Things to keep in mind about backtesting

In the last lesson, we briefly learned how to backtest. Backtesting with Excel is one of the basics of quantitative investing, so it is a skill set that anyone who is interested should know how to do!

However, there are a few things to be aware of when backtesting, and I will explain them this time!

 

1. Backtesting is just backtesting!

Many strategies have good backtest results, but there are times when they don't yield returns when they are put into practice. Backtesting is past performance, and past performance does not reflect the future. At this point, it is important to consider whether this is a valid strategy and counter-attacks any cognitive bias when looking at the strategy with a logical thinking style!

 

2. The backtest, excluding the fee and slippage, is a half backtest!

When conducting a backtest, you must include the cost of trading. The 0.1%~0.2% commission seems to be small, but if you trade 100 times or 200 times, you eat a lot of the overall rate of return. In particular, the higher the frequency of trading, the greater the amount of commission you pay, so you must reflect this and backtest!

Slippage means that the transaction cannot be concluded at the desired asking price. This is often the case in asset classes with low liquidity. Personally, when investing in the cryptocurrency market, I am aware that risk can be spread by holding and distributing 5 universes (coins to invest), but 70% of the market capitalization of Bitcoin and Ethereum in the cryptocurrency market Considering the market specificity of occupying more than one share, only Bitcoin and Ethereum are traded. These two have a lot of fluidity, so the slippage is quite good, about 2-3 ticks!

 

3. Check if you have over-optimized and check again!

When backtesting the strategy, it is meaningless to draw a pretty picture by inserting the 23-day moving average as well and entering the number that shows the best performance in the past data, such as stop loss -2.375%. The reason is that these numbers may have worked well in the past, but there is no guarantee that the same numbers will work well in the future.

 

4. Backtest database

Backtesting may be a strategy that worked in 2020 or didn't work before, and conversely, a strategy that worked before but didn't work in 2020. Alpha investment through market timing, not beta investment through asset allocation, is not eternal. However, when analyzed as a long-term time series, if alpha has been created for a long period of time, it should be viewed from a probabilistic perspective with a high "probability" that will work in the future. Therefore, when conducting backtest, it is recommended to simulate at least the latest years of data.

 

5. In the cryptocurrency market, even for the same coin, the performance of each exchange is different.

Unlike the stock market, the cryptocurrency market has individual exchanges in each country around the world. Therefore, the user base of each exchange is different and the trading volume is also different. The important thing is that the price movement of the same coin is not the same for each exchange, and if you backtest the same coin on different exchanges with the same strategy, the results are all different.

***Binance has the highest trading volume at the moment. Binance has relatively low fees and plenty of liquidity.

 

Above were the points to be aware of when performing the backtest!