The Kraken Candle: We can do better, lots better
The magic candle
On September 13, 2019, a very strange bar appeared on the chart of the Bitcoin price on the Kraken exchange. It looked as though the price of Bitcoin had fallen, within the day, from around $10,250 all the way to $8,000, then rallied to $12,000, then returned to $10,250 or so. Or, perhaps, it went up first then came down. It was impossible to know because, as one began changing the time scale on the chart, from daily to hourly, from 5 minutes to 1 minute, the bar looked the same. The high was $12,000. The low was $8,000.
And that’s all we knew. There was no news out of the exchange. I, personally, saw some discussion on LinkedIn and Twitter, but nothing more came from Kraken all day.
Then, on September 14, the exchange released this string of tweets, describing and partially explaining what had occurred.
Wait. What?! Stops were triggered and filled because of a test?
That there was a bug in a new order type was not a complete shock. One expects bugs from time to time. But we were left with that rather odd statement “stops in the range of $8–12k were triggered” and “Stop orders were correctly filled”.
It was a test. So, why were stops triggered? Remember fire drills when you were a kid in school? I don’t seem to remember them burning your books and your backpack after you left the classroom. The idea, if I understood it, was to get you ready for the fire, but NOT suffer the effects of the fire. So why were the stops filled again?
There were a few tweets in reply from users, many of whom were not exactly happy about this chain of events. There were a few deflecting and defensive tweets from the exchange.
And the message to customers who felt they had been stopped out in error was to submit a support ticket.
And what about people who had long leveraged positions on Kraken, something Kraken allows? Were they liquidated due to the test? It appears that, yes, they were. But again, why? If the “candles were hollow”, why should a leveraged position be liquidated?
The even more magical disappearing candle
But then, on September 16, 2019, there was another miracle. The chart was altered and there is no more crazy candle.
So, who was stopped out? Who took what loss? What happens next?
It’s awfully hard to tell, because there are no further tweets from Kraken. There is nothing (at the time of this writing, anyway) on the website. And there were no emails to customers. It quiet out there…… too quiet.
So, where does this leave us?
First, there was a lot of confusion. No one knew who was stopped out and where.
Second, there were undoubtedly losses, but they were losses at the expense of the customer and seemingly caused by the exchange itself.
Third, there were conspiracy theories. Did Kraken clear out orders before some big move? Did they owe something to whales who were going to scoop up the stopped out Bitcoin? I would assume not, but it’s hard to dismiss conspiracy theories when you just cost your customers money and you aren’t talking.
Fourth, this sort of thing makes the whole cryptocurrency industry look bad — immature, poorly planned, and not ready for the big time.
Fifth, that bar disappearing from the chart does not improve things. Are we really changing the past? I mean, maybe if every customer account goes back to status quo ante then OK. That would make sense.
But, lastly, how do we know?
Ladies and gentlemen, I believe that we can do better. This improvement can take two forms — better communication and structural changes.
Communication
Announcements: This much is obvious. Communication should happen as soon as possible. Of course, there are limits and we should not be surprised that the exchange had to do some research. My guess is that there was a 24 hour period heavy on lawyers and coders. I understand that when you release the first announcement is something of a judgment call, but the sooner the better.
Transparency: Kraken had a problem. Kraken divulged the problem. So far, so good. But the information release was fairly limited and followed by silence. What was behind the silence from Saturday through Monday? Did management limit what public relations could say? Did lawyers? Letting people know that you are working through some issues and will have more answers by a specific date would have been a really good idea.
Clarity: This was a failure on Kraken’s part. What on earth does “the candle was hollow” followed by “stops were triggered” mean? This response is internally inconsistent and, as such, left more questions than answers. If you can’t give customers the whole story right away, don’t say something that leaves them scratching their respective heads.
Expectations: Now we get to the crux of the matter. You really need to tell people what to expect before it happens. Here are a few suggestions: 1) we had a problem, 2) more information will be coming soon, 3) we’ll try to minimize any customer losses. Would that have gotten past the lawyers? I think it would have helped.
