The Two-Cent Position: Loss Aversion, Live
Somewhere on Kalshi right now, a trader is holding a “Cut 25bps” contract on the September Fed decision that he bought for forty-five cents in April. It is worth two cents today. He has not sold it. He is not going to sell it. And if you ask him why, he will not say “I think the Fed is about to cut.” He will say something closer to “I’ve already lost this much, might as well see it through.” That sentence is the entire subject of this issue.
Prediction markets are supposed to be the cleanest laboratory in finance for watching human bias meet a hard deadline. Every Kalshi contract resolves to a dollar or to nothing, on a known date, verified by an outside fact. There’s no earnings call to hide behind, no “long-term thesis” to retreat into. And yet the single most reliable pattern we track across every category: Fed decisions, House control, weather, awards season shows traders behaving as if the price they paid still matters to the market. It doesn’t. It never did. The market has no idea what you paid, and it isn’t waiting for you to get even.
How we got to two cents
Rewind to late April. The Fed’s own dot plot was penciling in one cut for 2026. Wall Street desks were split between one and two. And Kalshi traders had priced roughly a 60% probability of at least one 2026 cut: nearly double the 27% CME Fed Funds futures carried at the time. That gap was the widest divergence of the cycle, and it was loud enough that ProCap flagged rate-sensitive homebuilder and REIT names as a way to trade the spread between what futures believed and what the retail-heavy prediction market believed.
The prediction market was betting on a version of the Fed that leaned dovish. Then the Fed changed. Kevin Warsh was confirmed as chair, and by his first meeting as chair in June, the “zero cuts in 2026” contract jumped from 68¢ to 77¢ in a single session before the minutes were even released. By July, Kalshi traders had flipped to seeing 54% odds of a hike this year, not a cut. The entire premise the April longs were built on had inverted. And by the September ladder, the “Cut 25bps” contract that had traded in the 40s and 50s in the spring had collapsed to 2¢ bid, 99¢ ask on No.
Some traders are now pricing a 2026 hike, not a cut. None of that is the interesting part. Markets repricing on new information is just markets doing their job. The interesting part is what didn’t happen: open interest on that contract didn’t collapse to nothing. A meaningful slice of the April longs are still sitting in the position. Not because they’ve done new analysis and concluded 2¢ is cheap: ask around, and almost nobody will defend that number on the merits. They’re sitting in it because selling means admitting the loss is real, and holding lets them keep pretending it might not be.
The Disposition Effect
The well-documented tendency to sell winning positions too early and hold losing positions too long: first formalized by Hersh Shefrin and Meir Statman in 1985, building directly on Kahneman and Tversky’s Prospect Theory.
The mechanism: your brain doesn’t evaluate a position against its current fair value. It evaluates the position against your reference point: almost always your entry price. A gain relative to that reference point feels good but not linearly so; the pleasure curve flattens fast. A loss relative to that reference point hurts sharply, and the pain curve doesn’t flatten: it keeps punishing you the longer you sit in it. Selling at a loss converts a fuzzy, deniable “unrealized” loss into a sharp, final, realized one. So you don’t sell. You wait. Waiting costs nothing today, or so it feels: which is exactly the illusion.
We covered the theory side of this in Issue No. 3: loss aversion means losses register roughly twice as painfully as equivalent gains feel good, and sunk cost thinking convinces you that money already spent is still yours to protect. The disposition effect is where those two biases stop being abstractions and start showing up in your position blotter. It’s the difference between understanding loss aversion and watching yourself do it.
Why prediction markets make it worse, not better
You’d think a market with binary, verifiable resolution would cure this. It doesn’t: it just delays the reckoning. In equities, a losing position can drift sideways for years, giving you infinite runway to avoid deciding. On Kalshi, the contract has an expiration date. That should force discipline. Instead, what we see is traders treating the approach of resolution as a reason to hold rather than fold :a variant of the sunk cost fallacy we’d call terminal-value hope: “It’s already down this far, and it settles in six weeks anyway, so I might as well see what happens.” That logic would be sound if the six weeks changed the odds. On a contract sitting at 2¢ with the entire policy committee on record leaning the other way, it doesn’t.
The same pattern shows up outside rates. On the 2026 House-control sweep market, traders who built positions during the spring’s more competitive generic-ballot numbers have been slow to trim as the picture clarified, anchoring to their entry level on the Democratic Sweep or Republican Sweep contract rather than re-underwriting the position against this week’s price. Longshot bias compounds it further: a beaten-down contract at a few cents starts to look “cheap” purely because of where it used to trade, not because of any real edge at that price.
What actually fixes it
You can’t out-willpower a hardwired bias. What you can do is change the question you ask yourself before every hold-or-fold decision. AFG’s rule, unglamorous as it is: your entry price is not a variable in the decision. The only two variables that matter are the contract’s current price and your current, honest estimate of fair value. If you wouldn’t buy the contract fresh today at today’s price, you have no business holding it just because you already own it.
Delete the cost basis from the decision. Literally cover it if you have to. Ask only: at today’s price, is this contract’s edge positive or negative against my current probability estimate?
Set the exit before you enter.A stop-loss level or a “kill condition” decided in advance, before you have skin in the game, is a decision your calm self made for your losing self. Honor it.
Reframe the sale as data, not defeat.Closing a losing position isn’t a verdict on your judgment in April. It’s new information processed correctly in August. Treat it as the trade working, not the trader failing.
Watch for terminal-value hope specifically. If your reason for holding a beaten-down contract is “it resolves soon anyway,” ask what new information you expect between now and resolution. If the honest answer is none, that’s not a reason to hold: it’s the absence of one.
Size the next trade so this can’t happen again. Positions that are painful to close are almost always oversized relative to the conviction behind them. Right-sized bets are ones you can exit unemotionally.