This article focuses on understanding trend changes, support and resistance levels, and the techniques for identifying and trading them. Research shows that by using the method of identifying trend levels, annual returns ranging from 115% (using a hold-only trading strategy) to 220% (using a long-short trading strategy) can be achieved, as demonstrated by back-testing. This is no financial advice. Please do your own research.
What is a trend?
A trend in traditional finance and especially in the world of trading is defined as an order of higher highs (or lower highs) and higher lows (or lower lows) based on a fundamental high and low. So for having an up-trend, there will be needed at least two higher lows and one higher high based on an existing high and low.
Figure 1 shows the structure of an uptrend. The blue framed box shows the minimum structure needed to call it a trend. The longer the trend, the more higher highs and higher lows. The same schemata is for a downtrend with lower lows and lower highs based on an existing high and low. A trend change reversal occurs when the structure of the current trend does not continue. The reason for this could be a rejection at a major resistance level (the process of identifying these levels with the historical break-even analysis is the focus of the following two subtopics). For example, after the higher high in an uptrend, the next notable point on the chart is a lower low. This lower low is followed by a lower high. The higher high at the resistance level will then be the high as the starting point of the downtrend. These two points interrupt the existing trend and provide the fundamental structure for the next downtrend. Based on this scheme, a total of 37 trends were identified for the last 6 years. The table below shows the distribution of these trends.
Definition of historical break-even analysis
The approach to identifying the price levels at which a trend change occurs will be done with on-chain analysis of the Ethereum blockchain. In particular, one on-chain metric shows significantly good results for exactly this task. The metric is called historical break-even analysis. This indicator tracks the number of addresses with realized profits and losses, as well as the variation over time. By summing the dollar value of all sales for each address and subtracting the dollar value of all purchases, it is possible to categorize addresses into those that have realized profits, those that have realized losses, and those that were break-even. In total, there are three time series showing the percentage of all tracked wallets for each day over the last 6 years that made a profit, made a loss, or break-even on a particular day. The visualizations of the three time series are shown in the next subtopic in Figures 2, 3, and 4. Break-even means that the current price is equal to the average price of Ethereum (ETH) for these wallets. For further research, only the outliers, and the maximum values of these three single time series are relevant. This is because these outliers mark the price level that will be the main support and resistance level when trend changes occur.
How to detect trend changes?
Most of these price levels are identified by the break-even line. The following chart shows the price of Ethereum as a time series with the break-even values for each day.
The red box in Figure 2 frames the relevant area of outliers. The left y-axis gives the price of Ethereum in USD and the right y-axis gives the percentage of wallets that are break-even. Only days where the break-even value is greater than 5.5% are considered. Price levels with a break-even value lower than 5.5% introduce too much noise into the data. The relevant price levels are then defined by the range of open-to-close prices for these days. In total, only 20 price levels were identified. The days with a high break-even value are interesting because a large proportion of ETH-holders have Ethereum on their wallet at this price and/or because a particularly large number of new entrants have switched to ETH on these days. The next figure shows the profit line plotted with the price of Ethereum.
The red box frames the relevant days, which in turn marks the relevant price levels with the daily open and close. A threshold above 91.5% marks these days and shows that more than 9 out of 10 wallets tracked have made a profit from selling their Ethereum. It can be seen that this level was only reached at new all-time highs (ATH). The profit lines therefore indicate when the market is overbought and a correction is imminent. The next figure shows the Ethereum price plotted with the loss values.
As before, the red frame marks the days on which an extremely high number of wallets sold their ETHs and made losses. The 84% threshold also shows only a few days where this value was reached. The 91.5% threshold of the profit line and the 84% threshold of the loss line are both maximum values. Each time these thresholds were reached, there was a change in trend or at least a correction. Therefore, the historical prices of the days when we reached these thresholds mark the relevant price levels where support or resistance can be expected. A total of 26 price levels have been identified, ranging from $80 to $4800.
Evaluation based on different trading strategies
Regarding Table 1 at the beginning of this article, 34 of the overall 37 trends started at these price levels, and 28 of 37 trends ended at these price levels. The following table shows the distribution for the hit rate of starting and ending trends based on their category.
Table 2 shows that the long- and medium-term trends in particular start and end at the price levels based on the historical break-even analysis. This approach has been backtested as a long-only and long-short strategy against a hold-only strategy. The backtesting covers three years and is divided into three different market phases:
- 01.09.20 — 31.08.21 Bull market
- 01.09.21 — 31.08.22 Bear market
- 01.09.22 — 31.08.23 Consolidation
The results of the backtesting are shown in the next figure.
For backtesting purposes, each strategy starts with a deposit of USD 100. For the hold-only and long-only approaches, spot trading fees were used. For the long-short strategy, futures trading fees were used. The reason for this is that short trades cannot be executed on the spot markets. All trades were made without leverage. Each time the price reached a price level on the daily chart and the weekly close was inside the price level, a trade was executed for the long-only and long-short strategies. It can be observed that the hold-only approach performs best in the bull market and worse in the bear market. The long-only strategy performs almost as well as the hold-only strategy in the bull market. Even in the bear market, there was a small growth in the account balance possible. The long-short strategy was the best performer overall, particularly during the consolidation phase of the market. During the consolidation phase, when the market traded between two price levels throughout the year, there were no false signals and every trade was profitable. The following table shows an evaluation of the overall performance of each approach.
The Long-Short strategy offers the highest return and the best risk-adjusted return (as measured by the Sharpe Ratio). It appears to be the most effective of the three strategies at capitalizing on market movements. The Long-Only strategy also outperforms the Hold-Only strategy, suggesting that active trading only on the long side can generate better returns than just holding. However, it’s important to consider other factors such as tax implications and the time you’re willing to invest in planning and executing your trades.
This work was done in partnership between the University of Ulm and the AI startup Blockbrain. We’ve been able to dive deep into trading and risk strategies, uncovering insights that can benefit investors. To stay at the forefront of this rapidly evolving field and unlock exclusive content, make sure to sign up for Blockbrain.
About the author
Max is a Ph.D. student at the joint project “DeFi Risk Advisor AI” of the University of Ulm & Blockbrain, specializing in the convergence of finance, blockchain, and AI. Max has a deep interest and background in science and an interest in start-ups and distributed ledger technology.