Assay

Methodology

How Assay verifies crypto manager performance.

This document describes the methodology Assay applies to verify crypto and DeFi manager performance. It is the single source of truth for how returns, risk metrics, composites, and benchmarks are computed and disclosed in the Manager Universe.

Why crypto needs rigorous performance verification

Crypto asset management lacks the trust infrastructure that traditional finance built over decades. There is no equivalent of fund administration, GIPS verification, custody attestation, or independent NAV reporting at institutional standards. Managers raising capital rely on self-reported track records, ad-hoc audited statements, and personal trust. LPs allocate on faith — and have been burned repeatedly.

The closest existing solutions are databases (such as Crypto Insights Group) that aggregate manager-submitted data without independently verifying it, and on-chain vault platforms (Enzyme, dHEDGE) that solve part of the problem for on-chain strategies but have no benchmark methodology, no GIPS-aligned composite construction, and no coverage of off-chain CEX strategies.

Assay's position is straightforward: claimed returns are unverified, and unverified claims don't enable credible institutional allocation. The research page contains a full competitive landscape and the moat analysis behind that thesis.

Our methodology at a glance

  • Time-Weighted Return per GIPS, computed monthly from primary sources.
  • Pre-registered benchmarks, timestamped at engagement scoping and locked for the duration of the engagement.
  • Composite construction over all discretionary fee-paying accounts following a given strategy.
  • Liquidity-adjusted valuation for positions exceeding ADV concentration thresholds.
  • Wallet clustering for continuous detection of undisclosed external flows.
  • Custody attestation via whitelisted custodians or on-chain vault contracts.
  • Anomaly screens on return distributions, position concentration, and counterparty exposure.
  • Independence from manager equity, token holdings, and trade flow.

GIPS provisions we follow directly

The Global Investment Performance Standards published by CFA Institute are the global benchmark for investment performance disclosure. Assay applies the following GIPS provisions to crypto manager performance verification:

Time-Weighted Return (Provision 2)

Returns are computed using a Modified Dietz approximation at sub-monthly granularity, geometrically chained to monthly returns and annualized over GIPS-standard horizons (1Y, 3Y, 5Y, 10Y). External cash flows are neutralized: every deposit is treated as a contribution, every withdrawal as a distribution, and performance reflects only the change in value of capital while it was deployed.

Composite construction (Provision 3)

All discretionary fee-paying accounts managed under a single strategy are included in the composite for that strategy — no cherry-picking the winners. Assay enforces composite construction at engagement scoping: the manager declares which accounts and wallets belong to the strategy composite, and any account meeting the discretion criteria must be included.

Valuation (Provision 5)

Daily marks for liquid positions using VWAP across qualifying venues. Liquidity-adjusted fair value for positions exceeding concentration thresholds. Stablecoin valuation at peg with deviation disclosure when off-peg by more than 50 basis points. Airdrops and fork events marked at first liquid trade.

Disclosures (Provision 4)

Required disclosures include:

  • Gross and net returns; fee schedule
  • Benchmark description and rationale (locked at engagement)
  • Composite description (in-scope and excluded accounts)
  • Dispersion measure (asset-weighted standard deviation)
  • 3-year ex-post standard deviation once available
  • Presence and nature of leverage or derivatives
  • Custody arrangement

Locked-in-advance benchmarks (Provision 4.A.1)

The benchmark must be declared in advance and described in the disclosures. Assay enforces this cryptographically — timestamped commitment at engagement. The manager cannot retroactively change benchmarks to flatter their returns.

Adaptations for digital assets

GIPS was written for assets that close at 4pm Eastern, settle T+2, and trade on a small number of well-supervised venues. Crypto trades 24/7 across hundreds of venues with wildly varying liquidity, settles on-chain in seconds, and includes asset types (LP positions, staked tokens, structured vaults) that simply have no TradFi analog. Our adaptations are operational, not methodological — the underlying GIPS principles (TWR, fair value, no cherry-picking, locked benchmarks) carry over unchanged.

  • Reporting period close. Crypto markets do not close. We standardize on UTC month-end as the closing instant and use VWAP across the trailing 4 hours rather than a single last-print.
  • On-chain settlement. Wallet activity is settled on-chain in seconds; we treat inbound transfers as contributions and outbound as distributions, after running them through wallet clustering for related-party detection.
  • Composite scope across CeFi and DeFi. A single strategy may run across centralized exchanges and DeFi venues simultaneously. The composite includes both legs, with separate disclosure of CeFi vs DeFi exposure.
  • Position types beyond spot and futures. Liquidity-pool positions, staked tokens, points-earning positions, and locked vault tokens each have explicit valuation rules in the methodology supplement.

The market manipulation problem and our defenses

A manager running a strategy in a thinly-traded token can in principle manipulate their own mark by placing small orders at period close. They show big paper gains; in reality they couldn't actually exit at those prices. This is “marking your own book” — a known TradFi problem but more acute in crypto: many tokens trade on fragmented venues with wildly different prices, some tokens have $5–50M total daily volume (easy to move with a $100K order), 24/7 markets mean there is no clean close print, and reported volumes are often inflated by wash trading.

We have three primary defenses, plus several supporting controls.

