There's a huge amount of people in Crypto who think that Buy&Hold is the best strategy. That might be true (especially when effort vs results are considered) for more mature markets, but I felt that in Crypto, simple Trend Following should be able to catch the bulk of the big moves and avoid most of the bear markets, so I ran some tests.
I picked some of the most basic Trend Following strategies with reasonable/round parameters:
- MA Crossover (rx_xma)— when slow MA crosses over fast MA, go long. I will also use several different MAs for this — EMA, LRC (LinReg), HMA, WMA, KAMA, SuperSmoother, WVMA, TEMA, ZLEMA.
Parameters used: fast MA = 5d, slow MA = 40d
- Supertrend (SupertrendXm) — uses ATR to get band width, goes long if price over the band.
Parameters used: ATR period = 5d, band multiplier = 1.5/2/2.5
- Bollinger Bands Breakout (rx_bb_bout) — goes long when price over upper band.
Parameters used: BB period = 3d/5d, BB stddev = 2, MA = EMA/LRC/SuperSmoother
- High/Low Breakout (hilo) — goes long when price over previous X days high. Close position when under X days low.
Parameters used: period = 5d/10d
- Linear Regression Slope (expreg_slope) — goes long when slope of regression > 0.
Parameters used: reg period = 5d/10d/15d
- Single MA/Price Crossover (rx_xma_single) — goes long when price crosses over MA. Same MAs as rx_xma will be used (EMA, LRC, etc).
Parameters used: reg period = 30d
- Single MA Slope (rx_xma_slope) — goes long when slope of MA > 0. Same MAs as rx_xma will be used.
Parameters used: MA period = 20d, slope period = 1d
- Long only.
- 0.10% commission.
- Simple condition for entry (long), reverse for exit (close).
- No stoplosses or take profits.
- Use 100% of balance on each trade (will add position sizing in next post).
- 60m candle size, except for Supertrend (1d to get daily ATR).
Coins and date ranges used
I will use 25 of the ~TOP50 coins and 3 different date ranges. Specific coins for each date range depend on data availability. Some will use partial range data (starting later), specified in parentheses.
Jan 2017 — July 2020
Coins used: XBT, ETH, XRP (since May 18, 2017), LTC, BCH (since Aug 1, 2017), XMR, DASH (Apr 12, 2017), ETC, ZEC, GNO (since May 1, 2017), IOT (since Jun 12, 2017), REP
Jan 2018 — July 2020
Coins used: all of the above + BNB, EOS, XLM, ADA, TRX, NEO, OMG
Jan 2019 — July 2020
Coins used: all of the above + BSV, XTZ, LINK, ONT, QTUM, QSH
Here are the results, AVG profits over all tested assets.
- Charts seem pretty conclusive to me: Trend Following > Buy&Hold.
- You won't beat Buy&Hold like this on each and every asset that you test or trade live. But, over multiple assets, chances are you will on average.
- Trade frequency is a bit low (0.5 – 2 per week).
- Max drawdowns are still pretty big (60 – 80%). Need some filters, stoplosses and position/risk management.
- While it's highly unlikely to beat pure isolated BULL market (like 2017) with basic Trend Following, over longer periods with multiple big swings, strategies catch bulk of the moves and come out ahead.
Submitted September 29, 2020 at 12:04PM by __deandre