With any testing going forward, can you warn customers that tests are coming? For instance, does anyone want to have stop orders open on an exchange that is testing new code? Frankly, you probably shouldn’t have stop orders on a crypto exchange at all, but certainly not as they are testing or rolling out anything new.
Just tell people what the rules are, what to expect. Then customers can make a plan. If they leave stops in, well, that’s their decision. But it should be, as much as possible, an informed decision.
Re-writing history: That adjusted bar chart on Monday morning compounds the problem. Re-drawing the chart to take Friday’s move out is a form of communication. There still has been no clarifying email. There is nothing new in the Twitter feed. So, we are left to interpret the chart. Are all the stop orders that were filled reversed? That would certainly be the implication, but we don’t know yet. It seems like a bad idea to re-write history on the chart if you haven’t re-written history in your customers’ accounts.
Structural Changes
OK. I’m talking my own book in this section. This is something we are working on at Grey Swan Digital. If you wish to discount this section as you would something marked “paid advertising”, you may feel free to do so. I just don’t want to spend three pages complaining about Kraken’s failure of communications, then hide my own self-interest. Well, OK, maybe I do want to, but I won’t.
What if you could structure a market with no need for stop losses? What if you could have a form of leverage without margin calls? Certainly, you can do that by buying conventional options, but the seller has to worry about margin and risk. I’m talking about both sides, knowing their limits and never having to set stops. You could use stop losses, but they would not have the same importance they have now.
I don’t want to dig too deep into the how of it right now, but what would the advantages be?
First, you could relax about flash crashes. We know these things happen. They can be caused by legitimate economic or geopolitical events or worries. They can be caused by “fat fingers” or errors in execution. This has happened several times including in Bitcoin. It appears not to have been the cause of the May 6th 2010 crash. Now, thanks to this most recent event, we can add “an exchange testing code” to the list. Finally, there is probably some lovely black swan flash crash cause that we’ll all learn about… but only after it happens. One could trade for short term moves, with leverage, without worrying about being knocked out of a winning position by such short term imbalances.
Second, such a structure could also give some level of safety for exchange failures or other trading halts.
Third, this could also be very helpful for genuine hedging, either in crypto or in physical products derivative markets. One danger of hedging is that, yes you get leverage, but if the market moves against your hedge, you may owe money you don’t have yet. In such a liquidity crisis, you can be hit with a margin call, be liquidated, then lose the price protection of the hedge. The ability to leverage without margin calls would be extremely helpful for this scenario.
Given that one of the stated purposes of derivatives and markets is to allow producers and consumers to hedge, we should try to make this work for them, not increase their risk.
Of course, if one is trading an asset like Bitcoin, which has a completely asymmetric return profile (it might go up 10x, can only go down 1x), there will always be reasons to hold the actual asset. A safer leveraged product is part of a mix here, not the full solution.
Finally, safety built into the structure itself means less reliance on operational perfection — moving us closer to a “trustless” environment. We may never reach full “trustlessness”, but moves in that direction that remove the need for perfect institutions are good even if limited.
End of our solution
Conclusion
I don’t mean to beat up too much on Kraken (oh, now he tells us). I believe and hope that they will handle this properly for their customers. I don’t know yet, but believe and hope. I’m sure there will be communication in the coming week or so, as they sort through the mess. A professional exchange with good tools and that communicates well is a benefit to the ecosystem, so I really want them to get this right.
But overall, if the crypto arena is to be taken seriously, everyone needs to do a better job. We need to give people advanced warnings. Users should have a clear sense of the rules of the road. And there should be no papering over of mistakes (no re-adjusting price charts) until customer accounts are adjusted and any adjustments have been communicated to customers.
And, we should work to integrate new structural solutions like those I discuss above, so we can reduce our dependence on human and flawed systems. There are going to be errors, but having safer outlets for both leveraged hedging and speculation can improve the landscape, even if the details are the subject of another article.
