So Daniel sent us this one — and it's the kind of question that sounds simple but pulls a thread on basically everything most people assume about currency markets. He's asking about the Euro-Dollar rate, one of the most liquid pairs in the world, and what actually drives the price you see on your screen. His mental image, and I think most people's mental image, is that exchange rates are driven by trade flows, geopolitics, supply and demand from companies buying and selling goods across borders. But the reality is that institutional speculation absolutely dwarfs all of that. So he wants to know: to what extent does institutional capital actually move major currency pairs, how much do retail transactions even matter, who's speculating and on what time frames, and in a world of AI and algorithmic trading, is there any edge left for anyone who isn't a giant fund?
And the short answer, before we even get into the numbers, is that if you're looking at the Euro-Dollar rate and thinking it reflects trade between the US and Europe, you're looking at maybe three to five percent of what's actually moving the price. The other ninety-five percent is financial flows — speculation, hedging, portfolio rebalancing. It's not even close.
Three to five percent. That's the trade part.
That's the trade part. And that's being generous. The Bank for International Settlements runs something called the Triennial Survey, which is the authoritative census of global foreign exchange markets. The most recent one we have is from April twenty twenty-two, and it put total daily FX turnover at seven point five trillion dollars. Let that sit for a second — seven point five trillion, every single day. Of that, spot transactions, which is what most people picture when they think of currency trading, accounted for about two point one trillion. The rest is forwards, swaps, options — the derivatives market.
So the spot market, the thing people actually picture, is less than a third of the total.
Right. And even within that two point one trillion in spot, trade-related flows — a German car company paying for parts from a US supplier, a French wine exporter converting dollar revenue back to euros — all of that combined is estimated at under five percent of total volume. Some estimates put it closer to two or three percent.
Which means the price of the euro in dollars, the number that determines what your vacation costs and what companies pay for imports, is almost entirely set by people who will never take delivery of a single euro.
And the BIS breaks market participants into three categories. You've got reporting dealers — these are the big banks, the market makers. You've got other financial institutions, which is a catch-all that includes pension funds, hedge funds, insurance companies, mutual funds, and increasingly high-frequency trading firms. And then you've got non-financial customers — that's corporations, governments, the actual end users of currency for trade and investment.
And that middle category is the one that's exploded.
It's the story of the last decade. Other financial institutions now account for roughly fifty-five percent of total FX turnover, up from about forty percent ten years ago. These are institutional investors using currencies not as a means of payment but as an asset class. They're taking directional bets, they're hedging portfolio risk, they're running carry trades. And the inter-dealer market — banks trading with each other — has actually declined as a share of total volume. So the market has shifted from being bank-intermediated in the traditional sense to being driven by institutional investors who happen to execute through banks.
So walk me through the mechanism. A pension fund in Norway decides it wants exposure to the dollar. What actually happens to the price?
This is where it gets interesting, because it's not as simple as the pension fund placing a bet and the market moving. The pension fund calls its prime broker, which is typically a major bank — think JP Morgan, Deutsche Bank, Goldman Sachs. They say, we want to buy five hundred million dollars against the euro. The bank doesn't just go find someone who wants to sell five hundred million dollars. That rarely exists as a single matching order. Instead, the bank takes the other side of the trade itself — it warehouses the risk. It's now short dollars and long euros on its own book.
Which it doesn't want to be.
Which it absolutely does not want to be. Banks are not in the business of taking directional currency bets. So the bank then hedges dynamically. It starts selling euros and buying dollars in smaller chunks, through multiple venues, over some period of time, to neutralize its exposure. That hedging activity is what actually moves the price. The original five hundred million dollar order created maybe one and a half or two billion dollars in total market activity by the time the bank is done hedging. It's a multiplier effect.
So the pension fund's trade gets amplified through the banking system.
Amplified and also informationally transmitted. Other banks and trading firms see the order flow, they see the price starting to tick up, and they infer that someone big is buying dollars. They might front-run it, or they might pile on because they think there's momentum. This is what people in the market call flow toxicity — the idea that large orders contain information, and that information leaks into the market through the execution process. It's why large trades are almost never executed as a single block anymore. They're sliced up algorithmically to disguise the intent.
