Crypto Trading 2018 in Review: I backtested 32 strategies with full year of data

[deandree]

I was getting a bit tired of seeing a lot of unrealistic results/expectations posted in this forum and also people asking if they can get 10% a day, so I collected 32 open source Gekko strategies from Github and backtested them against full year, to show a bit more realistic view of how much profit could be made if I followed these strategies for a whole year (2018). Keep in mind - reality is always worse than backtest.

You can see my results HERE.

Backtesting a strategy means - I have a code that gets candles and indicators as input (Bollinger Bands, RSI, MACD etc), and then makes BUY or SELL call or does nothing. Based on those calls, Gekko platform calculates profits/losses against historical data (in my case - BTC from Jan1 2018 to Jan1 2019). That's it.

This way you can get closer representation to reality compared to just trading some random basic oldschool strategy you found on somewhere that says you can make X% day with it.

For example, here is how results for 120m candles look like:


Here are buy/sell decisions from some of the best strategies. Green/Red arrows are buy/sell decisions.





I you are interested, you can also check my profile for more posts like this.
https://medium.com/@deandree

[1715537130]

1715537130

[1715537130]

1715537130

[1715537130]

1715537130

[Marbigoad]

Interesting research, have you tried to test bigger candles like 1day / 2days etc?

[deandree]

Thanks, I have tried 16h / 24h with these strategies, result was worse. I have tried 2-3 days a while back with different strategy set, result was also worse compared to 120/240m. But that's just my result for a few specific runs. Haven't tried 1week, because I think it's way too slow for crypto market tempo.

[clrpod]

I'm going to look in to this in some more depth when I have the time but it answers some questions and disproves some estimations of potential profit people have. From what I could gather the best result was a 323% profit in the year. That works out to being about 0.4% per day if compounded. A far cry from the 10% people claim or even the 2-3% many people say is easy.

[kawetsriyanto]

I've just taken a look the strategies written on your article. It is quite complex strategies I ever read so far. Regarding to your research, I am interested to learn more about Turtle, RBB-ADX-BB and Bestone. Actually I am just starting to do trading, so don't really know well the most appropriate strategies to apply in current market condition. This could be very helpful for me.   

[wuvdoll]

I like to hear your current performance in terms of profits for current markets just to evaluate the strength of your strategies.
If time permits, please share about your open calls and target details so that we could follow and may decide to make use of your strategies.

Many times I have heard about back testing to analyze the newly designed strategies. In my understanding, traders who are not having enough time and patience for testing in real markets, going for back testing. Other than this is there any significance importance are there to follow a back testing ?

[CryptoNeed]

First thought was "yet another 'trading expert' who made profit in 2017" but this is actually a great work!

[deandree]

@clrpod
Yes, 323% was the best but that was on a single coin for single strategy in single timeline. If you take average from all coins, result is worse (like <100%). And since past performance is not indicator of future results, it most probably would have been even worse if you tried this now.

@kawetsriyanto
Thanks, I describe Turtle and Bestone in Part2, I suggest checking that and especially source code. Keep in mind you need to heavily backtest this yourself before thinking about going LIVE, don't take my results for granted!

@wuvdoll
I feel you're suspicios and that's good - you should always be very very skeptical when reading articles about trading. I'm NOT saying "Follow strategy X and get +300%", that's not how it works. Past results don't guarantee anything in the future, but at least they serve as a bit of reality check. Because if you get negative 80% in backtest, you can be pretty sure that this is not working.
As for strategies, they are not mine, they are open source, I've collected them from Github, forums and Discord channels. And I'm not recommending them. Also I'm not saying they are bad. This is just an experiment to show you a bit more realistic results.
As for importance of backtesting - every serious algo trader backtests the hell out of strategies. How can you not? Imagine manufacturing a car and not testing it, just putting pieces together and hoping it sticks. Bad plan.

@CryptoNeed
Thanks, glad you enjoyed the post!

[Upgrade00]

Great analysis. Lots of traders and wanna be traders are made to believe earning percentages which are unrealistic, and they are sometimes blind ti the possibility of losses which is inevitable when starting out.
I am not an active trader as yet, still trying to polish my skills. And always open to new strategies.

[deandree]

Great analysis. Lots of traders and wanna be traders are made to believe earning percentages which are unrealistic, and they are sometimes blind ti the possibility of losses which is inevitable when starting out.
I am not an active trader as yet, still trying to polish my skills. And always open to new strategies.

Thanks! It's sad, but those unreasonable expectations mostly come from "Trading Gurus", who don't actually trade but sell ebooks, courses and bots for unreasonable prices. On the other hand, market is a zero sum game, the more such "10% per day" traders, the higher chances for you to win, theoretically.