In October 1998, during the yen carry unwind, retail brokers operating the Asian session faced a specific test — could their clients keep trading the following Monday. The published record of that week is unkind. What survived the tape was a small cohort of accounts that had been open more than eighteen months, the kind of number no broker prints on a landing page. That number, if I had known to ask for it in 2013 when I first funded an offshore account from Jakarta on a friend's recommendation, would have saved me the two years and the three blown deposits that followed.
Why This Is Actually True
Here is the strongest version of the argument I am going to spend the rest of this piece complicating. Regulation is the honest broker metric. Not longevity, not spreads, not leverage caps — regulation. Because a broker cannot fake a licence from the JFSA or the MAS or the FCA. The paperwork exists in a public register. You can look it up. The counterparty has to segregate client funds under a specific statutory regime, submit to periodic audits, publish accounts, and — critically — post a bond or contribute to a compensation scheme. When the tape goes bad, the regulator forces a resolution process that protects retail balances up to a stated cap. This is not marketing. This is administrative law.
I concede this argument almost fully. If you fund an account at Saxo Bank APAC in Singapore, you are sitting under the MAS wholesale market framework that has been in place since 2008 and inside a Capital Markets Services licence regime that has, in practice, done what it says it does. Interactive Brokers Asia sits under HKMA's parallel oversight and the SFC's Type 3 licence. OANDA Asia holds a MAS licence. IG Group Asia is regulated through MAS in Singapore. When Korea's FSC tightened retail forex leverage in 2009 after the accumulator-linked losses of 2008, the enforcement mechanism was licensing. When Japan's JFSA capped retail leverage at 25:1 in 2010 — following the 2005 FX law that first brought retail forex under formal supervision — the enforcement mechanism was licensing.
The tier-1 regulator field on any broker page is, on its own, a strong signal. AvaTrade lists ASIC. Exness lists FCA. FBS lists ASIC. FXTM lists FCA. HF Markets lists FCA. Those are not decorative. Each of those registrations means a specific, published audit trail. The steel-man version of "just trust the regulator" is: the state has already done the diligence for you, and the state has enforcement authority you do not have, so let the state's work substitute for yours.
But here is what the licence field does not tell you — whether the average account funded there survives the second year.
Where It Breaks Down
The regulator tells you what is legal. It does not tell you what happens to the median depositor. And in retail forex, those two questions have never been the same question.
Consider what the disclosed numbers actually are. Every FCA-authorised broker publishes a "percentage of retail investor accounts that lose money" figure on its landing page — a rule from ESMA's August 2018 product intervention measures, adopted by the FCA when it kept the framework post-Brexit. Look at the range across the operators any Asian-session trader would consider. The published loss ratios cluster between 68 and 84 percent. That is the FCA-audited number. That is what tier-1 regulation buys you. The broker is honest about it because the regulator forces disclosure. But the number itself is not something the regulator can improve — it is a property of the product, the leverage on offer, and the average client's holding period.
Now stack the broker fields from the grounding side by side. Exness offers max leverage of 2000:1 outside its FCA entity. FBS offers 3000:1. FXTM offers 2000:1. HF Markets offers 1000:1. AvaTrade offers 400:1 and prohibits scalping. Minimum deposits: one dollar at Exness and FBS, five at HF Markets, ten at FXTM, one hundred at AvaTrade. Spreads on EUR/USD standard accounts run 0.7 pips at FBS, 0.9 at AvaTrade, 1.0 at Exness, 1.2 at HF Markets, 1.5 at FXTM. Pro accounts compress those to 0.0 to 0.1. All five list at least one tier-1 regulator. All five offer Islamic accounts.
The regulator field is genuine at each of them. What the regulator does not tell you is which of those combinations — leverage, minimum deposit, spread — produces an account that is still open in eighteen months. A one-dollar minimum with 3000:1 leverage is regulated. It is also arithmetically hostile to survival. On a $100 deposit at 3000:1, a fifty-pip move against you on a standard lot is the account. That is not a broker failing to segregate funds. That is the maths of the product the broker is legally permitted to sell you.
The 1998 October yen unwind is the archive case. USD/JPY moved from the 136 area to around 111 across a compressed window that included one particularly brutal 24-hour stretch. Retail carry positions built up through 1997 and 1998 — long AUD/JPY, long NZD/JPY funded in yen — were closed at whatever bid the venue could show. The regulated Tokyo houses that survived that week did so because their retail books were smaller and their leverage lower than what the offshore venues were offering to the same clients. The regulator did its job. The regulator's job was never to make the median account survive Monday.
The Rule I Use Instead
Ask the broker for average client lifespan. Not the loss ratio — the lifespan. How long does the median funded account stay funded before it hits zero, is withdrawn to zero, or goes dormant.
No broker publishes this. The FCA does not require it. The MAS does not require it. It is not on any landing page in the grounding above and it will not be on any landing page you visit. That is precisely why it is the useful number. A metric that cannot be gamed by marketing is a metric worth chasing.