1. Volume-weighted average pricing across venues

Position marks use VWAP across all liquid venues over a defined window (typically the trailing 4 hours of the reporting period) rather than the closing tick on any single venue. This neutralizes most close-of-period spoofing. Our venue qualification list is published as a supplement to the methodology, and we exclude venues whose volumes fail wash-trading screens (CoinGecko Trust Score, Kaiko, CCData).

2. Liquidity-adjusted valuation haircuts

If a manager holds $10M of a token with $500K daily volume, marking the full position at market price is fiction — they couldn't exit at that price. GIPS already requires “fair value” valuation. We operationalize this by applying a haircut to illiquid positions based on days-to-liquidate at realistic market impact. This is what serious crypto funds already do internally; we make it standard and auditable.

3. Position concentration disclosure

Any position larger than 5% of the token's average daily volume or 10% of the manager's AUM is flagged in the verification report. LPs see the concentration. They can decide what to make of it.

Supporting controls

  • Liquidity scoring on the Manager Universe. Median market liquidity of held positions is displayed on every manager row. Two managers with identical Sharpe are not equivalent if one runs in $1B-daily-volume markets and the other in $5M-daily-volume markets.
  • Wash-trade-adjusted volume. We do not take reported exchange volumes at face value. Volume figures used in valuation are filtered using third-party adjusted volume data.

Benchmark selection

Benchmarks fall into three categories:

Beta-style (broad market)

Bitcoin (BTC), Ether (ETH), Bitwise 10 Large-Cap (BITW), CoinDesk 20 (CD20), MarketVector Digital Assets 100, CF Cryptocurrency Ultra Cap 5 (Kraken/CF Benchmarks), CCI30, S&P Cryptocurrency Broad Digital Market.

Sector / thematic

DeFi Pulse Index (DPI), Metaverse Index (MVI), Bitwise Crypto Industry Innovators (BITQ), Coinbase 50.

Risk-free / cash equivalents

3-month T-Bill rate (USD), USDC lending rate (AAVE/Compound) for stablecoin-denominated strategies, published exchange funding rates for funding-rate basis strategies.

Composite / blended

Custom blends declared at engagement (e.g., 50% BTC / 50% ETH, monthly rebalanced) are supported and are timestamped at the moment of commitment. The benchmark cannot be retroactively changed.

Three-tier verification process

Self-Reported is immediate; the manager submits a tear sheet which is presented in standard format and explicitly labeled unverified. Account-Verified is typically completed within a week: read-only API and on-chain wallet linkage, TWR computed on the full reconstructed history. Assay-Verified is a 14–28-day engagement covering benchmark pre-registration, composite construction, custody attestation, continuous wallet clustering, anomaly screens, and the full set of GIPS-aligned disclosures.

Wallet clustering and external flow detection

Per GIPS, deposits and withdrawals are external flows that must be neutralized in TWR computation. The risk in crypto is that a manager could route money between wallets they secretly control, mark inbound transfers as legitimate external contributions, and inflate apparent returns.

We require pre-declaration of all wallets the manager controls at engagement scoping. Wallet clustering analysis (similar to Chainalysis/Arkham methodology) runs continuously to detect deposits from wallets that exhibit behavior consistent with related-party flows: shared funding sources, transaction-pattern correlation suggesting common control, or shared infrastructure (custodian, relayer, exchange deposit address). Any flow flagged by clustering analysis is held aside and reviewed before being treated as a genuine external contribution.

Custody verification

The custody arrangement must be disclosed in the engagement memo. Assay supports whitelisted custodians (Fireblocks, BitGo, Anchorage, Copper, Ledger Enterprise) via API attestation, and on-chain vault contracts via direct chain reads. Withdrawal permissions are not granted to Assay; we hold read-only access. Any change in custody arrangement during the engagement triggers a re-attestation.

Anomaly detection methodology

Beyond TWR and clustering, we screen verified strategies for return-distribution anomalies that correlate with fabricated track records. Indicators include implausibly low monthly return variance relative to declared strategy type, monthly return autocorrelation patterns inconsistent with the instrument set, Sharpe ratios above 4 over multi-year horizons (suspicious in any asset class, more so in crypto), and concentration of returns in a small number of months that do not align with identifiable market regimes.

Flagged strategies are not de-listed automatically; they are placed in the “Under Review” tier while the verification team runs targeted follow-ups. The methodology supplement enumerates the specific tests, threshold parameters, and remediation paths.

Independence and conflicts-of-interest policy

Assay receives no compensation from managers beyond the disclosed verification fees. We hold no investment positions in any verified manager or in any tokens forming a substantial portion of any verified strategy. Our staff is prohibited from trading in tokens held by verified managers. The conflicts policy is published on this site and audited annually by an independent third party.

Internal access to manager trade data is logged. Verification team members must complete annual training on the conflicts policy and sign personal trading attestations. Violations are grounds for termination.

Methodology revision process

Any change to methodology that materially affects how returns or risk metrics are computed is published with a 30-day notice period. Verified managers receive direct notification with a chance to raise objections before the change takes effect. Once effective, prior periods are not re-stated; the change applies prospectively from a defined effective date, and the verification report annotates affected periods.

This page is the canonical methodology document. Material changes are versioned, dated, and archived; the history is available on request.