The Little Bot SNAFU That Could

29 May 2018

At around 8:45 UTC on April 28 2018, in two minutes the Bitfinex BTCUSD price swung suddenly and violent between 9500 and 8750 and generated one of the highest volume candles in months. The 9500 and 8750 prices used for the apparent stop hunt didn't seem correlated that well with trends up to that point, and were hard to explain at the time that they happened.

Since then, that candle has been a goldmine of information, predicting trendlines both near the local top and relevant to now the falling wedge borders we've followed weeks later. That candle is, as best I can tell, the result of a mistake by a really important algorithm in the BTCUSD market that controls a very large amount of capital on Bitfinex, which has inadvertently given away a great deal of information about its plans for May and likely at least early June based on what it did in two short minutes on an April morning.

This particular post isn't so much to post any prediction (though based on the other lines I have on the chart, I am indeed expecting to find bottom and see a falling wedge reversal in this region in the next couple of days) and more to highlight the observation and a give general reminder that BTCUSD is a market whose trends are controlled by relatively few large players. It appears that, given 20/20 hindsight, we can say sometimes those players give away their game by accident, and if you can recognize those mistakes, you can use them to predict some tradeable trendlines based on only 2 points.

I'm an amateur and you definitely shouldn't take anything I say as financial advice. I'm interested in any feedback.

30 May 2018

Not

Posting this recent example from the ETH 1m chart as an addendum to this post since I can't publish on the 1m charts. Really worth being on the lookout for algorithmic oddities.