So the price you see on your phone is basically the exhaust of giant institutions trying to hide from each other.
That's a remarkably good way to put it. And then there's the portfolio rebalancing channel, which Daniel alluded to indirectly. This isn't even speculation in the traditional sense. A Japanese pension fund that holds forty percent of its assets in foreign equities — when those equities go up in value, the fund becomes overweight foreign currency exposure. To rebalance, it has to sell foreign currency and buy yen. That's not a bet on the yen. It's mechanical. But it's massive. We're talking about trillions of dollars in global pension assets that get rebalanced monthly or quarterly. These flows are one of the biggest drivers of daily price action, and they have nothing to do with anyone's view on where the euro should trade.
So you've got pension funds rebalancing, hedge funds taking directional bets, banks hedging the inventory they took on from both of those groups, and high-frequency firms trying to sniff out who's doing what. And somewhere in the background, a German car company is actually trying to pay its American supplier.
And the German car company's trade, in most cases, doesn't even hit the open market. Their corporate bank internalizes it. The bank nets it against other corporate client flows and only hedges the residual. So the trade component is not just small — it's often invisible to the broader market.
Which brings us to retail. Daniel asked about the individual or small money manager with short-term speculative positions. Where does that person fit into this seven point five trillion dollar ecosystem?
They're a rounding error. I don't say that to be dismissive — I say it because the numbers demand it. Retail FX trading, through brokers like OANDA or IG or Forex dot com, is estimated at three to five percent of global spot volume. And spot is less than thirty percent of total volume. So retail is maybe one to one and a half percent of the overall market. And even that overstates their influence on price discovery.
Why?
Because most retail brokers internalize their order flow. When a retail trader buys a hundred thousand euros against the dollar, the broker doesn't necessarily go into the interbank market and execute that trade. The broker takes the other side. They're running what's effectively a book-making operation. They know that statistically, most retail traders lose money — the data from European regulators shows that seventy to eighty percent of retail FX accounts lose money. So the broker's business model is to be the counterparty to losing trades and hedge only the net exposure that represents actual risk to them.
So the retail trader isn't even participating in the same market. They're playing a game against their broker, who has better data, better execution, and a structural edge.
And the broker, when it does hedge, is doing so in the institutional market at institutional prices. The retail trader sees a spread that's been widened, execution that's been delayed by milliseconds, and a price that may or may not reflect what's actually happening in the interbank market. It's not a level playing field — it's not even the same sport.
That's a grim picture for the small trader.
It is, and I think it's important to be honest about it. The quintessential image Daniel mentioned — the individual staring at five-minute charts, trying to scalp a few pips — that person exists, and there are millions of them. But they are noise in the price discovery process. The dominant speculative time horizon in FX is actually medium-term — days to months. Asset managers and hedge funds running macro strategies or carry trades are not looking at five-minute charts. They're looking at interest rate differentials, central bank policy trajectories, and capital flow data.
Let's talk about the carry trade, because that's a perfect example of what you're describing. Daniel mentioned it indirectly, and it's one of the most important speculative forces in the currency market.
The carry trade is beautifully simple in concept. You borrow in a currency with low interest rates, convert it to a currency with high interest rates, and pocket the difference. For years, the classic version was borrowing in Japanese yen, where rates were near zero, and buying Australian dollars or New Zealand dollars, where rates were higher. You earn the interest rate differential — the carry — as long as the exchange rate doesn't move against you by more than that differential.
And when it works, it's a slow, steady grind of profits. When it breaks, it breaks spectacularly.
August twenty twenty-four was the most recent example, and it was dramatic. The yen carry trade had become one of the most crowded trades in the world. Institutions had been borrowing yen at near-zero rates and buying higher-yielding currencies for years. Then the Bank of Japan started signaling rate hikes, and the yen started strengthening. When the yen moves against the carry trade, it creates a feedback loop — traders have to buy yen to cover their short positions, which strengthens the yen further, which forces more covering. In a single day in August twenty twenty-four, dollar-yen moved three percent. That's an enormous move for a major currency pair. It was not driven by trade flows or geopolitics. It was the unwinding of a speculative institutional position.
Three percent in a day. For context, a typical daily move in dollar-yen is maybe half a percent.
Right. And the flash crash in sterling in October twenty sixteen was even more extreme — a six percent drop in a matter of minutes. That was triggered by a combination of algorithmic trading and institutional stop-loss orders cascading into each other. No retail trader caused that. No trade flow caused that. It was the plumbing of the institutional market having a seizure.
Which brings us to the AI and arbitrage question. Daniel's observation is that in an era of easy access to AI, simple arbitrage opportunities should be long gone. And I think that's basically correct. But I want to understand what's actually replaced them.
Let's define what we mean by arbitrage first, because there are different kinds. The simplest form is locational arbitrage — buying euros in London where they're cheap and simultaneously selling them in New York where they're expensive. In the nineteen eighties, you could do this. Today, the latency between London and New York is measured in microseconds, and the margin is fractions of a pip. That trade is completely dead for anyone who isn't a high-frequency trading firm with a direct fiber connection and a server colocated at the exchange.
And AI didn't kill that — fiber optics and colocation did.
What AI and machine learning have done is change the nature of the edge that institutional players have. The firms that dominate spot FX now are not traditional banks — they're quantitative trading firms like XTX Markets, Citadel Securities, Jump Trading. XTX Markets alone now handles roughly eight percent of global spot FX volume. That's more than most major banks. They're not doing simple arbitrage. They're using machine learning to predict very short-term price movements based on order flow patterns, correlation breaks, and statistical relationships that are invisible to a human trader.
So the edge has shifted from being faster than the next guy to being smarter about pattern recognition.
And having more data. These firms see order flow from multiple venues. They can detect when a large institutional order is being worked through the market, and they can adjust their pricing accordingly. It's not arbitrage in the traditional sense — it's more like a very sophisticated form of market making that extracts information from the flow itself.
And the retail trader with an AI tool they downloaded last week is competing against this?
They're not competing. They're providing liquidity to the people who are competing. The retail trader who thinks they've found an arbitrage opportunity has almost certainly just encountered a spread widening, a requote, or a latency difference between the price their broker shows them and the price that actually exists in the market. It's not an opportunity — it's an illusion created by the retail execution environment.
That's brutal. But I suspect it's also true.
It is true, and the data backs it up. The European Securities and Markets Authority publishes periodic reports on retail trading outcomes. The consistent finding is that the vast majority of retail traders lose money, and the average holding period for a retail FX trade is measured in hours, not days. Meanwhile, the institutional speculators who actually move prices are holding positions for weeks or months.
So we've got this layered picture. At the top, institutional investors — pension funds, hedge funds, insurance companies — are driving price formation through directional bets, portfolio rebalancing, and carry trades, mostly on time horizons of days to months. Below them, the banks are intermediating, warehousing risk, and hedging dynamically, which amplifies the price impact. Below that, high-frequency trading firms are providing liquidity and extracting information from order flow on microsecond time horizons. And somewhere at the very bottom, retail traders are placing bets that mostly never leave their broker's internal book.
And the key thing to understand about the structure is that the derivatives market absolutely dwarfs the spot market. FX swaps alone account for about three point eight trillion dollars in daily turnover. That's half the entire market. These are instruments used primarily by institutions to manage funding and hedge positions. When a European pension fund buys US bonds, it typically hedges the currency exposure using an FX swap. That swap transaction shows up in the BIS data as foreign exchange turnover, even though no one is taking a directional view on the euro. It's hedging, but it's hedging that creates volume and moves prices.
So even the distinction between speculation and hedging starts to break down when you look at it closely. A pension fund hedging its US equity exposure is not speculating on the euro, but its hedging activity affects the euro-dollar rate just as much as a hedge fund's directional bet would.
