LY Traders · working draft
A co-founder of Kalshi, the largest prediction market, was asked on camera this
month about a striking number: 75% of people lose money on the platform. She
did not deny it. Her defence was comparative — that the number is never quoted
next to how many people lose money day trading.
The interview is worth watching, and the candour is rarer than it should be. Most
venues bury their loss statistics; this one had its own figure put to it
publicly and answered without a flinch. But the deflection is where the real
lesson sits, and it is not the lesson either side of that interview intended.
Day trading's loss rate is genuinely terrible — the best-known academic work on
Taiwanese day traders found that only about 1% earned consistent profits net of
fees, year after year. So when a prediction market says "we look like day
trading," it is citing a true number.
It is not a defence. Standing next to another bad number does not make yours
good — it establishes that a whole category of retail activity returns most of
its participants to zero. The correct reaction to discovering you are in
famous company is not relief.
What matters more is what the two numbers are made of. A day trader can lose
for years without knowing it: the feedback is noisy, a bull market hides a bad
process, one lucky week feels like an edge. Nothing settles. On a prediction
market, the event settles on Friday. The question was binary, the outcome is
public, and the ledger is legible within days. The 75% is not a different
species of loss. It is the same loss, with the noise stripped out.
This is the under-discussed property of prediction markets, and the reason
traders should pay attention to them even if they never place a contract.
Every trading venue hides its verdict behind noise. Did the strategy fail, or
did the regime change? Was the loss a cost of the edge, or a broken rule? A
trader can ask these questions for a decade without an answer, because the
sample never closes and the confounds never stop arriving. That ambiguity is
not an accident of markets — it is the medium in which self-deception survives.
A prediction market removes the ambiguity by force. The contract is binary.
The resolution date is fixed. The evidence arrives whether or not you are
ready for it. A venue where the average participant's verdict lands within
days is the most honest losing venue ever built — honest not because it is
fair, but because the feedback cannot be postponed.
The instinctive taxonomy says: stocks and indices are investing; event
contracts are gambling. The 75% figure seems to confirm it.
But the taxonomy does not survive contact with the same venue's winners. On
the other side of that 75% sits a quarter of participants who profit — and the
durable ones are not luckier, they are structured differently: market makers
harvesting the spread, arbitrageurs closing cross-venue mispricings,
participants with genuine informational or analytical edges on the specific
event. The same instrument, the same contracts, the same Friday settlement —
and opposite results, divided not by what they traded but by whether they had
a verified, repeatable edge against their actual counterparty.
That dividing line runs through every market ever invented. Most retail
traders are on the gambling side of it while holding instruments with
dignified names. The prediction market did not create the gamblers; it just
issued them a venue where the truth arrives on schedule.
Not "avoid prediction markets" — that is the moral-panic reading, and it
misses the mechanism. Three things instead:
any definable period whether your process works, you are not being patient —
you are postponing the verdict indefinitely. Define the sample in advance:
how many trades, over what window, against what baseline.
cannot articulate why the person selling to you might be wrong, you are the
one being harvested. This question is uncomfortable in every market and
unavoidable on a binary contract.
interview noted the influencers posting easy-money wins while 75% lose
underneath. The feed shows the survivor; the platform knows the denominator.
That gap is not a prediction-market problem — it is the ambient condition of
all retail trading content, and it scales with whatever venue is currently
growing fastest.
The 75% figure comes from the venue's own reporting and a Bloomberg interview;
its measurement window and definition of "lose money" are not ours to
reconstruct. Day-trading loss rates are estimates with their own methodological
debates. None of this is investment advice, and none of it is a prediction
about any market or venue — including the one being dissected.
The uncomfortable summary: prediction markets did not invent retail losses.
They invented the receipt.
The market changes. The rules don't.
The venue keeps proving the point. Within days of this essay's draft, Kalshi launched single-stock perpetual futures — CFTC-cleared, settling twice daily — and reported, by its own account, over $300m traded on the first morning. The instruments keep multiplying. The receipt arrives faster with each one, and the same dividing line from section 3 ships with every new contract: the same instrument, opposite results, decided by who holds a verified edge against their counterparty.
