JASCoder

The extra effort you have expended to provide the community such quality content is greatly appreciated. I know how that last 5% of polish can often consume a disproportionate amount of effort ! Well done John. Our respective research is complementary, we could certainly gain some synergy and avoid redundant effort by collaborating :) Most of your shared research and analysis are constant with my own observations and conclusions, so I will respond to your content on the points left unaddressed, or at odds with my own... You have have focused on the ledger activity, which I found both fascinating and supportive of my own guesses and observations. My efforts have been more narrow in focus, monitoring the behavior and drilling down into measuring the cost slippage computed from the dual order books of the BS/Bitso exchanges. I will address the following points : --- ODL's apparent behavior at the more granular level Volatility and its impact on the efficacy of ODL processes My current theory on ODL's operational design TRAFFIC PATTERNS When you observe the XRP sales at Bitso, in a ten-minute granularity, it appears relatively uniform through out the hours of the day, except for a precipitous drop to near-zero on Saturday morning, but then resumes on late Sunday (generally). There is, however, the occasional spikes and drops which deviate from the established pattern. I have been able to correlate some of these events with apparent arbitrage opportunities for profit, or excessive cost slippage periods. My efforts to improve the accuracy and precision for measuring the cost profile has been currently a bit stymied by stale and inaccurate FX rate data (which I hope to soon have resolved!). VOLATILITY IMPACT In the past have posted here on XRP Chat some captured charts and conveyed analysis of a few anomalous events of valuation spikes and drops, as well as occasions of passive observations. In one instance, the steady process of XRP sells at Bitso didn't even reflect any apparent impact from a sudden dumping of XRP and valuation drop at Bitstamp. This was actually surprising, but viewing the slippage metrics showed a reduced cost friction, so it's not really surprising in that context. I have not seen any real evidence that volatility causes ODL to be less effective, or at least to show any discernible impact. But we should do more research into this, which I'm hoping to do in the future as well. MY THEORY OF ODL OPERATION What we cannot know: the depth of reserve fiat capital held ready on-account at any ODL corridor's two exchanges. That of course, is managed by the market Takers and Makers. It might be wisest to assume such data is also unknown by Ripple's ODL operations (to keep our analysis simplified). 1. The buy/sell trade events are not directly driven by remittance customers. It makes zero sense it could be the case, given the uniform pattern of trading volume we can monitor from the exchanges' own exposed market data. As you surmised John, there must be some mechanism for value pooling of a "nostro/vostro style" performed by the managing ODL processes which Ripple operates. 2. It is very possible the ODL does act upon any detected arbitrage opportunities, to both acquire additional XRP or sell down some surplus on occasions when friction is extra low or into a profit range. 3. ODL's capacity to leverage the dual markets for moving cross-border value is constrained by both the depth and re-balancing latency of the market makers. So perhaps the ODL processes are monitoring in realtime the ever fluctuating slippage profile, and they meter their transactions accordingly. They are incentivized to not over-run the market's capacity of course, to avoid exceeding their "pain threshold" on costs (my guess that is constrained to a maximum tolerance of about 1.2 percent slippage cost) - and they don't want to be too low with their automated trading and discourage the market makers from devoting their capital resources to Ripple's ODL corridors. Cheers friend ! ( Please do DM me here John, we should compare notes moving forward )

Happy to share interests friend Yes indeed, the Bitso XRP sells volume now runs at a mean of about 1.7m an hour, wasn't long ago it was between 500-800 k This could be a high-volume week. Just a few hours ago we saw 5.4m XRP sold in one hour at Bitso.

As a long time poker player (and fan) myself, I'd be remiss to not share this humorius piece on Bilzerian from Doug Polk. heh

Here is another interesting event - one well-timed large XRP buy at BS, an apparently associated sudden sell of XRP at Bitso. In the (black) BS chart, you see a long stable period on this quiet Sunday (ODL wise). Then a punch down, a slow retrenchment up, then a spike buy for ODL use (see spike in Bitso volume below). So if you were designing a system like ODL to buy/transfer/sell XRP in a steady stream of programmed trading, what kind of algorithm and automated processes would be most cost effective ? Bitstamp Buy event (see green volume spike of 2.4m) : Bitso XRP Sell event (same time period) :

Hey John, thanks so much for putting the time and energy into sharing your own research, analysis and theories. I plan to fully read, digest and respond (here) to your blog entry. I had just hopped on here to post an intriguing event with ODL into the "Bitso thread," then return to my coding (so close!). Perhaps you'll find it interesting too - as IMHO it's yet another clue to how ODL's backend is programmed (which I've been trying to figure out).

