Everything directional reduced to beta
We tested directional strategies the hard way: momentum, mean-reversion, breakout, trend, and every combination across timeframes and coins. Then we did the one test most people skip — we subtracted the market’s own return (a beta-strip).
After removing beta, every directional “edge” went to zero or negative. The win rate could be anything; the expectancy could not be made positive (see why an 87% win rate still loses). The conclusion is uncomfortable but clean: for retail, on public data, direction is not predictable enough to beat fees.
So we asked a different question: is there anything that makes money without predicting direction at all?
One thing survived: selling volatility
There is. It is called the volatility risk premium (VRP), and it is the difference between what option buyers pay for protection and what actually happens.
On five years of BTC options data (Deribit 30-day ATM IV, 2019–2024), implied volatility (the price of options) averaged about 61%, while realized volatility (what the market actually did) averaged about 53%. That gap — roughly +8.4 percentage points, positive 79% of the time — is an insurance premium. Buyers overpay for protection; the seller of that protection collects the difference.
The seller is not betting up or down. The seller is the insurance company. You win when the world turns out calmer than the premium implied — which, on average, it does.
Why it’s a real edge and not disguised beta
The return is uncorrelated with market direction (correlation to BTC ≈ +0.03, same proxy model), and it survived the beta-strip that killed everything else. Two honest qualifications, both from our own audits:
- That +0.03 is a tautology, not evidence — a symmetric straddle is uncorrelated with direction by construction. The exposure actually taken on is volatility (correlation −0.93 against the realised move): a different risk, not a direction-free free lunch (
research/findings/vrp_audit_bias_20260620.md:23,48). - The “Sharpe near 2.0” figure (defined-risk condor capped at 10% loss, proxy model, 2019–2024) was refuted the same month: the cap truncated losses without paying for the long wings. Re-derived with a true condor payoff and realistic cost, the central estimate is Sharpe 0.92 (
research/findings/vrp_audit_math_20260620.md:5,25).
The honest caveats
This was a strong lead. It is now a killed one, and the kill came from our own data:
- The real-quote backtest has since run, and it says NO-GO (2026-07-22, OOS-confirmed 2026-07-23). On real Deribit bid/ask snapshots a 7DTE iron condor earns a negative Sharpe before cost (ATM fly −0.09, 1SD condor −0.29) and −0.49 / −0.68 after spread and fees, cumulative −34% / −31% (
research/findings/20260722_vrp_condor_realfills.md:3-8). A 2026-08-01 pre-screen of a two-leg strangle is still not a GO (research/findings/strangle_cost_prescreen_20260801.md:3). The strategy page is now markedkilled. This bullet used to say the 2026-08 data-depth milestone had been reached and the backtest was under review — it had already run, six weeks earlier, with the opposite answer. - It needs options small enough for a small account. On OKX, ETH options fit: one contract is 0.1 ETH — about $246 of notional at ETH $2,457 (OKX instruments API, 2026-09-05) — so a condor capped at about 10% of notional risks about $25, roughly 0.5% of a $5k account. A BTC contract is 0.01 BTC, about $797, only about 3.2× larger. Corrected 2026-09-05: this line used to say
~$1,700 of notionalandBTC options are ~37× larger— it counted a whole ETH as one contract and used the BTC/ETH price ratio as a size ratio. The dollar amounts are a snapshot and do not update themselves. - Defined-risk only — iron condors with a hard maximum loss, weekly loss limits, and a circuit breaker. Never naked.
We left this article up on purpose, with the verdict written into it. The premium is still measured and still there — the part that failed is the vehicle we chose to harvest it with. That distinction is the whole point of publishing verdicts instead of signals.
Gambler vs house
A directional trader is a gambler — even a skilled one converges toward the cost drag over time. A premium harvester is the house — small, frequent, uncorrelated edges that compound. You do not need to be right about direction. You need to be paid for bearing a risk other people want to offload.
That is the argument. What we have not found is a vehicle that survives real bid/ask — the condor did not, and no substitute has cleared a pre-screen. You can verify every one of the directional failures yourself on the simulator, and read the unedited verdict on each strategy we ran, this one included. We show what fails. Including when the thing that fails is the one we were most hopeful about.