FX Value (Purchasing Power Reversion)

Thesis (edge)

The basic intuition is old and simple. If a basket of goods costs far more in one country than another once you convert at the market exchange rate, then that currency is expensive. Over long horizons, trade flows and capital flows tend to push the exchange rate back toward the level that makes prices roughly comparable. Buy the currencies that are cheap on this measure, sell the ones that are expensive, and collect the drift back to normal.

That is the theory. The honest version of the story is far more humble. Exchange rates deviate from purchasing power parity for years, and the reversion, when it comes, is slow and unreliable. Academic estimates of how long it takes for half of a misvaluation to disappear cluster around three to five years. That is an eternity in trading terms.

So why bother? Because the signal is genuinely different from everything else in the FX toolkit. Carry and momentum both tend to pile into the same strong, high yielding currencies. Value pushes in the opposite direction. It is the signal that tells you to stop buying the currency that everyone loves, right at the point where it has become expensive. Used alone it is frustrating. Used as a counterweight it earns its place.

Where it works (regimes)

Value pays off after a long trend exhausts. The years following an extended dollar bull market, when the dollar is stretched against nearly everything, are the classic environment. The reversal from the 1985 dollar peak and the unwind after the 2001 dollar peak are the textbook cases, and in both the value signal had been screaming for years before it was proven right.

It works badly during the trend itself. While a currency is grinding richer, value tells you to short it, and you lose steadily. This is not a bug in the implementation. It is the fundamental nature of the strategy. You are betting against the current direction, which means you are usually early and often painfully so.

It also fails when a cheapness is not a mispricing at all but a permanent change. A country that suffers a lasting productivity collapse, or loses a key export market, should have a permanently weaker currency. A model anchored to a historical average will keep insisting it is cheap, and it will keep being wrong.

Signals

• Real exchange rate: take the nominal exchange rate and adjust it for the difference in inflation between the two countries. If a country has had 30 percent more inflation than the United States over a decade but its currency has not fallen, it has become expensive in real terms.
• The anchor: rather than trusting a theoretical fair value, compare today's real exchange rate to its own long-run average, typically over five years or more. This asks a more modest and more answerable question: is this currency unusually expensive relative to its own history?
• Misvaluation score: express the gap in standard deviations so that currencies with different typical ranges can be compared on the same scale.
• Terms of trade adjustment: a commodity exporter whose main export has genuinely tripled in price deserves a stronger currency. Without this adjustment, the model will short exactly the currencies with the best fundamental story. This single correction removes a large share of the strategy's worst trades.

Portfolio construction

Rank the universe by cheapness. Go long the cheapest group, short the richest, weight by inverse volatility, and cap single-currency exposure.

Rebalance quarterly, or even semi-annually. The signal moves at the pace of inflation differentials and multi-year trends, which is to say barely at all from one month to the next. Rebalancing monthly means paying spread repeatedly for a signal that has not changed.

The most important construction decision is not the weighting scheme. It is the entry rule. Because value is chronically early, many practitioners require that a cheap currency has at least stopped falling before buying it. That single filter, combining value with a slow trend confirmation, historically improved the timing considerably without destroying the underlying logic.

Risk model

The dominant risk is time. A value position can be underwater for three or four years and still be correct. Most capital, and most patience, does not last that long. That is arguably why the premium persists at all: it is compensation for enduring a very uncomfortable wait.

The second risk is structural break. Purchasing power reversion assumes the long-run average is meaningful. If the economy underneath has permanently changed, the anchor is fiction and the trade is simply a losing directional bet with an academic justification attached.

The third risk is that value positions can be short high yielding currencies, which means you are paying carry while you wait. You lose a little every month for being right eventually. Track the carry cost of the book explicitly, because a value portfolio that bleeds 3 percent a year in negative carry needs a very large reversion just to break even.

Costs & implementation

Trading costs are low, because this is G10 FX and turnover is minimal. This is one of the cheapest strategies to run in the entire systematic universe.

The real implementation difficulty is data quality and definition. Inflation indices are revised, they measure different baskets across countries, and they do not capture traded versus non-traded goods cleanly. Two reasonable analysts can build two real exchange rate series that disagree about which currency is cheap. Do the work on the data, and be suspicious of a signal that flips when you make a small methodological change.

Failure modes

• Confusing cheap with good. A currency is usually cheap because the country has a problem, and sometimes the problem is real and permanent.
• Ignoring the negative carry. Financing a value position can quietly cost more than the eventual reversion pays.
• Rebalancing too frequently and paying costs on a signal that has not moved.
• Anchoring to a fixed theoretical PPP level rather than a rolling average, which makes the model insist a currency has been mispriced for twenty years.
• Abandoning the strategy after three bad years, which is exactly the point at which the signal is at its most stretched and most likely to pay.
• Backtesting on a period that happens to contain one giant reversion and attributing the whole result to the model.

Our Notes & Suggestions

Do not run this on its own unless you have a very long horizon and a very forgiving source of capital. Its natural home is inside a multi-signal FX book, where it sits beside carry and momentum and quietly stops them from becoming a single crowded bet on whichever currency is currently fashionable.

A practical setup: use value to set the strategic tilt, use momentum to time the entry, and use carry to size. When all three agree, take a full position. When value disagrees with the other two, take a smaller one. When value is the only signal firing, be patient, because you are early by definition.

Above all, be realistic about the horizon. If your reporting cycle is quarterly and your patience is annual, a signal that reverts over four years will make you look wrong for a long time before it makes you look right. That mismatch, not the model, is what usually kills this trade in practice.