Written by OKEx Insights | Powered by Kaiko
A PDF of the following report has been included at the bottom of this page so that readers can view, download and share it at their convenience.
Despite all the setbacks, global shocks, disruptions and crises it brought, as it nears its end, 2020 is shaping up to be a great year for cryptocurrencies and Bitcoin, propelling the latter to its former highs — a prospect that appeared far too distant only months prior.
When Bitcoin rallies, the market tends to forget previous, long-drawn bearish stretches, and the sentiment shifts to manic euphoria as quickly as a thousand-dollar-candle appears on the BTC/USDT chart. And while it is exciting to see Bitcoin go up, analyzing its price action against market behavior can help paint a somewhat telling picture of how market participants drove or reacted to various price ranges, surges and pullbacks.
For this report, we’ve collaborated with blockchain data firm Kaiko once again and analyzed data from the BTC/USDT market on OKEx between August and November 2020. Our focus here will be on filtering and batching trading data based on both amounts and trade directions (whether they were sell or buy orders). By charting such trades against the price, we aim to assess how various market segments behaved as Bitcoin surged to its new all-time high by the end of November.
Before getting to the actual analysis, however, we’ll introduce the methodology, and highlight the challenges faced in determining trade directions on any market. The discussion here requires an understanding of order books, makers and takers, and bids and asks — all concepts that are addressed in our in-depth guide to trading.
Methodology
Since even a single market on a high-volume exchange like OKEx can execute hundreds of thousands of trades in a day, any digestible analysis needs purposeful data filtering in order to reveal insights otherwise buried. For this report, we collected daily trade data from the OKEx BTC/USDT market between Aug. 1 and Nov. 30, 2020.
Trade ranges and market personas
This data focused on the number of daily trades executed on the BTC/USDT market, their amounts, directions and the overall volume-weighted average BTC price for the day. However, since such a dataset includes millions of values, we further grouped these transactions by amount-based ranges. These ranges, apart from simplifying the visual representation of this data, also serve as descriptive categories of market participants.
All daily transactions on the OKEx BTC/USDT market were hence grouped into five ranges: transactions under 0.5 BTC, between 0.5 and 2 BTC, between 2 and 5 BTC, between 5 and 10 BTC and, finally, 10 BTC and above. While these ranges are largely arbitrary, they do generally line up with established market personas, such as retail traders, professional traders, large traders/whales and institutions.
Retail traders typically include speculators, casual day traders and small investors. They are often seen following market trends, as opposed to setting them. Professional traders, on the other hand, often trade for a living, and they use advanced trading techniques and tools, including technical analysis and algorithmic trading.
The distinction between large traders, whales and institutions is harder to make, as there are no strict thresholds for these traders. Typically, a whale is an entity holding enough coins to be able to move market valuations by selling a large number of them all at once. While a single whale may easily execute a few $100,000 trades a day, a large trader with only about $100,000 in trading capital can theoretically do the same. The same is true for institutions, which are known to conduct large buys and sells — but those trades could also be attributed to whales. Hence, while we discuss these market participants, our analysis refrains from specific attributions.
That being said, it is also important to acknowledge the gaps in this dataset and the relevant assumptions. Firstly, this data only spans one exchange and one market, and even though OKEx is one of the largest in the industry, it does not represent the entire space.
Secondly, while we can assume that all trades above 5 or 10 BTC are very likely to be either whales or institutions, we can’t assume that whales and institutions trade only above these thresholds. In fact, it is in their interest to execute large trades in smaller batches so as to not affect market liquidity and, consequently, the market price.
Thirdly, this data does not account for over-the-counter services, which are often used for large transactions. These services place multiple, so-called child orders, often across days, to fulfill the parent (i.e., larger) orders, which makes it difficult to identify or attribute these trades accurately.
Determining trade direction on crypto exchanges
Even though we often come across expressions like “increased institutional buying” or “massive selling” in the course of general market discussions, in reality, all buying and selling has a counterparty. This means there cannot be “increased institutional buying” without there being sellers, nor can there be “massive selling” without there being buyers. To put it simply, on an exchange, any time you buy a coin or a token, someone else sells it, and vice versa.
