xRapid - Separating Signal from Background

hmatejx

hmatejx

Read more posts by this author.

Read More

Story originally posted on Medium.

As fans of XRP know, xRapid is Ripple’s solution for on-demand liquidity, aimed at payment providers and financial institutions dealing with cross-border money transfer. There is a lot of controversy surrounding whether anyone is actually using xRapid in a commercial setting and what the transaction volumes look like.

Utilization of XRP by xRapid is currently considered as one of the key factors of XRP’s long-term value proposition. It is therefore interesting to explore the public ledger history for signs of xRapid activity with the eventual goal to quantify this activity and monitor its evolution.

Note: information and conclusions below might be wrong or incomplete, limited by the author’s knowledge. This is the author's first attempt "measure" xRapid. Please reach out if you would like to contribute to the search.

Summary of what is known

Below follows a brief recap of what is known about the functionality of xRapid and it’s current state of adoption.

a) The xRapid process

As explained by Asheesh Birla in a recent YouTube video,
maxresdefault
an xRapid settlement consists of the following steps:

  1. deposit of fiat to an exchange and purchase of XRP,
  2. transfer of XRP to another exchange, and
  3. selling of XRP and withdrawal of fiat from the 2nd exchange.

Only the second step involves an XRP transaction.

b) The xRapid ecosystem

As evident by the recent flood of news, the xRapid ecosystem is rapidly expanding. Below is the list of known xRapid partners (exchanges and customers).

Existing Testing or implementing
Exchanges Bitstamp, Bittrex, Bitso, Coins.ph SBI VC, UAE Exchange, Binance
Customers Cuallix, Mercury FX, IDT, Zip Remit, Sadad Bahrain Catalyst Corporate Federal Credit Union, MoneyGram, Western Union, Viamericas, Cambridge Global Payments, Currencies Direct, Kuwait Financial House

One glance at the impressive collection of XRP/FIAT pairings below shows that the list shown above is not comprehensive. Therefore, the tentative xRapid volume determination shown later down below probably underestimates the true overall volume.

c) A sample xRapid transaction

Ripple’s David Schwartz has posted a live video of a demo xRapid transaction.
xRapid_demo
He later also provided the corresponding transaction hash.

Inspecting this particular XRP transaction between Bitstamp and Bitso reveals the destination tag of the receiving account: 25490962 (keep this number in mind for later).

Methodology

It is difficult, if not impossible, to distinguish xRapid transactions from ordinary Payment transactions on the XRPL. I briefly describe my attempt in the following subsections.

a) Simplifying assumptions

Some reasonable simplifying assumptions need to be made in order to detect a signal within the noisy background.

Transaction characteristics

  • It is reasonable to expect that xRapid customers will have their own separate accounts setup at partner exchanges.
  • Additionally, as there is no obvious benefit of having multiple accounts simultaneously, all xRapid transactions initiated by a customer to a destination will share the destination tag!
  • It is also likely that all xRapid transactions from a customer in a given corridor will be for comparable amount of XRP.

Scope restriction

  • I limit my current search to uni-directional transactions, as remittance flows are highly directional. While it is true that marker makers will eventually send the XRP back to the originating exchange, this process will probably follow a different pattern.
  • I analyze the potential remittance traffic to Mexico and Philippines as these were the most promoted corridors. Moreover, I don't have the required data to consider other corridors at the moment.

b) Technical details

The technical details of are collected here.

Known exchange addresses

With the help of Bithomp (and Google), I was able to find these:

Querying the XRP ledger

Thanks to the relentless work of @WietseWind, it is now possible to query the XRP ledger using Google’s Big Query platform. I've made two queries of the XRPL:

  • The first query obtains all non-zero successful XRP payment transactions between the 4 known xRapid partner exchanges (excluding payments from the an exchange to itself).
  • The second query adds the “leg” field (directional pair of source and destination) and calculates the difference in time and amount for a given transaction and its closest preceding or following transaction (within the same destination tag and leg subset).

Analyzing the transactions

I visualized the transaction data using Tableau Desktop public edition.

Results

The results are summarized below step-by-step.

Figure 1 shows the overall daily activity of inter-exchange payment transactions from 2017 onward (before that time there is scarcely any inter-exchange traffic). The transaction count is shown in yellow, and the total XRP amount per day in blue. It is obvious that the 2018- bear market has had an impact.
1_weC5nAWwNvPi10N7MhgOiA
Figure 1. Daily total amount and count of inter-exchange payment transactions since 2017.