Here is the math I do when I cannot get the number directly. Take the disclosed loss ratio — let us use 75 percent as a middle-of-the-road FCA-published figure. Assume the losing 75 percent lose their initial deposit on average, and the winning 25 percent are still trading. Assume the broker's disclosed year-over-year active-client growth is, say, 20 percent — a figure any competent broker will cite in an interview if you ask. If the active client base is stable or growing at 20 percent and the loss ratio is 75 percent, then the churn rate is at least 75 percent minus 20 percent, or 55 percent of accounts turning over each year. That implies a median account lifespan under two years. Probably closer to fourteen months.
Now compare that against what a broker charges you. Exness Pro at 0.1 pips on EUR/USD, on a standard lot, is roughly one dollar of spread per round trip. On $1,000 deposit at 100:1 leverage — a conservative use of what is offered — a working strategy trading twice a day pays about $500 in spread across fourteen months. That is 50 percent of deposit in transaction cost alone. The account is not going to survive on skill because the arithmetic of the cost floor is already eating half of it before the trading result is counted.
The rule I use, then, is this. Longevity is the metric that cannot be faked because no broker prints it. If I cannot get the raw number, I reconstruct it from disclosed loss ratios and disclosed growth. Then I ask: on the deposit and leverage combination I am actually going to use, is the transaction-cost floor consistent with an account that lasts eighteen months. If yes, the licence field matters. If no, the licence field is a formality.
When the Old Rule Still Wins
There is a version of the regulator-first argument I do not fight. If you are choosing between a licensed broker in the MAS or FCA register and an unlicensed venue with no public accounts and no compensation scheme, take the licence every single time. The floor matters. The floor of licensed regulation — segregation, audits, resolution — is what makes the difference between losing your account slowly to spreads and losing your account instantly to a counterparty that closes on a Friday night and never reopens. Regulation is not the whole game. It is the price of entry to the game.
Where the licence-first rule wins outright is when longevity data is genuinely inaccessible and the deposit sits below a compensation-scheme cap. Under the FCA's FSCS, the £85,000 cap makes the regulator's floor mechanical and enforceable. Under Japan's JFSA framework, segregated trust arrangements do similar work for retail balances. If your deposit is small enough that the compensation-scheme cap covers you in full and you have no visibility into churn, the licence is doing more work than the churn number ever could.
FAQ
Why does no broker publish average client lifespan?
Because no regulator requires it and the number is bad. The ESMA-derived rule that produced the "72% of retail accounts lose money" line was a hard-fought disclosure and it stopped there. Lifespan is a downstream metric — it captures the same information plus the effect of holding period, cost floor, and platform frictions. A broker that published a fourteen-month median would be inviting a comparison against a competitor showing eighteen months, and the industry has coordinated silence on the metric.
Can I estimate lifespan from the disclosed loss ratio?
Partially, yes. If a broker discloses a 75% retail loss ratio and separately claims 20% year-on-year active-client growth, the implied churn is at least 55% of the book turning over annually — a median lifespan of roughly eighteen months, and likely shorter for accounts under $500. The estimate is rough because it treats every closed account as a loss. But it gives you a floor to test the broker's own marketing claims against.
Does high leverage automatically shorten lifespan?
Not automatically, but in aggregate it does. Leverage does not force position size — the trader chooses that. But offered leverage of 2000:1 or 3000:1, combined with minimum deposits of one to ten dollars, produces an account population where the median user is holding position sizes that a fifty-pip move liquidates. The 2010 JFSA cap at 25:1 was a direct response to Japanese retail lifespan data the regulator had gathered internally.
Is Asian-session broker regulation weaker than European?
Not weaker — differently structured. MAS Singapore's Capital Markets Services regime, HKMA's oversight of licensed corporations, JFSA's post-2005 FX law, and Korea's FSC framework each impose segregation, capital, and reporting requirements comparable to FCA or CySEC on the substantive points. The differences are in leverage caps, retail eligibility, and compensation-scheme structure. For an Asian-session trader, MAS-regulated Saxo APAC or OANDA Asia is not a downgrade from a UK book.
What did 1998 actually change for retail brokers?
It shortened the industry's tolerance for undisclosed leverage on the yen crosses. Retail carry positions liquidated across October 1998 exposed the gap between what regulated Tokyo houses were showing on published books and what offshore venues were letting the same clients hold. The tightening that followed — culminating in the 2005 Japanese FX law and the 2010 leverage cap — was slow, but the direction was set that week. Regulators started asking for the numbers brokers did not publish.
If I only have $100 to start, does any of this matter?
It matters more, not less. A hundred-dollar deposit at 500:1 leverage on a fifteen-cent-per-pip spread has a mathematical lifespan measured in weeks under any realistic trading frequency. The broker's licence protects the hundred dollars from theft. It does not protect it from arithmetic. If starting small, the useful comparison is between deposit size and total round-trip cost per week — not between licences.
What questions actually get an honest answer from a broker sales desk?
Ask what percentage of accounts funded twelve months ago are still funded today. Ask what the median deposit size is. Ask what the average time from first deposit to full drawdown is. You will rarely get numbers, but you will get texture from the pauses and the redirects. A desk that pivots hard to spread and platform features when asked about lifespan is telling you what the number is without saying it.