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 3 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
New month has started, so it’s time to backtest again and see how/if things have changed. This time I will be testing all 19 strats from previous two parts and see how they come up against August market.
So the last two parts were sad to say the least, but on the bright side, this was also one of the saddest periods in Crypto lately, so if we can survive that, we should be good later.
Overall, August was mixed, just like everything in Crypto world lately. Started with huge drop, but at the end there was a swing back up.
If we compare the up/down percentage change for TOP5 coins, we get a good estimate of how August compares to June and July:
My test setup remains the same except for two things:
- I will no longer be using VEN coin because of Bitfinex VEN/VET token swap. Since the ratio changes, I really don’t want to re-calculate all my imported data, because this will make previous VEN results hard to compare to each other and complicate things. So that leaves us with 19 coins for this test suite.
- Since it was visible in previous tests that bigger candles brought better results, this time I will also include 120m and 240m candles, which makes it 6 timeframes for each strategy/coin — 5m/10m/30m/60m/120m/240m. Let’s see if increasing candle size brings any improvement.
Basic Strategies (from Part 1)
Let’s start with Basic strategies:
Not much to comment here. Only a few things that are not totally bad:
- TMA is the best from this basic suite, shoving some promise on longer timelines (120m/240m), but not too much.
- CCI (30m/60m/120m) and UO (60m/120m) could be worth exploring.
Other than these two, there are no good news.
Advanced Strategies (from Part 2)
Just like in Part 2, 5m/10m candles are pretty useless. Except for IWannaBeRich and MACD-CCI-RSI (Part 2 winner) — I find their ability to get even something close to profitable in 5m/10m pretty surprising.
Next, 30m/60m candles, MACD-CCI-RSI again in the top spot, IWannaBeRich starts to struggle, bestone and RSI_BULL_BEAR_ADX up next but still far from profitable, although seems worth exploring a bit with Genetic Algo or just manual parameter adjusting. I particularly like relatively small losses of RSI_BULL_BEAR_ADX, means if you improve the upside, downside should be already taken care of.
On 120m/240m we finally see some progress. Best results so far overall, considering all 3 Parts. Unusual suspect appears — Turtle. Althouth still not profitable on average, is shows much better results that I expected. MACD-CCI-RSI basically almost not generating any signals here.
bestone doing really good actually. Supertrend not bad, although losses still bigger than wins. IWannaBeRich also worth exploring.
Interesting statistic— trend following methods dominate the mean reverting ones in this timeline.
Let’s take a closer look at how some of the best results were achieved.
Supertrend 240m
Very pragmatic. Catches the long term uptrend nicely. Also, best part of the profit comes from last trade which is not actually a full troundtrip (just long signal), but Gekko counts that into profits, fair enough, makes sense.
But on bad test cases (ZEC/ZRX) we can clearly see the problems with it — getting whipsawed and false breakouts.
Also, looks like it has a bug. It starts each test with back-to-back buy/sell combo. When debugging, looks like first calculated variable is always 0 which is lower than candle and signals long, while next update is real number and more commonly larger than candle, which signals short. I don’t think this is intended in this strategy.
Bestone 120m
Even less trades, just a single one actually and not even completed roundtrip, basically only thing is does is catches long term uptrend and it does it well. Keep in mind that you can’t directly compare these unfinished roundtrips with strats that end with a trade, because you don’t know when the strat would exit, could be few % below when we see in these results.
When looking at bad test cases (ETC/ETH), we see that is handles false trends pretty bad and stays in trade too long which causes the big loss. Could be handled better with some kind of stoploss.
MACD-CCI-RSI 60m
Not much to say here, single good trade without even a full roundtrip, again. Only makes trades in 3 of 19 coins, so no bad cases to analyze here.
Summary
August brings better results than Jun/Jul, maybe because they were combined together.
Advanced strategies > Basic strategies, but more visibly than in Jun/Jul. Difference even more in higher timeframes.
Adding 120m/240m timeframes did bring some improvements.
Best strats this time — bestone, MACD-CCI-RSI, IWannaBeRich.
