Check out my latest app — TradingGYM. It’s a trading simulator that helps you practice trading with a much faster feedback/learning loop and minimal lookahead bias. Once you open the app, you will see a chart with a random asset at a random point in time. Make trades and fast-forward time to see how it played out.
This is Part 2 in multi part series:
- Part 1: Basic strategies, introduction, setup and testing vs June-July market.
- Part 2: Advanced strategies and where to find them, testing vs June-July market.
- Part 3: Basic and Advanced strategies testing vs August market.
- Part 4: Neural Network strategies description and backtests against September market.
- Part 5: Neural Network strategies backtests against October market.
- Part 6: Did Neural Network strategies predict November 14th price drop?
- Part 7: Crypto Trading 2018 in Review: 17 Advanced + 15 Neural Net strategies tested
- Part 8: Intro to Statistical Arbitrage in Crypto — Pairs Trading
- [NEW] Part 9: Crypto Trading 2019 Half Year Review: 17 Advanced + 15 Neural Net strategies tested
Three weeks ago I backtested a few basic Gekko strategies to come to a conclusion — basic is not enough. The next question is — can we find something profitable by just browsing around the web and testing what we can find?
Where to look for strategies?
- xFFFFF GitHub repo, contains tons of strategies but have not been updated since end of May
- Just search GitHub by “Gekko strategy” and similar keywords
- Gekko Community Forum
- Gekko Discord Channel — a lot has been shared in the channel, but you have to dig a little.
My selection process was mostly random this time, although I tried to pick some popular strats that already have a name in community and also keep balance between trend following and mean reverting strategies.
For this round of testing, I have chosen 10 strategies, 5 mean reverting and 5 trend following. All props to authors, I tried to find the original source as best as I could.
NEO
- Mean reverting
- Found here
- Uses SMA (slow/fast) to determine trend (BEAR/IDLE-BULL/BULL) then uses RSI thresholds (depending on trend) to make long/short calls. Also ROC extra setting to check if trend very-BULL or not-so-BULL (IDLE-BULL).
RSI_BULL_BEAR_ADX
- Mean reverting
- One of the most popular strats circulating around, found here with a long long discussion Has a few variations (without ADX, with stoploss). Uses SMA (slow/fast) to determine trend (BULL/BEAR) and then uses RSI thresholds (depending on trend) to make long/short calls. Uses ADX to get trend strength and adjusts RSI thresholds based on that.
- Due to available candle data and history requirements, I had to reduce SMA_long (from default 1000 to 500) to fit into 20 day limit, since I have data available from May 10.
bestone
- Trend following
- Found here
- Uses 4 indicators (EMA, STOCH, MACD, RSI) to determine the trend, if trend persisted long enough, goes long/short.
rsidyn
- Mean reverting
- Found here
- Uses RSI and keeps track of last highs and lows, calculates median and peak for highs/lows and if RSI over/under, goes short/long.
Supertrend
- Trend following
- Found here.
- Uses just 1 indicator — ATR (volatility) to calculate upper/lower bands and then makes long/short decisions based on that.
EMADIV
- Mean reverting
- Found here
- Uses just EMA and 2 thresholds (up/down %), if EMA over/under threshold, goes short/long.
RSI_WR
- Mean reverting
- Found here
- Uses RSI and WILLR (momentum), if both over/under threshold, goes short/long.
MACD-CCI-RSI
- Trend following
- Found here
- Uses 3 indicators (MACD, CCI, RSI) to follow the trend, if persisted long enough.
IWannaBeRich
- Trend following
- Found here
- Uses 3 indicators(BB, RSI, MACD) indicators to follow the trend.
Turtle
- Trend following
- Found here
- Calculates high/low based on X last candles, if detects breakout, tries to follow in the same direction
Results
Ok so we have 5 trend following and 5 mean reverting strategies, let’s go and find out how they have performed in last 2 months (Jun 1 — Jul 31).
My setup and methodology is pretty much the same as Part 1:
- 5/10/30/60 minute candles
- Default settings for each strat (except for RSI_BULL_BEAR_ADX which I explained above)
Here are the results:
Analysis
Overall still looks bad, we are bleeding cash here. On the bright side, this is better than Part1, so we are going in the right direction. Two big things to notice:
- In current conditions, Trend Following > Mean Reverting
- In current conditions, Big candles > Small candles
If we look at individual strats, we have just a few positive outliers here. I will use the sum profit of all coins here as a measurement for sake of simplicity.
So first one — MACD-CCI-RSI 60m candles with +31.12 %. keep in mind, most of that is +20% on BTC and +11 on BCC. Actually it’s not even making any moves in most of the coins so that’s basically all it has done this period.
Next is IWannaBeRich with 60m candles which manages to archieve +3.28% in total over all coins.
Other than these two, the rest are down and down heavily.
If we look closer at how MACD-CCI-RSI 60m BTC, result was archived, we can see that very few trades very performed.
If we compare SUM of trades made per strategy, this is even more visible. Maybe that’s something to learn from this — less trades = more win.
Now, don’t get this wrong. Just because strategy has bad result here doesn’t mean it’s useless or bad or whatever. Maybe the strategy wasn’t meant for current market. Maybe default settings doesn’t work anymore. There is a million possible reasons. You cant just mindlessly (like I just did) plug them in and hope for the best. Learn what they do, play with settings and see what you can come up with.
