Compilation of the original text: Deep Tide TechFlow
Compilation of the original text: Deep Tide TechFlow
secondary title
TL;DR
As token engineers, our goal is to understand how different elements of token unlocking design—namely, the size, frequency, periodicity, and distribution of unlocking—affect the stability and long-term health of token prices.
We collected and analyzed data from more than 5,000 different unlock events and came to the following conclusions:
Small unlocking events (i.e. increasing circulating supply from 0% to 1%) have no material relationship to price.
Larger unlocking events (i.e. increasing the circulating supply by more than 1%) have a clear negative correlation: as the unlocking size increases, the price decreases.
Tokens that have unlocked a majority of their supply (over 70%) have significantly lower volatility and higher relative prices, whereas tokens early in the unlocking period have relatively lower prices.
Protocols with more allocations to private placements (e.g. teams, investors) perform slightly better than public placements (e.g. ecosystems, communities). However, in our opinion, this outcome is not high enough to be a top consideration for token engineers.
From these conclusions, we propose three aspects that founders need to consider in terms of token economics, briefly summarized as follows:
Consider limiting the size of the unlock to no more than 1% of the circulating supply. Prefer daily or weekly unlocks over quarterly or yearly unlocks.
Reconsider inclusion of large-scale unlock events. These events can generate significant and unnecessary price pressure.
first level title
Target
While the impact of massive token unlocking is present in nearly every project that raises funds, it is an understudied aspect of token design. These unlocks can generate considerable volatility, causing confusion for the community, token designers, and traders.
However, when done correctly, token unlocking helps align incentives among stake holders. So what is the best way? We dig into the data to understand how token unlocking has historically impacted price action and overall protocol success. Can we glean generalizable principles from the data? How can founders act on these insights?
Our goal is to gain general insights into how different unlocking design elements (size, frequency, period, and distribution) affect token price stability.
first level title
data set
We collected and validated data from 20 protocols, including Uniswap (UNI), Galxe (GAL), and BitDAO (BIT), resulting in over 5,000 distinct unlock events for analysis. Data verification includes viewing on-chain data and/or confirming the token unlock plan and protocol core team.
Vested percentage as of April 2023:
first level title
analyze
Initially, we investigate the direct relationship between token unlock size and historical price changes. To get a general trend context for the data, we collected normalized price data for each unlock and plotted the mean and median lines on a single chart. We look at one month of price data, unlocking a 15-day window before and after occurrence.
On average, the token price was about 13% higher in the first 15 days of unlocking and rose moderately thereafter. The median line shows a stronger price downtrend, dropping another 5% after unlocking. We attribute the deviation between the median and the mean to some anomalies in extreme market events. Therefore, we believe the median is a better proxy for marginal unlocking.
These results fit basic economic intuition: If supply increases rapidly without a corresponding change in demand, prices should fall. However, the situation is clearly more complicated. Unlocking schedules are often public, providing opportunities for narrative building and event trading. More recently, due to a short squeeze*, some token unlocks have become a bullish narrative. This anticipatory activity further complicates the situation. Is there a way to alleviate it? Is it worth the relief?
(Shen Chao Annotation: Short squeeze, one of the terms in the stock market, when short sellers are forced to close their positions (empty positions) and continue to buy the underlying stock, the demand for the stock in the market far exceeds the market liquidity situation, due to insufficient supply. leading to a sharp rise in prices)
first level title
secondary title
analysis of the day
We first assess the market response to unlocking by analyzing single-day price changes. Below we plot the percentage change in price (compared to the previous day) versus the percentage change in circulating supply. Essentially, this attempts to measure the relationship between unlock size and price impact. If our correlation shows a negative relationship, then it means that greater unlocking is associated with greater price declines.
We classify two different types of unlocking: private offering unlocking and public offering unlocking. Private equity unlocks include teams, collaborators, investors, and advisors, while public equity unlocks include treasury, ecological funds, community distribution, and airdrops. In total, we collected 2187 public sale unlock events and 4546 private sale unlock events.
When comparing private and public unlocks, we find that public unlocks are small—typically less than 2% of total supply—while private unlocks have a larger and negatively correlated size and price impact. This makes sense, since unlocking to insiders often includes big events like Cliff and/or Quarterly unlocks, some of which may be sold upon receipt of tokens.
Isolating the private placement unlock leads to the following conclusions:
A notable feature is the cluster of points between 0% and 1% of supply unlocked, suggesting that different relationships may exist at different unlock size ranges. To investigate this, we looked at unlocks between 0% and 1% and greater than 1%.
For the first cluster (0%-1% unlocked), we found no linear relationship. This is in line with the judgment that small unlocks have less impact on price, but one still expects the price impact to increase as the unlock size increases. Instead, it seems that any unlock size in this range generally has a similar impact.
For the second cluster (unlocked volume greater than 1%), we see a stronger negative relationship, indicating that as the unlocked size increases, the price decreases. This is consistent with our hypothesis that larger unlocks lead to larger price drops. Since there are several exogenous variables in the token price, we knew we wouldn't expect a high correlation, but thought 16% would be meaningful for the token price data.
secondary title
longer time window
An obvious downside of single-day analysis is the lack of price data before and after unlocking. We assume that there are both forward and reactive effects when token unlocking occurs, which may take days or weeks to manifest. For example, short sellers could initiate a position weeks before a major unlock event, or it could take days for an insider to exit a position due to illiquidity. To investigate this, we performed a similar correlation analysis for longer timeframes, including 3 days, 1 week, and 15 days before and after unlocking.
