Estimating fluctuation for HBD/HIVE + cat tax

in SteemLeo2 years ago

You can profit from the ups and downs in the market even if these are just tiny. But how do you measure and predict market fluctuations? A straightforward technique would be to makes use of fundamental tools in statistics. In this post will apply this technique to the HBD/HIVE internal market.

We start with measuring fluctuations when the price remains approximately constant. Here is a internal market graph of the close and median prices:

This is from a couple of days before the big HIVE price jump and as you can see the average HBD/HIVE close price is cruising at around 0.15. There are a couple of jumps which we might remove from our data set since we are only interested in small fluctuations. After removing extreme values we could just plot a histogram of the prices and then try to fit a distribution. But this will not give a estimate market since we have left out what drive the market: volume. So instead of just considering price we consider weighted prices by volume. In doing so the volume determines the influence to the estimate. For the weighted prices by volume plot a hist and fit some distributions:

Not a perfect fit. But definitely a reasonable fit. So how to use this from a trading perspective? For a Hive sell pick a value towards the right tail of a distribution since the probability is high that the price will fall and for a Hive buy pick a value towards the left tail of a distribution since the probability is high that the price will increase.

But how would this work if there is a trend change? Over the last few days Hive has been bullish so we can have a look at that data. We are still interested in pumping profit out of small fluctuations but now the price moves up so the previous technique does not work since the price moves in a certain direction. To remove this price movement we can just consider a local average and measure the fluctuations w.r.t. to that average. Then we end up with this:

It is definitely, not as nice as before. Sudden jumps lead to more unpredictable behaviour. In addition, the time range of the jump was a bit short so you will see this reflected in the results. For applications I am highly scepticle that this gives useful information during sudden changes.

Anyway, this is a nice technique to add your trading toolkit if you have a large data sets to work with and the pattern changes are not too extreme.

All the data was extracted using the wonderful beem package.

Enough complicated words here is cat tax: =^.^=

Posted Using LeoFinance

STEEM 0.28
TRX 0.05
JST 0.038
BTC 36237.21
ETH 2395.27
USDT 1.00
SBD 3.77