In this new appointment, we will explore a strategy called “bias”, which analyzes the repetitive behavior of a particular market. The strategy identifies periods of time or, for example, days of the week in which it is more profitable to buy or sell a given underlying.
In particular, this type of strategy uses information derived from all repetitive behaviors that occur over a predetermined period. Depending on the duration of the trade, three different macro categories of distortion strategies can be distinguished:
- Monthly or “Seasonal”
In the specific case of this article, the analysis focuses on a relatively fast time horizon, intraday, and we will test the idea on the second most capitalized cryptocurrency in the world: Ethereum.
Ethereum (ETH) Trading Bias Strategy Test
We continue with a test on the Unger Academy® software, the Bias Finder™, which will make it possible to evaluate the average historical trend of the Ethereum price in a very simple way. Figure 1 shows this trend from 2018 to 2021.
Almost at a glance, one can see how there is a bearish trend in the early hours of the session that starts at midnight (GTC), which quickly resumes in the late morning, around 11:00, and lasts until the finale. hours of sitting in a kind of harmonic cycle, composed of dips and rises, that follows day after day.
It is clear that the analyzed trend is only an average result and therefore is not equivalent to a guarantee that every day of the year Ethereum continue this trend. However, it is true that the sum of the movements recorded on ETH led to these results, so it is worth examining the times detected by the software.
In the following figures, in particular, using this information to create an automatic strategy, we can see how by buying every day at 11:00 and closing the trade a quarter of an hour before the end of the session, at 23:45, we immediately get an excellent profit curve (Figure 2).
Back testing the strategy
The trade size used in this backtest is $1000. The backtest starts in 2017 and ends in the early days of 2023.
The pain point of this strategy is with the average trade, which is only $3.27 (or 0.327% of the position value). This is definitely not a fact that leaves you safe to work with this system in the real market, as the commission cost and the slippage (ie the difference between the theoretical and the effective price) that the trader would have to pay could be calculated around $2 (0.2% of the position value), so the net average trade remains a measly 0.127% ($3.27-2=$1.27).
In any case, this result is interesting because it should be remembered that the system as it was designed stays on the market for only a few hours and executes a trade every day of the backtest. In short, a bit like staying on the market for 12 hours yes and 12 hours no, continuously, every day of the year. The fact that you cannot aspire to extremely full average trades becomes more understandable, but you can always try to improve the value by adding a condition that limits the number of trades in history and makes the strategy more efficient and selective.
The images below show how adding a condition found in the proprietary pattern list improves the results. In fact, by isolating trades on days where today’s session high is at least 0.75% higher than the previous session’s high, the strategy moves from an average trade of $3.27 to $7.50 (0.75% of the position).
This condition identifies another confirmation situation in addition to the hourly signal in which the current session’s trend is bullish, at least compared to the previous day. A long trade will be opened (buy/buy) only if today, by 11:00 AM, the market shows strength compared to the previous day. Some other confirmation of the short-term market trend.
Even the profit curve is taking more graceful shapes with good results in all the years of the backtest, including 2022, which we know was very difficult for ETH and for all cryptocurrencies in general. Also, 2022 was left out of the initial backtest run by the software. Hence the periodout of sample”, part of the time not taken into account in the backtest, was more than positive and this only reinforces the basic idea of the system.
However, the strategy could be further refined or slightly modified, for example by adding a stoploss (point of maximum loss acceptable to the system) or takeprofit (point of maximum profit at which profit can be collected) or other conditions capable of further limiting the number of trades. This is just the first idea to show what are the tangible advantages of working with an automated strategy compared to more traditionalbuy & hold‘.
On the other hand, it must be recognized that 5-6 years of backtesting may not be enough to make a definitive judgment about a strategy, mainly because it is a biased strategy, usually one that could hide more pitfalls than more classic entry types like e.g ‘trend following’. What is certain is that the more than 450 trades obtained in the backtest period represent a reliable statistical sample.
As further evidence of the work done, we now continue to use the same strategy in the main cryptocurrency market as well, bitcoinswhere good results continue to be achieved as a result of what was seen on Ethereum (Figure 7).
This additional test increases the efficiency of the work done on ETH and adds even more robustness to the strategy.
Cryptocurrencies are certainly a young market, still immature, but starting to mature over time. Strategies away from the more traditional ones, such as distortion, are starting to send reassuring signals in these financial products as well, suggesting that these markets could be used in the context of a well-diversified portfolio.
Until next time!
Author: Andrea Unger