In such a scenario, how can any trade be labeled as a buy or a sell? How can we determine if the market is biased toward buying or selling? This is where the concept of makers and takers comes into play.
Summarily put, makers add liquidity to any market by placing limit orders that sit on the order book, available to be filled by takers. Meanwhile, takers place market orders that, when filled, remove the limit orders (previously placed by the makers) from the order book, thereby reducing liquidity.
A common convention used in order to determine the direction of any trade is to consider it from the taker’s perspective. If a taker places a market sell order, which is then filled by a maker’s limit buy order, that trade is considered a sell transaction. Similarly, if a taker places a market buy order, which is then filled by an available limit sell order, the transaction is counted as a buy.
Our data partner for this report, Kaiko, has a detailed article on their work in terms of normalizing cryptocurrency trade data, and we use their taker_side_sell variable to determine the daily number of sell and buy transactions on the OKEx BTC/USDT market.
Visualizing data
With our daily trading data grouped into amount-based ranges and further filtered with trade-direction data, we were able to move toward visualizing this data. In order to simplify this process, we decided to calculate daily percentages of buying and selling transactions, as well as to calculate their net differences.
The next step involved charting these values against Bitcoin’s price to see how market behavior changed as the price of BTC appreciated. Since we’re considering aggregate values, we chose volume-weighted average price, or VWAP — a way to measure the average price that an asset was trading at throughout the day — to accurately reflect the daily price change corresponding with the number of trades.
In some of the charts that follow, readers will be able to see Bitcoin’s price for each day, coupled with the percentage of buy and sell transactions. A second set of charts will then show the net difference between buying and selling trades on a daily basis, and will more accurately represent the changing sentiment of market participants belonging to each of our grouped categories.
Additionally, to add another critical perspective, we will look at sets of charts that follow Bitcoin’s price progression through this period, from around $10,000 to nearly $20,000, and note how the daily net buying (or selling) changes with price across these ranges, irrespective of chronological dates.
Finally, to conclude, we will highlight some of the key findings from these datasets by comparing them against each other in a table format — putting the insights extracted from our analysis into perspective.
Trades under 0.5 BTC — Retail traders
Trades under 0.5 BTC, as expected, represented the largest volume of daily transactions on the OKEx BTC/USDT market. These trades can be valued anywhere between $10 (0.001 BTC) and $5,000 (0.5 BTC) if BTC is priced at $10,000. We take $10,000/BTC as the reference price for this report because of its psychological significance as a key support level and because the period being analyzed in this report started with BTC trading around this level.
Looking at the entirety of this data, between Aug. 1 and Nov. 30, we have 122 days. When comparing the number of buying and selling transactions (as per the method explained above) for each day, we learn that traders in this range were net sellers for 65 days out of these 122, or 53.28% of the time. This means that on 65 days, selling trades were higher in number compared to buying trades, while buying trades were dominant on 57 days.
A daily chart (above) for the percentage of selling and buying trades against the volume-weighted average price shows that the buyers and sellers in this range (0.5 and below) were largely balanced, albeit a bit biased toward selling. We will now look at the net percentage difference against the price to identify more specific trends.
The chart above shows the net percentage difference between buying (positive values) or selling (negative values) on a daily basis, compared with the volume-weighted average price.
We can see here that the traders in this range were mostly selling when Bitcoin traded around $11,000 and above in August (shown by the dominance of values below 0.00%), possibly expecting a correction toward $10,000. This was seen in the shift toward net buying (values above 0.00%) starting from September, when Bitcoin dropped to $10,000 levels.
This buying took a backseat again as BTC traded above $11,000, all the way to $13,000, at which point it picked up again. However, buying interest largely peaked around $15,000, after which retail traders have seemingly been indecisive, mostly selling during the Thanksgiving crash (on and around Nov. 26) and cautiously buying on the bounceback.
In the chart below, we see the same daily net percentage difference against ascending VWAP values — simply, the lowest to highest VWAP values over our selected time period. This chart shows us how traders in this category reacted to price changes, regardless of chronology.
Once again, we can see that most of the buying was around $10,000 levels, while the majority of the selling was between $11,000 and $13,000, following which we see more buying until $15,000 levels. From there onward, retail traders appeared uncertain as to the market direction, but they followed the trend by selling on dips and buying recoveries.
