Circulating Supply Study for XRP

This was originally sent to subscribers of Messari’s Unqualified Opinions. Click here to become a subscriber and access all of our historical content.

You can view the original post here

Florent Moulin — March 19, 2019

Executive Summary

In January, we found material discrepancies between what Ripple reported as “distributed supply” of XRP, and what can be fairly considered “circulating supply” a metric we define as liquid supply (no contractual or programmatic restrictions) that is held outside of the company, affiliates, and creators and/or founders of a given cryptoasset. This report is a confirmation of many of our January findings.

Key Findings:

1. Ripple’s quarterly Transparency Reports and the Ripple Data API rely upon inconsistent and vague methodologies. They may mislead investors because key information regarding sales and and estimated trading volumes significantly understate the pace of selling pressure from Ripple and its founders and affiliates. We find that XRP’s circulating supply inflated by 11.5% in the past six months vs. the 4.5% implied by its API.
2. Quarterly XRP sales as a percentage of global trading volume may be significantly understated due to Ripple’s inclusion of fake volumes from exchanges known to facilitate extensive wash trading or fee-less trading. As a result, we have closely followed Ripple’s preferred exchanges for asset liquidation, and note that many of these exchanges have been excluded from Ripple’s own data API. The company appears to rely on CoinMarketCap’s headline volumes figure rather than the volumes from the exchanges upon which it currently trades.
3. Circulating supply data on Messari will soon be made available for all major assets. In the interim period, we have not changed our liquid supply assumptions for XRP.
4. We’ve been working closely with major indices as well as passive funds to provide them with more accurate data on supply & trading volumes. We recommend others review this report and reach out to us with questions regarding our methodology.

XRP Sales — A Historical Background

Last Wednesday, Ripple published its “Q1 2019 XRP Markets Report.” These transparency reports have been published quarterly by Ripple since Q4 2016. The company’s goal is to “share regular updates on the state of the market including quarterly sales, commentary on previous quarter price movement and announcements of new third-party wallets, exchanges, validators and third party liquidity providers.”

There are two types of XRP sales the company discloses:

Programmatic Sales: These come from the company’s “available-for-sale” XRP reserve, and can be sold at a rate of no more than “0.25% of average daily volume on defined exchanges.”

Direct sales from XRP II: These come from direct bespoke sales to institutional clients, and include (per Ripple’s 2016–2017 disclosures) reselling restrictions to ensure partners do not purchase XRP at a discount from the company and then immediately market sell on exchanges.

In Q1 2019, Ripple made $169 million from these combined sales, with two thirds of that sold on defined exchanges and the rest coming from direct institutional sales.

Since Q4 2016, over $890 million worth of XRP has been sold by Ripple. As you can see, Q1 2019 was also the highest revenue quarter ever from sales of XRP.

The company’s Q1 2019 reported sales represented 0.32% of the global XRP exchange trading volume from the quarter, up from 0.24% in Q4 2018, according to data from CoinMarketCap.

We believe these published figures tell an incomplete story, however. Investors may draw incorrect conclusions from the disclosures as we have found several material issues with this existing data:

• Ripple communicates XRP sales only in USD terms, limiting visibility into XRP inflation. We have limited visibility into the actual amount of XRP sold via direct XRP II sales because we do not know how heavily the company discounts encumbered distributions. This is important, as we have calculated implied discount rates for ongoing strategic XRP sales.
• Ripple uses inaccurate volume data from CoinMarketCap to calculate the % of quarterly volume its sales represent. It has been widely understood that reported volume metrics on CoinMarketCap are inflated due to their inclusion of overseas wash trading and zero-fee trading volumes. As a consequence, the “total sales as a % of total volume” presented by Ripple can be misleading. The percentage would be much higher if known fake volumes were removed from the total.
• Ripple only communicates what is sold through XRP II & Programmable sales. In our January XRP report, we uncovered that over 40% of XRP reported circulating supply was not in fact, circulating. This included holdings from co-founder Chris Larsen (5.9B XRP), co-founder Jed Mcaleb’ (6.7B), and the affiliated Ripple Foundation for Financial Innovation (2.5B). All of these holdings contain daily selling restrictions “based on a percentage of the previous 24 hours total trading volume on designated exchanges.” We believe this targeted figure is 1% of daily trading volume, but are unclear as to which designated exchanges are defined by the company. Quarterly founder and affiliate distributions and sales are not disclosed in Ripple’s Transparency Reports.