For the early days of the unlock event, we calculated the price change between the day of the unlock and the first day of the window period. Thus, the symbolic meaning of correlation is inverted. Positive correlation means higher price before unlocking, negative correlation means lower price. If we believe our intuition is correct, then we would expect to see a positive correlation in the days leading up to the unlock, as expected selling pressure leads to price drops leading to the unlock. Again, we should see a negative correlation within a few days of unlocking as the price continues to drop.
Likewise, we found little relationship between 0% and 1% unlocking. The line of best fit is generally flat both before and after unlocking, indicating that for unlocks of this size, neither type of analysis exhibits a persistent price impact.
And for unlocks with greater than 1% of the circulating supply, we get the following relationship:
As a preliminary caveat, we strongly state that causality cannot strictly be proved, but instead take these results as possible evidence of correlation for consideration and further study. Furthermore, we restricted the correlation analysis to a two-week time window, since long-term measurements will increasingly include unknowable or unmeasured factors. Finally, there may be some token unlock overlap as the time window increases, but we believe these effects will be minimal. With these caveats in mind, our results should serve as a useful, evidence-based indication that prices generally tend to unlock around prices in some way.
At first glance, these results seem to support our hypothesis. Usually a positive correlation occurs before unlocking, indicating a higher price before unlocking. After unlocking happens, we see a negative correlation, indicating a drop in price.
In the 3-day and 7-day time windows before unlocking occurs, we see a relatively strong correlation between price and unlock size (24% and 23%). Furthermore, the correlations were -7% and -15% during the 3-day and 7-day periods after unlocking. This suggests stronger price pressure up to a week before unlocking occurs, most likely due to the anticipation of publicly known unlocking events. On the other hand, our data is closer to the regression line a few days after unlocking, indicating more confidence in this result.
secondary title
model analysis
We cannot rely on empirical data for long-term analysis because there are many exogenous factors over too long a time period. However, we can use Agent Based Models*, which are closed systems with discrete variables, to understand long-term effects.
(Deep Chao Annotation: A method used to simulate the actions and interactions of autonomous agents (independent individuals or common groups, such as organizations and teams)Computational model, evaluate the role of the agent in the system as a whole by image display)
We simulated three different situations and a control group. In the control group, no tokens were distributed to institutional investors. We then introduce an institution holding 8% of the token supply for our experiments: tokens are unlocked every other day, every month, and every 6 months.
Overall, we find similar results to empirical results, suggesting that larger unlocks produce larger and longer-lasting token price drops. Over longer time frames, we're seeing greater variance and variation as the unlock size increases.
Correlation Summary
first level title
Unlock period
Another hypothesis we have is that some token recipients may sell after receiving tokens, which means that early unlocked tokens may be at a lower liquidity rate than most of the tokens that have been or are fully unlocked. There is greater and more persistent selling pressure in the sexual environment.
To evaluate this, we split the dataset into two parts: mostly unlocked (>= 70% unlocked) and mostly locked (<70% unlocked). There are 9 tokens that are mostly unlocked and 11 tokens that are mostly locked.
It should be noted that these coins fell into these categories at the time of publication, so we examine the most recent period to gauge how they have performed in terms of current vested state percentages. We look at 4 months, from January 15th to April 15th of this year. To measure performance, we focus on two metrics.
mean variance. Instead of variance or standard deviation, we use the coefficient of variation, which is a measure of the standard deviation divided by the mean. This allows for direct comparisons of volatility between different assets without being skewed by token price changes. Essentially, we measure how much divergence and volatility occurs in the price of a token over a time period.
Average price change. The percentage change in price from the beginning to the end of the time period. We use this metric to understand how a token is performing compared to the market.
After running the same metrics on Bitcoin and Ethereum, we have the following results:
first level title
Token distribution
As our aggregated statistics show, the average distribution of tokens is 63% public and 37% private. Again, we split the 20 tokens into two parts: "more public allocation" (8 tokens) and "more private allocation" (12 tokens), and analyzed them in a similar way to above.
Our hypothesis is that tokens with more private sale allocations experience greater volatility as narratives can be built around large private sale unlocks, especially sell-offs from “insiders”. After performing the same analysis, we came to the following conclusions:
January 15 - April 15, 2023
October 1, 2022 to January 1, 2023
Summarize
Summarize
Overall, the results of our simulation studies show similar conclusions to the empirical ones. An unlocked amount greater than 1% of the circulating supply will result in a negative correlation between unlocked size and price. Interestingly, for smaller unlocks, the impact is almost non-existent. We also found that protocols with a majority of tokens already unlocked were closer to the market and outperformed those tokens that were in an earlier unlock cycle. Finally, tokens with a higher proportion of private placement allocations have slightly lower volatility and better price performance.
These results provide some conclusions for the project founders.
1% rule:Founders seeking to reduce token price volatility should unlock fewer tokens relative to the circulating supply. Our data suggests that unlocking less than 1% of supply is not associated with price impact, implying that unlocking on a block, daily or weekly schedule may be preferred over quarterly and annual unlocking events. Another significant advantage of smaller, more frequent unlocks is that the underlying sell pressure is more evenly distributed rather than concentrated on large events.
Reconsider allocation:We observed that the price drop was most pronounced before a large token distribution. In some cases, this has resulted in a 20% drop in prices that takes up to 2 months to recover. Reducing the distribution of large events helps eliminate any reinforcement of negative stories and unhealthy leveraged trading practices. Founders should consider the possible ramifications of unlocking a year's worth of tokens, rather than making large distributions the default token design. US teams may need to delay unlocking by 1 year to maintain compliance with existing regulations. To achieve this, teams can start unlocking tokens after a year, rather than unlocking a year's worth of tokens in one event.
Original link
