[JumpCrisscross]

The people choosing which coins to include and how to weight them have tremendous power. They, and their buddies, will be tempted to (a) re-constitute the index to favour assets they own or (b) buy and sell ahead of re-constitution using insider information. “Painting the tape” is a problem with proper indices; here, someone on the GDAX side could conceivably just mess with the records.

[e12e]

> someone on the GDAX side could conceivably just mess with the records.

... On the blockchain?

I suppose you mean the pointers to which users controls what, in the internal db? Ie: straight up fraud. I'm not convinced having access to a large order lists doesn't already open the door to insider trading?

[JumpCrisscross]

> ... On the blockchain?

The index is market capitalization weighted. "The market capitalization of each constituent asset is calculated as the price of the asset multiplied by the supply of the asset" [1], where the "price for each constituent asset is the last trade price on the GDAX USD order book" (2.6).

The market capitalization, and thus weighting schema, is an entirely internal product of Coinbase's data.

> I'm not convinced having access to a large order lists doesn't already open the door to insider trading?

Currently, an insider could (a) give their orders better execution than the market or (b) foment a spike/crash by "painting the tape". This is risky and leaves a paper trail.

"The Coinbase Index Committee consists of one member representing Coinbase, and two unaffiliated independent members who have experience in index creation and oversight" (7.4). Those people are powerful. They can take economically-significant actions based on their subjective determinations. Convincing one of them to (a) tell you how they're rebalancing or what assets they're adding/removing when or (b) rebalance in a way that helps your portfolio can be done entirely over a glass of beer. Still risky and illegal. But less detectable than before.

This structure of incentives (small group of subjective decision makers operating on the basis of internal data) historically fails. (See the Libor scandals [2], where we only nailed those involved because they coördinated over instant messages.) Even if the first batch of three are honest, all it takes is one bad apple to spoil the bunch.

[1] https://am.coinbase.com/documents/cbi-methodology.pdf § 2.4

[2] https://en.wikipedia.org/wiki/Libor_scandal