And this is why I think the most important thing for anyone trying to understand FX markets is to let go of the trade narrative entirely. Exchange rates are not primarily driven by who's buying more goods from whom. They're driven by capital flows — who's buying whose bonds, whose stocks, whose real estate. The trade balance is a tiny fraction of the capital account flows that actually move currencies.
Which means if you're trying to understand where the euro-dollar rate is going, you should be watching interest rate differentials between the Fed and the ECB, not the eurozone trade surplus.
The interest rate differential is the single most important driver of medium-term currency moves. It determines the carry, it determines where capital flows, and it's what institutional investors are actually trading on. The Commitment of Traders report from the CFTC, which shows positioning in currency futures, is a much more useful indicator than any trade statistic.
Let me pull on another thread Daniel raised. He mentioned smaller investors using FX as a trading strategy. If the institutional market is so dominant, is there any viable approach for someone who isn't a pension fund or a hedge fund?
I think there is, but it requires being honest about what your edge cannot be. Your edge cannot be speed — you will never be faster than XTX Markets. Your edge cannot be information — you will never know more about order flow than the banks that see it. Your edge cannot be cost — you will never get better execution than an institution trading in size. So what's left?
Time horizon and risk tolerance.
If you're willing to hold positions for months, and you're willing to tolerate drawdowns that would get a professional fund manager fired, you can potentially exploit the fact that institutional investors are often forced to be short-term in their thinking. A pension fund has monthly rebalancing requirements. A hedge fund has quarterly redemption windows and risk limits. An individual with no external constraints can sit through volatility that institutions cannot.
So the retail edge, if it exists at all, is being able to be patient in a way that professionals can't afford to be.
That's the theory. Whether most retail traders actually have that patience is a different question. The data suggests they don't. But structurally, it's the only edge available.
And what about the AI question from the other direction? Daniel seemed to be asking whether AI has democratized anything or just concentrated power further.
The evidence so far points to concentration, not democratization. The firms that are winning in AI-driven FX trading are the ones with the most data and the best infrastructure — the Citadels and XTXs of the world. AI tools available to retail traders are mostly marketing. They're not giving anyone an edge against the institutional market. They might help with chart pattern recognition or backtesting, but they're not competing with the machine learning models running on institutional infrastructure.
So the gap is actually widening, not narrowing.
I think that's the uncomfortable truth. Every technological advance in trading — from electronic execution to algorithmic routing to machine learning — has initially been sold as a democratizing force. And each time, the primary beneficiaries have been the institutions that already had scale. The retail trader gets a slightly nicer interface, but the underlying structural disadvantage remains.
There's a parallel here to what we see in other markets. The same dynamic plays out in equities, in fixed income. The technology that was supposed to level the playing field ended up tilting it further.
Because technology amplifies the advantage of scale. If you have more data, better infrastructure, and lower costs, a new technology makes you more effective. If you don't have those things, the technology doesn't close the gap — it just makes the gap more visible.
Let's zoom out for a moment and talk about what this means for understanding the world. If exchange rates are driven by institutional capital flows rather than trade, that changes how we should interpret currency moves in the news.
Completely. When you see a headline that says "Euro falls on trade war fears," what's actually happening in most cases is that institutional investors are adjusting their portfolios in anticipation of how a trade war might affect interest rates and growth, which then feeds into currency positions. The trade war is real, but the transmission mechanism is through capital flows, not through the trade balance directly.
So the headline is not wrong exactly, but it's skipping several steps in the causal chain.
And those steps matter, because if you skip them, you end up with a model of currency markets that doesn't actually predict anything. You'll be watching trade statistics and wondering why the euro isn't moving the way you expected. Meanwhile, the real action is in the interest rate swap market and the CFTC positioning data.
Which brings us to something practical for listeners who aren't trading but are trying to understand what's happening. What should they actually be watching?
