technology does not do is remove the human from the decision. It supports, tests and occasionally contradicts a judgement built over a career of watching these specific markets. Nine years of logged decisions is a proprietary conditioning set. The mathematics that reads it is available to anyone. The record it reads is not.

Structure and symmetry are not a forecasting method. They are a way of organising attention: knowing what state a market is in, recognising when its structure is genuinely offering something, understanding what would prove the idea wrong, and having the discipline to wait for the days when state, probability, tradeability and risk actually line up.

Most of the time, they do not. Recognising that is not a limitation of the framework. It is most of the value in it.

Vadantia Quant FX is a liquid alternative strategy focused on FX majors and minors, gold and silver, live since 2017. The strategy is implemented through separately managed accounts, with a fund structure in development. Performance reporting is produced independently by Fundpeak, net of fees and transaction costs, on a non-compounded basis.

FOR PROFESSIONAL AND INSTITUTIONAL INVESTORS AND ELIGIBLE COUNTERPARTIES ONLY

This material is for information purposes only and does not constitute legal, tax, investment, financial or other advice, a personal recommendation, an offer or solicitation, or a financial promotion. It is not a signal or copy trading service. Capital is at risk and leveraged products may lose more than the amount invested. Past or historical performance is not a reliable indicator of future results. Vadantia FZCO acts as a strategy provider and investment adviser to a regulated asset manager and does not guarantee any return.