From what Tiny (who's gone now?!!) had shared with me in past exchanges, he said from his own research he found that the trade actions we see occurring at the ODL corridor endpoints are not driven directly by customer transfers, but the paid in fiat submissions are pooled and metered by Ripple's processes to be buy/transfer/sell payloads, handled in a roughly constant flow over a 24 hour period; this is to prevent out-running the arbitrage re-balancing dynamics between the endpoints and realize excessive friction we assume. Our best guess is that Ripple is maintaining their own fiat capital pools at the pay-out endpoints, which acts as a buffer - much like an electric capacitor does. The hourly volume metrics we see seem to bear this out - as you certainly would not expect remittance customers to maintain a 24 hour period of same sized transfers on weekdays. I hope I'm making sense. You can glean more details on this from our past posts here in this thread IIRC (tho his name will now show as "Guest"). Cheers.

Large ODL event a few hours ago... In the hour of 17:00 on both charts, you see the XRP buys at Bitstamp (just two volume tics over 3.5m / $1.3m) and the Bitso selling volume of 4.3m - note is slight price upwards tic at BS, but the pricing at Bitso showed no perceptible nudge. As you can see, the hourly sell volumes at Bitso have been fairly relentless the last 36 hours... Bitstamp: Bitso:

@Dogowner5 Thanks for chiming in friend - you appear both astute and informed @Frisia - The slippage "friction" is not the total cost for Ripple's ODL service, but it is both significant and a bit variable, at least these days. If you think about it, the fluctuating dynamics of the FX rates must certainly influence the balancing actions of the network's arbitrage processes. I've come to suspect these processes are leveraging some centralized or authoritative aggregator of of the FX metrics. I'm hoping IG's feed will provide this need...

The other item from Bitstamp is they are establishing a totally-USA based service, much like Binance did. This will really sped up the latency for fiat transfers.

Howdy thar Xrpiet - thanks for your interest in my ODL research ! You can see from the two charts below (black is price/volume action at Bitstamp over the last few hours, and the blue is the raw xrp hourly sells volume at Bitso), you can clearly see at hour 15 a flurry of heavy buying at Bitstamp, and corresponding selling action at Bitso. Viewing the Bitstamp pricing you see a small upwards bump as well, and interestingly, at Bitso, there was no perceptible pricing impact. Over the last many "crisis weeks" I've continued to monitor this ODL corridor, and it's been heartening to note that the steady usage of ODL continues to strengthen. JFYI, I continue developing more "scanner" tools, in particular for slippage metrics. But I ran into a frustrating discovery, in that the FX rates play more of a significant impact on the slippage pricing than I'd realized. I'm pursuing better sources for more a granular and accurate API source (IG is really good, but they won't give me access to their APIs because I'm not in the UK. They are working on providing that for USA "soon" I'm told). Oh, and the Visa-dev's solution want $3000 to set me up. :-/ I also pursued adding the coins.ph order book to my charts, but they won't give me an API key until I have renewed by passport (a CB requirement). Sigh.

It will be ugly. IIRC both student loans and car loans are both over a trillion USD - each.

Given the historical trading today, for thoroughness, and analysis, here are some of the metrics... The last two months of XRP volume selling at Bitso: The last 36 hours of hourly volume trade: The Bitso trade pricing range of the same hourly marks as above: And the most recent pricing/volume action at Bitstamp (the omfg moment of the year):

That is for sure friend ! Just take a look at today's hourly XRP volume at Bitso (bottom is their hourly prices high/ave/low) on trades.

IIRC the "paper gold" liquidations have occurred in the past to provide deleveraging capital, and it's quickly bounced back. One thing I believe, is very very little physical gold and silver has changed hands. You can bet the CBs and Chinese et al have VERY firm grips on their precious.

The fact gold dropped $80 in four hours this morning, imho, CLEARLY shows a desperate need for cash. Leveraged longs, and such create for DIRE exposure to Black Swan events as we're seeing unfold this week.