It is insightful to compare the inter-exchange activity with the overall ledger payment activity, as shown in Figure 2. The proportion of inter-exchange transactions, in comparison to total number of payment transactions, ranges from ~ 0.5% (recently) to ~ 2% (in the past). The recent trend of increasing number of all payment transactions , which is not reflected proportionally in the inter-exchange traffic, can be interpreted as a possible sign of XRP adoption among users, as well as for other use cases. Or it could just be increasing spam.
1_q7yvbTnQtugfykCsXFHphw
Figure 2. Daily count of XRP payment (from xrpcharts).

In order to find interesting patterns, each transaction is depicted individually in the scatter plot on Figure 3. Transactions are colored by the directional exchange pair. Note the peculiar “banding” pattern of Bitstamp ⇨ Bittrex (bright green) and Bittrex ⇨ Bitstamp (purple) transactions. I have no clue what to make of it… arbitrage or market making?
1_btUR07G1j9u0btlktrDXdQ
Figure 3. All inter-exchange payment transactions since 2017, colored by the (directional) exchange pair. Note: Vertical axis is shown in log scale.

The situation becomes more interesting on Figure 4, showing the inter-exchange traffic in the direction of the promoted xRapid corridors (⇨MXN, ⇨PHP). Overlaid on the plot are three vertical lines representing important dates for xRapid.

  • The gray line shows approximately the date when Cuallix was announced as the first institution to use xRapid.
  • The green line marks the date when three additional xRapid partner exchanges were announced.
  • The yellow line marks the date when xRapid was declared to be commercially available.

1_l6O7O9HY1Cvg1KNGO49xUg
Figure 4. All inter-exchange payment transactions for Bitstamp ⇨ Bitso, Bitstamp ⇨Coins.ph, Bittrex ⇨Bitso, and Bittrex ⇨Coins.ph since 2017, colored by the destination tag. Note: Vertical axis is shown in log scale.

Interesting observations:

  • It is surprising to see that the 25490962 destination tag for Bitstamp ⇨ Bitso became highly active shortly after the announcement of Cuallix (remember, this tag was also used in the live xRapid demo transaction). Moreover, there appears to be a period of testing by the same entity shortly prior to that announcement (the half-pink/half-gray worm-like cluster of transactions in the upper pane). What is really strange about this tag is that the typical amount per transaction is rather small, ~25–50 XRP.
  • Another interesting tag in Bitstamp ⇨ Bitso is 30106572, which also became active shortly after the Cuallix announcement. This tag transfers larger amounts of XRP, most often exactly 20k per transaction.
  • Likewise interesting are destination tags 5 and 967 in the Bitstamp ⇨ Coins.ph leg. Tag 5 became active almost immediately after Coins.ph was announced as an xRapid partner, and 967 shortly after xRapid was commercially available. Destination tag 5 is up to date the most active entity, transferring up to 5k XRP per transaction. There appears to be a ceiling configured at 5k since December 2018.

Inspired by these findings, I’ve made Figure 5 showing the weekly count of transactions for the top 5 most active destination tags (by total count) for each directed exchange pair (the vertical axis is shown in log scale). I find it perplexing that I cannot find any interesting traffic in the Bittrex ⇨ Bitso leg that would fit what one would expect of xRapid (could it be that Bittrex is not yet utilized by xRapid customers?). In the Bitstamp ⇨ Bitso and Bitstamp ⇨ Coins.ph legs, we find the same tags as already identified before.

1_p-wvPOKcWaXeKTRpmHaSWQ
Figure 5. Top 5 most active Destination Tags per directional exchange pair. Shown is the weekly TX count in log scale.

From the above analysis, only destination tags 25490962, 30106572, 5, 967 match the xRapid expectations. The monthly XRP volume for these tags is shown in Figure 6.
1_QN2B9V0dTnlTGf5VTl6fng
Figure 6. Tentative xRapid traffic towards Mexico and Philippines. Y axis shows the monthly total XRP amount. Note that the Feb 2019 is biased due to partial data.

Of these, only tags 30106572 and 5 have any appreciable volume, transferring ~ 1M XRP per month. The other two are only transferring thousands per month. The total monthly volume of all 4 identified tags is finally shown in Figure 7. Behold, the tentative xRapid signal!
1_J4bpbd5j1Y7sVt7NLhlf-g
Figure 7. Tentative xRapid traffic towards Mexico and Philippines (combined). Y axis shows the monthly total XRP amount. Note that the Feb 2019 is biased due to partial data.

Conclusions

The above analysis reveals a rather modest amount of monthly traffic that can potentially be attributed to xRapid. It will be very interesting to follow its development throughout 2019 as xRapid utilization intensifies and branches into additional corridors.

However, to cover these additional corridors, more information about the additional partner exchanges will be required.

Crawl, walk, run ;-)


Did you like this post by hmatejx?

Send some love: