Our Blog

Backtesting in Crypto: Why Your Data Source Decides the Result

Every backtest produces a number, and most of them are wrong for reasons that have nothing to do with the strategy. Only about one in four coins listed in 2014 survived to 2021, corrected price series hide errors a live strategy would have hit, and data from a single exchange teaches a strategy that exchange's quirks instead of the market. This article covers all four biases, with the academic evidence behind each, and sets out the five properties backtest data needs.

Crypto Market Data for AI Agents: How Agents Access Prices, and Why It Changes What "Accurate" Means

A trader who sees Bitcoin at $5,400 knows something is wrong and checks another source. An AI agent receives the number and acts on it. This article explains how agents actually find and use market data through the Model Context Protocol, how they pay for it through x402, which crypto providers already run MCP servers, and why the requirements for price quality rise, not fall, once a human leaves the loop.

What Is Price Accuracy in Crypto? The Accuracy Triangle Explained

In a market without a consolidated tape, "the price" is an estimate, and different sources produce different estimates at the same moment. This article defines price accuracy, introduces the Accuracy Triangle of breadth, depth and latency, shows how to measure deviation in basis points, and walks through two documented failures, Pyth in 2021 and Mango Markets in 2022, to show which side of the triangle broke.

Crypto Market Data: A Complete Guide

Most guides to crypto market data are a list of data types, and this one is too, in its first half: trade data, order books at three depths, OHLCV, derived prices, reference rates, derivatives data, on-chain data and metadata, plus how each is delivered and where crypto breaks the expectations of anyone arriving from equities. The second half covers what the lists leave out: where a price actually comes from, the five questions that separate a single-exchange quote from a filtered market-wide aggregate, and what each user group needs the data to answer.

How Oracle Price Feeds Work: A Publisher's View of Getting a Price On-Chain

A smart contract cannot see the outside world. Everything it knows about prices it has to be told, and it acts on the number without question. This article follows a price from an exchange trade to an on-chain value: who aggregates it, who signs it, how the network combines submissions, who validates them, and how push and pull delivery differ. It compares the eight networks Blocksize publishes into, walks through what each does with a bad submission, and reviews the exploits at bZx, Harvest, Cream and Mango that shaped oracle design. The angle is the publisher's: the network can remove one bad submission, but it cannot improve what a publisher sends in.

Crypto Reference Rates and Benchmarks: What Makes a Price a Reference Rate

A US spot Bitcoin ETF does not value itself off an exchange price. It uses a reference rate: one figure, computed across several exchanges over a fixed window by a published method, from an administrator regulated for the purpose. This article explains what separates a price from a reference rate and a reference rate from a regulated benchmark, walks through how the CME CF Bitcoin Reference Rate is calculated, sets out the range of rates from once-daily settlement to continuous, and names the five properties every reference rate needs.

Fair Market Value in Crypto: Choosing the Right Price for Tax

Every taxable crypto event needs a value in local currency at the moment it happened. The usual advice is to pick one exchange and use it consistently. That meets the consistency test and skips the harder one: whether the exchange was ever representative of the market. This article sets out what the four major tax authorities actually require, where a fixed exchange produces a consistently wrong number, what a reasonable method looks like, and how a defined exception keeps a method consistent without locking it to a bad price.

Net Asset Value in Crypto: What a Defensible NAV Actually Requires

A traditional fund strikes NAV against a closing bell. A crypto fund has no bell, no official price and no single market, so every published NAV rests on decisions most investors never see: when to value, from which source, how to handle a token that barely trades, and who owns the policy. This article walks through all four, shows where NAV calculations fail in practice, and sets out what a NAV needs to survive an audit, an allocator's due diligence or a dispute.

Crypto Prices Explained: VWAP, Bid/Ask, Mid, Mark, Index and State Prices

Ask what Bitcoin costs and you get a number. Ask which number and the conversation gets interesting: last trade, best bid, best ask, mid, VWAP, and for perpetuals a mark and an index price, plus a state price on decentralised exchanges that needs no trade at all. This article follows one Bitcoin through every one of them at the same moment, shows how each is calculated, explains why positions are liquidated on mark price and not the chart, and ends with a decision table for which price to use for what.

Execution Quality in Crypto: How to Know You Got a Fair Price

Every firm that executes in crypto should be able to answer one question after every trade: was that a fair price? Most can't, because there is no consolidated tape and no official reference to measure a fill against. This article covers what the MiFID II and FINRA definitions actually require, why measuring against your own liquidity provider proves nothing, how pre-trade and post-trade benchmarks work, why the best displayed price is not the executable price, and where real-time market data changes the answer.

Rethinking Crypto Data: Single-Venue Prices vs Market-Wide Fair Value

Institutions arriving in crypto bring an assumption from equities: that speed is what separates good market data from bad. In a market with no consolidated tape and no central authority, that assumption fails. This study explains why integrity has to be engineered through aggregation, shows how far even the deepest exchanges deviate from market-wide fair value, and sets out what oracle-grade validation means for anyone valuing, hedging or settling against crypto prices.

Market-Neutral Staking Is Changing: How Oracles Improve the Yield–Risk Profile for Liquidity Providers

Crypto Market Data: Why Oracles Are the Unseen Backbone of DeFi

Discover how blockchain oracles secure DeFi by delivering reliable market data for lending, derivatives, and asset tokenization. A comparison of Chainlink, Pyth, RedStone, Supra, and SEDA.

DoubleZero and Jito – Behind Solana’s Latency and Throughput Improvements

What is DoubleZero? Discover how Solana’s new fiber backbone tackles its biggest bottleneck — and what it means for speed, reliability, and staking.

Solana’s Next Leap in Performance - Firedancer

Discover Firedancer, Solana’s new high-performance validator client. Learn how its modular architecture, parallel processing, and security improvements reshape reliability and staking efficiency.

Solana Staking Reliability: Past Challenges and Present Stability for Institutional Stakers

Solana has shifted from repeated outages in 2021–22 to over 16 months of uninterrupted uptime. Reliability is now crucial for institutional staking, with validator choice directly impacting yield and risk, and operational maturity proving as important as raw performance.

How to Stake Solana (SOL) Safely in 2025: The Complete Guide

Staking Solana (SOL) in 2025 is easier than ever - but choosing the right method is critical to protecting your tokens and maximizing rewards.