Given its diversity, this range represents the largest chunk of market participants, including speculators, day traders and casual investors. Our data essentially shows that, in August, retail traders were not expecting the price to stay above $11,000 for long, and that they were seeking opportunities to buy below this level. After the September drop, however, they have been following the price surge and have been net buyers on most days — all the way to the new all-time high.
Trades between 0.5 and 2 BTC — Professional traders
Trades between 0.5 and 2 BTC represent the second-largest volume on a daily basis. These trades can be valued anywhere between $5,000 (0.5 BTC) and $20,000 (2 BTC) if BTC is priced at $10,000. For our purposes, we attribute this range to professional traders.
Looking at the 122 days between Aug. 1 and Nov. 30, traders in this range were net sellers for 80 days, or 65.57% of the time (compared to 53.28% for the retail range). Moreover, while retail traders were mostly net buying from September onward (i.e., the number of buying-dominated days in the month were higher), professional traders in this range only became net buyers in October and November.
The visual representation of daily sells and buys (in percentage terms) in this category against the VWAP shows a bias toward selling, as compared to the more balanced trend seen in the retail range.
The net percentage difference compared to the VWAP also highlights the predominant selling pressure throughout August and September, as the price dropped from $11,000 levels to $10,000 levels and subsequently recovered. The shift in sentiment only came in October, as the price broke through $11,500, after which professional traders have mostly been on the buying side, especially in November.
The first major buying peaks are seen on Oct. 18 and Oct. 21 (corresponding with $11,500 and $12,500) while similar selling action was seen on Nov. 1–2 and Nov. 10 (corresponding with $13,500 and $15,300 levels). The majority of the buying in this range started after the $15,000 price level, however, and continued, for the most part, until the all-time high — with the exception of the Thanksgiving crash.
Charted against ascending VWAP values, the net percentage difference shows mostly selling activity until about $12,000, buying interest around $13,000 levels, subsequent selling until $15,500 and then a major shift toward buying post-$16,000.
This range (0.5 to 2 BTC) represents the beginning of relatively larger trades and is likely to include professional traders who use technical analysis and charting techniques alongside algorithmic trading. This could explain the pattern here, somewhat, since a Fibonacci retracement, drawn between the 2017 high and the 2018 low, shows $12,000, $13,000 and $16,000 as price levels corresponding with the 0.5, 0.618 and 0.786 Fibonacci levels, as shown below.
Trades between 2 and 5 BTC — Large traders and whales
Transactions in this range are valued anywhere between $20,000 (2 BTC) and $50,000 (5 BTC) if BTC is priced at $10,000. While these figures are not indicative of whales, per se — as traders not holding millions of dollars worth of BTC can also make these trades — they do mark the threshold from which we can start considering large traders and whales as participants.
Interestingly, in this dataset, of the 122 days between Aug. 1 and Nov. 30, traders were net sellers on 86 days, or 73.50% of the time. Moreover, unlike professional traders — who became net buyers in October and November — the traders in this range were net sellers throughout.
Our percentage values chart above shows selling trades in this range overtaking buying trades for the majority of the period, with some exceptions. Notably, a more balanced approach can be seen during the time in which Bitcoin was trading under $11,500.
The net percentage difference chart above shows buying interest around $11,500 levels (with the highest peak in mid-October recorded at this level), followed by $15,000. Notably, after $15,000, the selling pressure only intensified in this range, with most selling trades recorded from Nov. 21 to Nov. 30, often as the price rose.
The net percentage against the ascending VWAP chart further confirms our observations, with notable selling post-$15,000 and especially after $18,000, all the way up to the all-time high.
This pattern indicates how large traders, and possibly whales, bought at low levels, around $11,000, and decided to take profits on the way up, especially near the all-time high, where they seem to have sold when retail and professional traders were buying.
View https://www.okex.com/ for the full report.
If you found this analysis insightful, you can view, download and share the PDF version of this report below:
OKEx Insights presents market analyses, in-depth features, original research & curated news from crypto professionals.
Not an OKEx trader? Sign up, start trading and earn 10USDT reward today!
was originally published in OKEx Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.