Three things. First, interest rate differentials — the gap between what the Fed pays and what the ECB pays. That is the gravitational center of medium-term currency moves. Second, the CFTC Commitment of Traders report, which shows positioning extremes. When everyone is already long the dollar, that's often a sign the move is exhausted. Third, large option expiries. When there's a big options barrier at a certain level — say, one point ten on euro-dollar — the hedging activity around that level can create a magnetic effect on the spot price. These are the real drivers. Trade statistics are a distant fourth.
And the BIS data we've been citing — the next Triennial Survey should be coming out soon, right?
The next one covers April twenty twenty-five and should be published later this year or early next. The key question is whether the share of other financial institutions has broken above sixty percent. If it has, that would confirm that the institutionalization of FX markets is still accelerating. My guess is it will.
Because the trend has been one-way for a decade.
And because the structural forces pushing in that direction haven't changed. Pension funds are getting bigger. Hedge fund assets under management keep growing. Passive investing, which requires constant currency hedging, continues to take share from active management. All of these trends point toward more institutional flow, not less.
So the market becomes more dominated by a smaller number of very large players.
Which has implications for market stability. When everyone is positioned the same way — like the yen carry trade in twenty twenty-four — the unwinding can be violent. The institutionalization of FX markets doesn't necessarily make them more efficient in the way economists mean. It can make them more prone to crowded trades and sudden reversals.
Because the diversification is an illusion. You have a hundred different pension funds, but they're all running similar rebalancing algorithms and similar hedging strategies. When they all need to go the same direction at the same time, you get a flash crash.
The GBP flash crash in twenty sixteen, the yen carry unwind in twenty twenty-four, the Swiss franc shock in twenty fifteen when the central bank removed the peg — all of these were institutional flow events. Retail didn't cause any of them. And yet the retail trader is the one who gets blamed for being irrational.
The narrative is always about irrational speculators, but the actual systemic risk comes from institutional crowding.
And the institutional crowding is itself a product of the incentive structure. A pension fund manager who loses money because everyone else lost money in the same carry trade won't get fired. A pension fund manager who loses money on a contrarian bet that no one else made will get fired. So the rational thing is to stay with the herd.
Which is the oldest story in finance, really. Just playing out in a seven and a half trillion dollar daily market.
And now: Hilbert's daily fun fact.
Hilbert: The caramba, a string instrument long attributed to indigenous Lenca musicians in western Honduras, was widely cited in interwar ethnomusicology as a pre-Columbian survival. In nineteen thirty-eight, researcher María Dolores Torres demonstrated that the instrument was actually introduced by Spanish missionaries in the eighteen century and bore no connection to Lenca musical traditions — the earlier attribution had been based on a mislabeled museum specimen and copied uncritically for decades.
A mislabeled museum specimen creating decades of wrong scholarship. That feels uncomfortably on-brand for academia.
At least they corrected it eventually. Most fields just double down.
So where does this leave us? We've established that the FX market is overwhelmingly driven by institutional speculation, that retail is noise, that simple arbitrage is dead, and that AI has concentrated power rather than distributed it. The open question, I think, is whether that trend continues or whether something disrupts it.
The next BIS survey will tell us a lot. If other financial institutions cross sixty percent of total volume, that's a market that's becoming more concentrated, not less. And I think the technology question is still open. It's possible that genuinely decentralized finance — not the marketing version, but actual peer-to-peer settlement — could change the structure. But we're years away from that being relevant to the major currency pairs.
For now, the practical takeaway for anyone listening is pretty clear. If you're trading FX, understand what pool you're swimming in. Your edge, if you have one, is not speed or information — it's patience and a different time horizon. And if you're just trying to understand why currencies move, ignore the trade headlines and watch the capital flows. The market is a mirror of global finance, not global trade.
And that's the thing that's hardest to internalize, because the trade story is so intuitive. Goods cross borders, money changes hands, exchange rates adjust. It makes sense. The reality — that a pension fund in Norway rebalancing its portfolio has more impact on the euro-dollar rate than a German car exporter — is deeply counterintuitive. But it's true.
Thanks to our producer Hilbert Flumingtop. This has been My Weird Prompts. If you want to send us your own questions, email the show at show at my weird prompts dot com. We'll be back soon.