These methodological omissions make it difficult for investors to properly estimate the correct amount of XRP that has entered circulation in any given quarter. We believe we can take the first step towards understanding true circulating supply with our revised methodology.

Liquid vs. Circulating Supply

We based many of the findings from our January report on Ripple estimated present and future “Liquid Supply” on known selling restrictions and historical Founders’ agreements. Our subsequent Cryptoasset Supply Methods defined Liquid Supply as the number of units that currently exist on-chain and which are not known to be encumbered by any contracts, programmatic, legal, or otherwise.

Another important supply tranche in our classification schema is Circulating Supply, which excludes illiquid and liquid tokens held by asset creators and their affiliates, including founders. This is analogous to the “free-float” method of calculating public equity shares outstanding. Liquid Supply available for sale by asset creators and their founders and affiliates are not included in our Circulating Supply calculations until they are sold or distributed to third parties.

Due to the centralized nature of Ripple’s XRP distribution and the size of Ripple’s XRP treasury, we believe a study of both Liquid and Circulating Supply is required to get a sense of ongoing XRP distribution rates and the implied dilution that causes. Indeed, Liquid Supply gives us an indication on what CAN be distributed while circulating supply tells us what IS effectively distributed.

Estimating Circulating Supply is difficult. It requires identifying all Ripple-affiliated treasury wallets and founder accounts. While this can be done In theory through ledger forensics, accuracy depends on a project’s willingness to be transparent and share key wallet addresses with their communities. (This is what our registry members are currently doing with Messari.) Less precise, but directionally correct estimates can also be derived by reading, analysing, and interpreting on-chain transactions.

In this report, we use our supply and volume methodologies combined with on-chain analysis to better estimate how much XRP has entered circulation in the past six months.

We also cover how this newly circulating supply has been distributed and why the use of CoinMarketCap volume data by Ripple is inconsistent relative to its own reporting and distribution practices.

Our overarching goal is to provide a comprehensive view on the current XRP Circulating supply in order to more accurately calculate the true circulating market cap of XRP.

I) XRP Ledger analysis

Our ledger analysis starts with the identifiable “Genesis” addresses at Block 32,570 in the XRP Ledger. Transaction data before block 32,570 cannot be reconstructed as CTO David Scwartz detailed “In January of 2013, a bug in the Ripple server caused ledger headers to be lost. All data from all running Ripple servers was collected, but it was insufficient to construct the ledgers. The raw transactions still survive, mixed with other transactions and with no information about which transaction went in which ledger”.

Starting with Genesis addresses allows us to identify most Ripple-affiliated wallet addresses, including founders & foundation wallets that have known selling restrictions, and liquidation wallets that are not officially bookmarked as Ripple wallets, but that are exclusively used by Ripple to transfer funds to exchanges / other known addresses.

Based on our analysis, we believe that more than 75% (75 million) of the current XRP outstanding supply (units identifiable on-chain) is currently managed by Ripple and its founders and affiliates. Circulating supply data is sourced from known on-chain addresses.

This may not include all founders’ and company treasury supply, as we may not have been able to identify all of the existing liquidation accounts. The value presented is our best estimate for the Maximum circulating supply of XRP. The actual circulating supply could be significantly lower (as relative trading volumes imply). Regardless, we will continue to review our data based on observable on-chain movements.

We have classified these known addresses in order to have a better understanding of where Ripple funds are currently located and where they may move in the future. We currently identify the following categories of XRP supply.

Supply Controlled by Ripple: Treasury holdings and tokens in escrow, including Foundation and Founders’ commitments.

Hot Liquidation: Supply that is located on addresses used to distribute the tokens to external addresses. These addresses usually transfer the funds very shortly after they received it.

Cold Liquidation: Supply that is located on addresses used to distribute the tokens to external addresses. These addresses usually transfer funds slowly. Some may be encumbered by selling restrictions.

Cold: Supply that is located on addresses that do not transfer funds to external addresses. The funds generally goes through a Hot or Cold liquidation address before being transferred.

Escrow: Supply that is currently located on addresses that have escrowed most of their funds.

Re-escrow: Supply that is currently located on addresses that have re-escrowed funds from initial unlocked escrow addresses.

Inactive: Supply located on addresses that been inactive since 2017, or that have never been used.

In total we’ve been able to identify 80+ addresses controlled by Ripple and its founders and affiliates. Here is a visual breakdown of how the 75M held by Ripple are managed, using the above categorization.

The next chart displays how many addresses are holding funds from:

• The Foundation & Chris Larsen: Chris Larsen contributed 5.9 billion XRP to the foundation and some funds are held in the same wallet.
• Jed Mccaleb: Ripple appears to have retaken control of Jed’s funds after the company and co-founder reached a settlement to restrict ongoing sales to 1% of average daily trading volume.
• Ripple: Known treasury addresses
• Unknown Genesis addresses: addresses holding funds since 2013. These 2 Genesis addresses each hold 1 billion XRP, and are believed to be tied to Chris, Jed, or co-founder Arthur Britto.
• Liquidation: Addresses exclusively used to send funds to exchanges from the addresses above.

Identifying these addresses is key to estimating how (to which address / exchange) and when Ripple, the Foundation (including Chris committed funds) & Jed’s XRP are effectively entering circulation.

II) The XRP supply has inflated at a much higher rate than what can be explained from the Ripple Transparency reports and Ripple’s API in the last 6 months

First let’s look at the Ripple Data API v2 (“the API”), which provided users with information on “outstanding”, “distributed” & “escrowed” supply. The API acts as a data source for applications such as XRP Charts and ripple.com. It allows anyone to access information about changes in the XRP Ledger, including transaction history, and currently powers data aggregators such as CoinMarketCap.

For this section we will focus on the “Get XRP Distribution” request for the API.

According to the API, 1.042 billion XRP were distributed in Q4 2018 & 751 million in Q1 2019, inflating the overall “distributed” supply by 4.5% in 6 months (from 40 to 41.8 billion).

(Note: there is one week, December 9th, where circulating supply decreases according to the API. Interestingly, this is not due to the company’s re-escrow policies. We do not don’t know if this is due to a token buyback or an actual bug from the API. Ripple representatives have not responded to repeated requests for clarifications and further information.)

Ripple considers XRP distributed to the Ripple Foundation for Financial Innovation (including Chris’s commitments), as well as Jed McCaleb’s contractually encumbered XRP as “distributed”, so liquidations from these accounts do not impact the figured delivered from the Company’s API.

We do attempt to use the data provided by the Ripple API, coupled with revenue in USD published in Ripple’s quarterly Transparency Reports, to better estimate the total XRP sold through the company’s Direct Sales program through XRP II, and any implied discounts on these sales.

Since funds controlled by Ripple are subject to selling restrictions based on reported volume, we can use the VWAP (Volume Weighted Average Price) of each quarter to approximate how many XRP have been sold through Programmatic Sales.

We assume every XRP released by Ripple enters circulation through Programmatic Sales or Direct Sales. As such, we can also calculate the average XRP sale price for Direct Sales by dividing reported Direct Sales revenue from Ripple’s Transparency Reports by the difference between the programmatically sold XRP estimate and the total XRP distribution returned via Ripple’s API. We then compare this result with the quarterly VWAP to estimate the implied discount of Direct Sales.

While the implied discount of Q1’19 is conceivable, an average 88% discount in Q4’18 seems high.

We believe this could be indicative of a materially large XRP distribution to a key strategic partner. For instance, this distribution and discount could coincides with the company’s legal settlement with partner R3, which had agreed in 2016 to exercise a purchase option on XRP at a price well below current spot.

Ripple does not disclose details surrounding these sales, nor does it explain how, when, or why XRP sold through Direct Sales are sold at a discount to market prices. This has led some people to erroneously calculate how many XRP have been distributed each quarter without comparing the implied results from the company’s Transparency reports to those viewable via Ripple’s API.

In fact, the only way to know how much additional XRP has been distributed each quarter is to conduct Ledger analyses of our 80+ identified addresses.

XRP Ledger Analysis

As expected, Ripple’s “Hot Liquidation” addresses receive most funds before they are ever sent to exchanges. One in particular has received XRP from both Ripple, its affiliated Foundation and Chris Larsen. We traced all funds transferred out of Ripple managed addresses in order to find out:

1. Is this the same as what is reported by Ripple’s API?
2. Where does this newly circulating supply come from?
3. How and where is this supply distributed?

As shown in the chart below, it appears more supply is entering circulation than what is displayed by Ripple’s own API: 73% more in Q1’19 & 23% more in Q4’18.

The distribution of XRP entering circulation according to our wallet analysis and classification suggests that over 50% of the Q1’19 difference between the API and what can be observed on-chain can be explained by Ripple’s stance that the holdings of Chris Larsen, the Ripple Foundation and Jed McCaleb are already “distributed” despite the fact that Ripple maintains some control over this supply via their contractually-imposed selling restrictions. This discrepancy in methodologies significantly impacts the implied inflation rate of XRP’s monetary base.

