Bitcoin Max trading bot
You want to invest in Bitcoin, but you feel a ‘buy-and-hold’ strategy is too risky? Then “Bitcoin Max” is the right bot for you.
The Bitcoin Max bot trades Bitcoin vs USDT. The bots scans the market 24/7 and enters a (spot) position when the algorithm identifies the start of an uptrend. The bot assumes that the trend will continue and stays in the position until it gets a confirmation that the trend is over, no matter how long that may be. Although this means the bot never buys the bottom and never sells the top, the strategy consistently outperforms the market over time.
LONG TERM RETURNS
The strategy of this bot has been extensively tested and validated in a process called “walk forward testing”. In contrast to “backtesting”, this process does not look back at what the optimal settings of the algorithm would have been, but tests the performance in a paper trading simulation with actual market data.
During the 10 year test period (1-1-2011 to 27-10-2021), the bot realized the following returns:
45x the market (If you started with 1 BTC on 1-1-2011, you would now have 45 BTC)
8.8 mln x your investment (If you started with USD 1 on 1-1-2011, you would now have USD 8.8 m)
The returns are net of 0.2% fees & slippage and without any leverage. The testing setup is explained in detail at the end of this page
Please note that this bot is optimized to beat the market in the long run (years). In bull phases of the market, the return of the strategy is slightly lower than the market. In bear phases of the market, the strategy doesn’t enter many trades. While the market retraces, it waits for the next bull run. This is where the bot realizes most of its outperformance.
A buy-and-hold strategy was profitable in 10 out of the previous 12 years and realized a compound annual return of 208% per year since 2011. However, in 2015 and 2019, this strategy resulted in a significant loss of -58% and -72%. If you would have started at the wrong time (e.g. 1-1-2018), your average annual profit would have been much lower.
The Bitcoin Max bot was profitable every year and realized a compound annual return of 338%.
If you have been reading about Bitcoin, you probably came across the notion of 4 year cycles of tops and bottoms in the Bitcoin price, driven by the ”halvings”. Roughly every four years, the number of Bitcoins that are produced by miners is halved. The reduced supply drives the price up over time. In this light, the optimal horizon to compare investment strategies is a four year period. This way, the volatile price swings during the cycle are cancelled out and we only evaluate the long term performance trend.
Investing in Bitcoin for 4 years with buy-and-hold strategy, resulted a return of at least 10x for almost every start date (If you invested USD 1 and waited 4 years, you would have USD 10) . Only an investment from Nov 1st 2017 to Oct 27nd 2021 would have a lower return of 8.7x .
In contrast, the Bitcoin Max bot realized a 25.7x return on investment during the same 4 year period. Furthermore, the chart shows that the bot outperformed a buy-and-hold strategy during every 4 year investment period.
SHORT TERM RETURNS
As the strategy is aimed at outperformaning the market in the long term, short term returns can fluctuate. The bot can underperform a buy-and-hold strategy for many months or even years (see the monthly returns presented in the chart below).
The underlying reason is that the momentum strategy leads to a large number of small losses. The money is made with a few big wins. Furthermore, the real outperformance is realized during bear markets.
Total trades: 371
Trade frequency: 2.8x per month
Winning trades: 46% Losing trades: 54%
Average profit per trade: 7.4%
RISK
To assess the risk associated with investing in the bot we look at the maximum drawdown. The maximum drawdown (MDD) is the maximum observed loss from a peak to a trough, before a new peak is attained. In other words, it is the maximum observed loss if you invested at the worst possible moment.
The maximum drawdown of the Bitcoin max bot is -35% (in 2021)
RETURN VS RISK
By comparing the annual return to the maximum drawdown of the bot we can determine the return/risk ratio RoMaD (Return Over Maximum Drawdown).
This approach is often used to assess the performance of hedge funds. In practice, investors want to see maximum drawdowns that are half the annual portfolio return or less, which results in a RomaD of 2 or higher. The Bitcoin Max trading bot has a RoMaD of 9.7, almost 5x the benchmark.
RoMaD = Compound annual return / Max drawdown = 338% / 35% = 9.7
A more prudent approach is to use the RoMaD over the past 4 years (in line with Bitcoin’s 4 year cycle). In that case we evaluate the performance from 27-10-2017 (bot value USD 342.7 k) to 27-10-2021 (bot valuu USD 8,819 k). The bot’s average annual return during this period was 125% per year. During this period, the Bitcoin Max trading bot had a RoMaD of 3.6, almost 2x the benchmark.
RoMaD = Compound annual return / Max drawdown = 125% / 35% = 3.6
TESTING APPROACH
To determine the optimal settings of a trading algorithm, many quantatitve traders use a process called “backtesting”. Backtesting refers to applying a trading system to historical data to verify how a system would have performed during the specified time period. By running these tests with different settings, the “optimum” settings can be discerned. Unfortunately, tweaking a system to achieve the greatest level of past profitability often leads to a system that will perform poorly in real trading. This over-optimization (“over fitting”) creates systems that look good on paper only.
All our algorithms follow the same generic momentum strategy, but the settings are different for every trading pair. To derive these settings without presenting unrealistic returns, we used a process called “walk forward testing”. In contrast to “backtesting”, this process tests the performance in a paper trading simulation with actual market data.
The simulation of the Bitcoin Max bot followed the following steps:
The initial settings of the algorithm were derived from backtesting on the dataset that preceeds the simulation (1-7-2010 to 1-1-2011)
The initial settings were then applied to actual market data to simulate the bot’s performance in the first quarter of 2011 (1-1-2011 to 1-4-2011)
In a first iteration, the settings of the algorithm were updated based on backtesting on the total available dataset in the simulation (1-7-2010 to 1-4-2011)
The updated settings were then applied to simulate the bot’s performance in the second quarter of 2011 (1-4-2011 to 1-7-2011)
This process of updating the settings each quarter was repeated until a first cycle of 4 years was reached. After that the settings were updated each quarter based on the optimal settings of the previous 4 years (instead of the full history) and then tested on the upcoming quarter
The combined returns of all simulations form the test results displayed here
DISCLAIMER
The test returns show that the strategy worked well in the past but unfortunately don’t guarantee that the bot will also outperform the market or realize a profit in the future.
Due to several optimizations the historic returns of the live bots do not always match backtesting returns 1-on-1.
Source of all trading data is the TrdingView Bitcoin index. As bots on different platforms trade on different exchanges (e.g. Binance and PanCakeSwap), small differences in price and slippage can occur