BOLLedOver
Introduction
BOLLedOver leverages classic signal strategies typically seen in equity program trading algos. Interestingly, in the crypto world, these statistical methods don’t mean quite as much as crypto is generally random and spastic. (I find some of the online analysis humorous—“setting bottom when we cross the 50 day moving average”—really? Trend following is measured in minutes, hours or days, not weeks or months. BOLLedOver uses the Bollinger Band method with various filters to insure good buying and selling opportunities. All my scripts use the same framework with different signaling strategies as I have found this a very useful way to “test and learn”. The framework allows configuration for “interesting” parameters to their underlying statistical functions. In addition, I have found that trade execution strategies are far more important than getting the right signals in place. In the live BOT version, my framework allows for MARKET orders only, and “chasing the book” which insures that you can place LIMIT orders attempting to be a “market maker” not “taker”. In addition, the framework takes a trailing STOP approach which eliminates a lot of risk on the down side.
With so much trading, fees and slippage can really make a difference. Some exchanges provide free trades (GDAX) if you make markets with your trading. This applies to both the BUY and SELL sides and proves very useful. A big mistake with newbies in this area is not considering cost (and slippage). My BOTs have yielded approximately 70-80% free trades (mileage varies based on statistical settings) with exchanges that have the market maker policy. Always, be conservative in back-testing the strategies with fee settings—this can quickly destroy profits.
I have tested extensively on Binance, Bitfinex and GDAX and leave it to you to backtest your favorite exchange—BOLLedOver trades a lot with some settings, so if your exchange isn’t quite as reliable or has holes in the data the BOT may not perform as advertised.
Key Features
• Designed for market maker trading
• Leverages classic statistical models in a unique crypto way
• Trades when market is sideways or heading up, sleeps when heading down
• Two to three trades daily (depending on settings and market action)
• Tune-able with ample knobs and levers.
Parameters
Stop Loss % (default 98%) STOPs will be placed and ratchet up following stops with each 1% increase in price action.
Stop Loss Trail % (default 96%) Second STOP starts here.
Bollinger Bands Time Period (default 6) indicates 6 candles in calculation. BOLLedOver runs best at 15 minutes periods. Try your own setting with plenty of backtesting.
Average Volume (default 18) filters buy and sell signals
Buy ROC Length (default 75) number of candles averaged for positive rate of change, which gives the go ahead to act on a BUY signal. When markets are heading south the BOT goes to sleep. You might get a STOP LOSS haircut (default 3%, 2 to 1 chance if you are in a position), but no trade chattering in whipsaw downward spirals after that.
Sell ROC Length (default 85) number of candles averaged for a negative rate of change, which gives the go ahead to act on a SELL signal. Note: the tighter the Bollinger Bands (e.g. 5) the less likely a SELL will process before a STOP LOSS is reached making this parameter useless in those cases.
MACD – the moving average convergence/divergence is used to check the validity of BUY and SELL signals
MACD Fast Period (default 10)
MACD Slow Period (default 24)
MACD Signal Smoothing (default 10)
Candles to Wait After Trade (default 10) set to 0 to turn off. Keeps trades from occurring consecutively in pump and dump environment.
BOLLedOver leverages classic signal strategies typically seen in equity program trading algos. Interestingly, in the crypto world, these statistical methods don’t mean quite as much as crypto is generally random and spastic. (I find some of the online analysis humorous—“setting bottom when we cross the 50 day moving average”—really? Trend following is measured in minutes, hours or days, not weeks or months. BOLLedOver uses the Bollinger Band method with various filters to insure good buying and selling opportunities. All my scripts use the same framework with different signaling strategies as I have found this a very useful way to “test and learn”. The framework allows configuration for “interesting” parameters to their underlying statistical functions. In addition, I have found that trade execution strategies are far more important than getting the right signals in place. In the live BOT version, my framework allows for MARKET orders only, and “chasing the book” which insures that you can place LIMIT orders attempting to be a “market maker” not “taker”. In addition, the framework takes a trailing STOP approach which eliminates a lot of risk on the down side.
With so much trading, fees and slippage can really make a difference. Some exchanges provide free trades (GDAX) if you make markets with your trading. This applies to both the BUY and SELL sides and proves very useful. A big mistake with newbies in this area is not considering cost (and slippage). My BOTs have yielded approximately 70-80% free trades (mileage varies based on statistical settings) with exchanges that have the market maker policy. Always, be conservative in back-testing the strategies with fee settings—this can quickly destroy profits.
I have tested extensively on Binance, Bitfinex and GDAX and leave it to you to backtest your favorite exchange—BOLLedOver trades a lot with some settings, so if your exchange isn’t quite as reliable or has holes in the data the BOT may not perform as advertised.
Key Features
• Designed for market maker trading
• Leverages classic statistical models in a unique crypto way
• Trades when market is sideways or heading up, sleeps when heading down
• Two to three trades daily (depending on settings and market action)
• Tune-able with ample knobs and levers.
Parameters
Stop Loss % (default 98%) STOPs will be placed and ratchet up following stops with each 1% increase in price action.
Stop Loss Trail % (default 96%) Second STOP starts here.
Bollinger Bands Time Period (default 6) indicates 6 candles in calculation. BOLLedOver runs best at 15 minutes periods. Try your own setting with plenty of backtesting.
Average Volume (default 18) filters buy and sell signals
Buy ROC Length (default 75) number of candles averaged for positive rate of change, which gives the go ahead to act on a BUY signal. When markets are heading south the BOT goes to sleep. You might get a STOP LOSS haircut (default 3%, 2 to 1 chance if you are in a position), but no trade chattering in whipsaw downward spirals after that.
Sell ROC Length (default 85) number of candles averaged for a negative rate of change, which gives the go ahead to act on a SELL signal. Note: the tighter the Bollinger Bands (e.g. 5) the less likely a SELL will process before a STOP LOSS is reached making this parameter useless in those cases.
MACD – the moving average convergence/divergence is used to check the validity of BUY and SELL signals
MACD Fast Period (default 10)
MACD Slow Period (default 24)
MACD Signal Smoothing (default 10)
Candles to Wait After Trade (default 10) set to 0 to turn off. Keeps trades from occurring consecutively in pump and dump environment.
릴리즈 노트:
Added CANDLES_AFTER_TRADE for crypto trading. Pump and dump activitiy affects a great trade with a new BUY right after closing the deal and subsequent DUMP, take a rest for a few candles.
릴리즈 노트:
This script logic will be available on cryptotrader.org/?r=2649 as a rentable BOT. You will need API keys for automated trading.