Our more comprehensive on-chain analysis shows that real inflation is likely to be 11.6%, compared to 4.5% reported by Ripple’s API.

How and where is this supply being distributed? And is this consistent with Ripple’s limitations on programmatic and direct selling?

III) Ripple trading volume estimates in their Transparency Reports are inconsistent with their own API, leading to material understatements related to the rate of programmatic selling.

We break XRP distribution down into three categories:

1. Transfers to exchanges. These transfers occur through a liquidation address or directly from the original tracked account. We have limited visibility on whether funds are sent directly to these exchanges or are re-distributed to other addresses / exchanges. We assume exchange transfers enter circulating supply because we have not been able to identify many incoming XRP flows from exchanges to Ripple-owned addresses.
2. Transferred to an external address or an exchange through a custodian. These transfers are completed by transferring funds to addresses activated and most likely managed by BitGo. Some of this supply is sent to exchanges after a few transfers in a short period of time but we do not consider it as liquidated on exchanges directly by Ripple since the receiver could be a third-party such as an institutional client.
3. Transferred to unknown addresses. These transfers are made to addresses not identified as managed by Ripple or by a custodial agent of Ripple, but are ultimately transferred to an exchange. These could still be internal accounts, though, such as that of Xspring, a Ripple-financed incubator.

In Q1’19 & Q4’18,, addresses managed by Ripple mainly sent funds to exchanges. Some 66% and 77% of the quarterly distributed supply, respectively. Those numbers include affiliated and founder addresses, which are subject to the same selling restrictions as Ripple. Which exchanges tend to receive funds? Here is a complete list:

In Q1’19 Ripple and its affiliates sent a significantly higher percentage of its distributed XRP to BW.COM, an exchange currently excluded from the RIpple API, and estimated to facilitate extensive wash trading (and inflate volumes). Other suspect exchanges that received Ripple-affiliated funds for the first time included Bitforex and Digifinex also excluded from Ripple’s market API.

Is this problematic?

Funds managed by Ripple (75% of the outstanding supply) are subject to selling restrictions based on a % of average daily volume on defined exchanges. Since Ripple uses CoinMarketCap volume for its Transparency reports, and since some of the exchanges upon which it sells XRP are not included in Ripple’s API, there is an inconsistency in terms of which exchanges are used as reference points for the company’s structured selling efforts.

Bitwise previously reported (and we confirmed) that CoinMarketCap trading volume data is largely overstated due to wash-trading practices on large international exchanges. The Bitwise report was focused on BTC but we believe its findings applied to many other trading pairs as well.

The TIE for example, used the weighted average trading volume per web visit for Binance, Coinbase Pro, Gemini, Poloniex, and Kraken as a baseline volume per user to calculate expected volume for various alternative exchanges. Using these expected volume figures, the company calculated the ratio of Expect Volume / Reported Volume (by CMC) to estimate potential fake volume on 100 exchanges.

BW.COM was among the 25 most suspect exchanges in terms of fake trading volume. These reports were released in Q1’19 at the same time that Ripple and its affiliates began sending a much larger portion of its funds to new and less reliable exchanges. While The Tie methodology is vague and insufficient for determining accurately what the exact impact of fake trading volumes vs. reported trading volume for a given exchange, we believe they are directionally accurate.

We are working to expand our “Real 10” benchmark to account for exchanges with high legitimate volumes even when there is evidence of large amounts of wash trading. The Tie estimates corroborates our own preliminary volume assessments of non-real 10 exchanges.

The most problematic finding seems to be that Ripple has increased its % of volume-restricted XRP sales on exchanges that have been otherwise identified as having a high proportion of fake volume and the reason for this is unclear.

Why we care

Other crypto projects retain treasury tokens that they sell to fund ongoing operations. None come close to Ripple in terms of the size of the positions they intend to unwind over time.

While we appreciate the steps Ripple has taken to be more transparent with their sales programs, the real numbers that impact investors, inflation and implied institutional discounts, are still not disclosed. Instead, the retail investing public is left guessing as to the actual dilution generated by incremental XRP sales. As a data provider we take these issues to heart, as our accuracy can have real world implications for investors. For now, we will continue to conduct deep research on Ripple and its relationship to the XRP Ledger, as well as other blockchains, to bring you our best estimates for these numbers.