This issue is a dive into our all-time high methodology.
Before we get to that, I want to mention . . .
Our Most Recent Flippening Podcast, With Hunter Horsley, CEO of the Hold 10 Crypto Index Fund
The most recent episode of Flippening focuses on Crypto Index Funds.
My guest is Hunter Horsley, Co-founder and CEO of Bitwise Asset Management, which operates the very first crypto-index fund called the HOLD 10 Index. Hunter’s index tracks the top 10 crypto assets weighted by an inflation-adjusted market cap.
The indexing approach is about collecting small fees from lots of investors, and about passive methodologies winning over an active strategy of picking coins. It’s a belief that a properly formed index can beat hedge fund performance without source code reviews, meeting founders, or creating value hypothesis.
Hunter’s thoughts and worldviews are sobering counterpoints to some of this podcast’s most popular interviews.
Back to our regularly scheduled program . . .
How We Calculate All-Time-Highs
by Nomics Co-founder and CTO Nick Gauthier
We compute all-time highs off of raw trade data when provided by exchanges, otherwise, we use aggregated OHLC candles from exchanges.
However, simply taking the highest priced trade from an exchange at face value poses problems when price manipulation is encountered. For example, on Kraken’s BTC/EUR market, these are the trades with the highest prices:
Timestamp Volume Price 2018-01-13 11:22:45.5896 0.00919439 20000 2018-01-13 11:22:39.3126 0.00906561 20000 2017-12-17 12:28:25.2543 0.00011985 16323 2017-12-17 12:28:25.2375 0.00209015 16322.3 2017-12-17 12:28:25.2211 0.18379 16322.2 2017-12-17 12:28:25.0783 0.046 16322 2017-12-17 12:28:25.2044 0.054 16322 2017-12-17 12:29:00.9303 0.1 16321.9 2017-12-17 12:28:30.1217 0.002 16321.9 2017-12-17 12:29:01.1342 0.0042 16321.8 2017-12-17 12:29:01.2693 0.4958 16321.6 2017-12-17 12:29:01.3993 3.79793483 16321.6 2017-12-17 12:28:24.7294 0.005 16321.6
As you can see, a few low-volume high-price trades were created (perhaps to manipulate the price, although there’s no way to be sure).
We take two approaches to prevent high price manipulation. First, we blacklist particular exchanges and markets that have a lot of manipulated trade data.
Second, we take the top five highest trades on any given exchange and market, and we throw out trades larger than five times the lowest trade in that list. For example:
Market A top 5 prices: [200, 199, 198, 195, 194]
Market B top 5 prices: [700, 5.00, 4.95, 4.94, 4.92]
For Market A, 200 < 5 * 194, so 200 is the high. For Market B, 700 > 5 * 4.92, so 5.00 is the high.
This is a very simple calculation and we believe it will start to break down if fraud an manipulation significantly scale up on crypto exchanges. (If manipulation does see significant increases, we can modify our calculations after public review or rely more heavily on our blacklist).
We are working towards a more robust all-time high calculation method that will do a more thorough job of detecting outliers. Until then, this simple method has been working well.
Closing Thoughts (From Clay)
We’re really proud of the transparency we provide around our ATH data (especially how we reveal the currency pair and exchange that triggered an all-time high for a cryptoasset):
To date, no cryptoasset pricing index has been especially transparent about the methodologies behind their calculations. We hope to change this with this second deep looks at how our indicators are computed. In this spirit, please consider this content as part of a conversation we’re having with the cryptosphere about the proper computation methodologies for global cryptocurrency metrics. If you have suggestions or constructive criticism, please leave your thoughts in the comments.
Also, if you believe a high to be manipulated, please contact us.