This article explains a performance-attribution framework. The worked numbers are arithmetic illustrations, not claims about any strategy. Nothing here is investment advice. Decisions are yours.
Beta is how much of a portfolio's return is just exposure to the market's own moves; alpha is what remains after that exposure is priced. In crypto the market factor has a name: BTC. Most "outperformance" in this asset class is high-beta BTC exposure wearing a strategy costume, and one regression strips the costume.
The frame, from zero
The attribution logic is borrowed from equity finance: regress a strategy's returns on a market benchmark's returns. The slope is beta, the strategy's built-in dose of market exposure; the intercept is alpha, return the market exposure does not explain. Equities regress against a broad index. Crypto's honest benchmark is BTC, or a majors basket, because the same one-factor correlation structure that runs through crypto's market sectors means BTC's moves are the tide every boat rides. The regression asks the only fair question: after accounting for how much tide this portfolio was floating on, did the captain add anything?
The arithmetic that deflates bull-market genius
The trap this framework exists to catch: in a year where BTC doubles, a portfolio of higher-beta alts at, say, beta 1.5 should return around +150% as its fair share of tide. If it returned +120%, it made spectacular money and negative alpha: the operator would have done better holding levered beta and skipping the stock-picking. This is the quantified version of bull-market genius: bull markets manufacture returns that feel like skill and regress into exposure. The same arithmetic runs mercilessly in reverse: a beta-1.5 book in a halving market should lose 75%, so losing only 50% is genuine, invisible alpha, earned while looking terrible. Alpha is regime-symmetric; screenshots are not, which is why sellers publish bull-year returns and never publish betas. The 95% lie runs on exactly that omission.
A worked case from our own experiment
Our transplant experiment produced a clean specimen: the majors-only variant finished its window at roughly the same multiple as simply holding BTC, at a meaningfully shallower maximum drawdown. In alpha-beta language: no return alpha over the benchmark, but a genuinely different risk path, which is a real and honest kind of value, RISK alpha rather than return alpha, visible only when both the return and the path are on the table, and the path is only legible against crypto's drawdown base rates. We publish it that way deliberately: a system that matches its benchmark's return while softening its path has to say so in exactly those words, because "our system returned 63%" without the benchmark line is how this industry launders beta into genius.
Leverage, funding, and the beta that isn't even beta
One refinement matters in a perp-dominated market: levered exposure is not free beta. A 2x levered BTC position carries beta 2 minus the costs of carrying it: funding payments, and the volatility drag that scales with the square of exposure. Over any long window, levered beta underperforms its nominal multiple of the benchmark, which produces the industry's saddest genre of track record: strategies with negative alpha against a benchmark they are secretly just a levered, taxed version of.
The null-model discipline
The practical takeaway is a habit, and it is the cheapest audit in crypto: before crediting any strategy, feed, or fund with skill, compare it to the null model, buying and holding BTC over the identical window, then to a majors basket, then ask for the beta. Three questions, all computable from public data, which is exactly why we publish our formulas instead of a secret sauce. Most of the industry fails at question one, which is why our own experiments print the BTC-hold comparison in their headline results rather than in a footnote. Skill claims that survive the null model deserve attention; the rest is the tide, invoiced as seamanship. Decisions are yours.