Hello all and welcome to my analysis of trading patterns in crypto markets!
Before we begin, there are several key assumptions we must establish:
The reference data is based on trading prices and volumes on Bitfinex. This is because I believe that Bitfinex is the leading exchange in the cryptocurrency market due to its popularity, despite higher recent BTCUSD volumes on Binance.
Only USD-based pairs are examined
The index is a dollar-volume weighted aggregate of margin-tradeable alt coins on Bitfinex, excluding BTCUSD
The correlation and beta calculations are on a 14-period basis, which are benchmarked against BTCUSD
Given limitations with the data, the index is examined only on a range from October. An second index, consisting of only ETHUSD and LTCUSD will be provided to show the longer timespan
In this snapshot, I have split the frame into three timeframes (A, B, C), which highlight a changing market dynamic over time.
Timespan A captures the peak of bitcoin's 2017 hype. The lowest correlation point is -0.65 which was accompanied by a beta of -1.65, while the highest correlation was 0.80 with an index beta of 3.44. During this span, a relatively volatile correlation (refer to the standard deviation of correlation oscillator) between the altcoin index and BTCUSD can be examined, with similar volatility patterns in the beta. The interesting component here is that, during the price rise, alts would follow BTCUSD when BTCUSD was rising, but experience inflows of money from BTCUSD when the latter was falling. As such, the market mentality was bullish - the money would flow from BTCUSD to alts and then back around. By examining the index beta over this period, you can see how alts outperformed BTCUSD when times were good and when times were bad.
Timespan B begins to paint a different picture, where the volatility in correlation and beta measurements is far less drastic. Over this span, the lowest correlation was 0.28, showing that directional movement between the index of altcoins and BTCUSD remained positive, even when weak. It is important to note the longer duration of high correlations during this period, which were accompanied by similar movements in beta.
Timespan C is where we currently are and, in my opinion, this is a very interesting setting in the market. As the chart shows, the lowest correlation was 0.50, while beta almost consistently remained above 1. The standard deviation of correlation had been only trending downwards from Timespan A, which is reflected in the relatively consistent correlation measurements over the period. The conclusion to be made is that altcoins, in aggregate, have become strongly correlated to BTCUSD, while a beta above 1 indicates higher performance in alts than BTCUSD.
TL;DR: chart BTCUSD and trade altcoins based on BTCUSD momentum. As the statistics show, the market mentality has shifted away from moving money from one crypto to another, but rather buying/selling alts and BTCUSD simultaneously.
Furthermore, we can examine a longer timeframe in the chart below:
In this chart, the correlation and beta, as well as the standard deviation of correlation, are based on a dollar weighted aggregate of ETHUSD and LTCUSD, which were the longest standing altcoins on Bitfinex. The narrative described above remains the same, where the correlations have become far stronger in recent times compared to their historical trends, while the beta has become stronger over time, gaining more ground above a measure of 1 during the high correlation periods.
The final point is an examination of dollar volume in BTCUSD compared against altcoin dollar volume:
As can be seen from the chart, the dollar volume in BTCUSD has been losing ground to the dollar volume in USD-based alt pairs (on Bitfinex)
ความคิดเห็น
⋅
As I ran out of space in the original post, the only closing comment is as follows:
The dollar volumes in BTCUSD are falling and more money is entering altcoins. If you are a trader, you should be following the money. If the correlations remain highly positive, there is no reason to be trading BTCUSD, as betas on alts far outperform BTCUSD.
For reference: Beta is a measure of the %-change on a 1% change in the benchmark. In this study, BTCUSD was the benchmark. So if the beta is consistently above 1 and correlations are near one, trading alts is like trading BTCUSD with higher leverage.
ความคิดเห็น
⋅
For those interested, examining beta calculations on a running basis is useful. Here is an example on a 1 hour basis:
If there is enough interest, I can publish the indicator.
ความคิดเห็น
⋅
Once the correlations in the cryptocurrency market break, I am under a strong belief that we will start to see the return of a real bull run. This is because weaker correlations between the money flows will imply that the money will be moving from one asset to another, as opposed to into all assets and out of all assets at once. This would be further reinforced by the market psychology, where a market downturn will start to appear less significant, and market players will maintain enough confidence to hold one asset even if another is falling. In aggregate, this change in money flow patterns will reinforce the confidence in the market, which will deter the bearish mindset.
Even tho you did a great analysis, you are comparing a single asset to an index composed of all volume weighted altcoins, in which i see 3 biais :
1 - diversified portfolio will statistically outperform single assets. Since the index is volume based weighted, under-performing altcoins variances are basically shaded in the end
2 - In phase B the alts volumes are heavily weighted by mainly ETH/XRP. Then when ETH/XRP volumes decreased in phase C, the "outperformance" (which is more a raise in volatility .. not much performance) you see is partially noises caused by low volumes pump & dumped alts.
3 - since most altcoins are heavily correlated to BTC, they are not independant at all (so considering a random vector variance to get your index variance - assuming you used that - is not valid)
I'd be happy to debate, and again, you did a nice analysis, thanks for sharing !
checkm8
⋅
@Meishin4, I believe you misread, but it is DOLLAR volume weighted. It is about the inflow and outflow of money in BTCUSD compared to an aggregate index consisting of alt-USD pairs. An argument can be made for nominal volume numbers, but this is about the patterns of money flow into and out of the "market".
Anoinvest
⋅
Great work ! Thanks for sharing this is great statistical confirmation for me
swiftnesses
⋅
Awesome analysis, would be interested in the indicator!
majic92
⋅
strangly alts have been selled before this pump on btc, i think somes whales have selled alt and buy btc, for rescue at 6K..probably a bull trap
ArShevelev
⋅
great
UnknownUnicorn2774089
⋅
well they did say in Jan 2018 is going to be the year of the alts
and look what happened!
train wreck
checkm8
⋅
@LeoKim, this is not about buying and holding, this is about trading. Alts have a consistently higher beta on a strong correlation to BTCUSD, meaning that the volatility on alts is higher than that of bitcoin and outperform its directional movements. As the study explains, alts are strongly correlated to BTCUSD, so unless BTCUSD breaks out into a bull market, neither will the alts. They have simply become more volatile proxies to bitcoin.
WaveRider1423
⋅
Nice job mate. Thanks so much for sharing. This is something I have suspected from a visceral level but your work here supports my gut feeling. All the best!
checkm8
⋅
@tuamen, yup - it was good to get a statistical confirmation of this. I've been mainly trading alts over the last year due to my belief in this, but the theory has never been more true :)
1 - diversified portfolio will statistically outperform single assets. Since the index is volume based weighted, under-performing altcoins variances are basically shaded in the end
2 - In phase B the alts volumes are heavily weighted by mainly ETH/XRP. Then when ETH/XRP volumes decreased in phase C, the "outperformance" (which is more a raise in volatility .. not much performance) you see is partially noises caused by low volumes pump & dumped alts.
3 - since most altcoins are heavily correlated to BTC, they are not independant at all (so considering a random vector variance to get your index variance - assuming you used that - is not valid)
I'd be happy to debate, and again, you did a nice analysis, thanks for sharing !