Algorithmic Trading Video Series #13 | Crypto Trading

Hi

Thanks for the initiative.

Please do you know if shorting on futures can be done with QC on Bitfinex ?

Cheers!

Hi Ya1,
QuantConnect/Lean does not support crypto-currency futures with Bitfinex or any other crypto-currency brokerage yet. Please check out the supported security types on the docs.
Best,
Varad Kabade

Hi Louis!

As always your video is helpful and to the point.
I did some minor tweaks on the code, swapped daily resolution for hourly (to speed things up)  and used wilr instead of the rsi (just for fun). 
I then let it run live on a small sub account of mine on bitfinex. Not for profit, but to see if it would work and place orders as expected and to see if bitfinex and qc work together. 

After one night, it placed market buy orders for three crypto-usd pairs. 
And one market sell order. 
Here it get´s weird, or I don´t get it, anyway, it seems that the algorithm is not getting the “holdings” from the exchange, but rather calculating them from what orders it thinks were executed. within one night it is off significantly from the balances on the exchange.

#qc13 cryptoalgo from Louis slightly modified for testing - don´t put  money on this!!
class CreativeRedHornet(QCAlgorithm):

    def Initialize(self):
        self.SetStartDate(2019, 1, 1)
        self.SetEndDate(2021, 9, 1)
        self.SetCash(650)
        self.SetBrokerageModel(BrokerageName.Bitfinex, AccountType.Cash) 
        
        self.Settings.FreePortfolioValuePercentage = 0.05
        self.positionSizeUSD = 99
        self.rsiEntryThreshold = 60 # enter position if rsi rises above this threshold
        self.rsiExitThreshold = 60 # exit position if rsi drops below this threshold
        self.wilrEntry = -50 #enter if above
        self.wilrExit = -50 #exit if below
        self.minimumVolume = 10000 # filters out symbols with 30 day avg Hour dollar volume less than this 
        
        # add data for all tickers
        universe = ['BTCUSD', 'LTCUSD', 'ETHUSD', 'ETCUSD', 'RRTUSD', 'ZECUSD', 'XMRUSD', 'XRPUSD', 'EOSUSD', 'SANUSD', 'OMGUSD', 'NEOUSD', 'ETPUSD', 'BTGUSD', 'SNTUSD', 'BATUSD', 'FUNUSD', 'ZRXUSD', 'TRXUSD', 'REQUSD', 'LRCUSD', 'WAXUSD', 'DAIUSD', 'BFTUSD', 'ODEUSD', 'ANTUSD', 'XLMUSD', 'XVGUSD', 'MKRUSD', 'KNCUSD', 'LYMUSD', 'UTKUSD', 'VEEUSD', 'ESSUSD', 'IQXUSD', 'ZILUSD', 'BNTUSD', 'XRAUSD', 'VETUSD', 'GOTUSD', 'XTZUSD', 'MLNUSD', 'PNKUSD', 'DGBUSD', 'BSVUSD', 'ENJUSD', 'PAXUSD']
        self.pairs = [ Pair(self, ticker, self.minimumVolume) for ticker in universe ]
        self.SetBenchmark("BTCUSD") 
        self.SetWarmup(1000)
 
    def OnData(self, data):
        
        for pair in self.pairs: 
            if not pair.rsi.IsReady:
                return
            
            symbol = pair.symbol
            rsi = pair.rsi.Current.Value 
            wilr = pair.wilr.Current.Value
            
            if self.Portfolio[symbol].Invested:
                if not pair.Investable():
                    self.Liquidate(symbol, "Not enough volume")
                
                elif wilr < self.wilrExit:
                    self.Liquidate(symbol, "below threshold. Getting out.")
                continue
            
            if not pair.Investable(): #if the pairs tradingvolume is too low
                continue
            
            
            if wilr > self.wilrEntry and self.Portfolio.MarginRemaining > self.positionSizeUSD: 
                self.Buy(symbol, self.positionSizeUSD / self.Securities[symbol].Price)
                self.Log("Yummy, yummy, cryptos for 99!")

class Pair:
    def __init__(self, algorithm, ticker, minimumVolume): 
        self.symbol = algorithm.AddCrypto(ticker, Resolution.Hour, Market.Bitfinex).Symbol
        self.rsi    = algorithm.RSI(self.symbol, 336,  MovingAverageType.Simple, Resolution.Hour)
        self.volume = IndicatorExtensions.Times(algorithm.SMA(self.symbol, 720, Resolution.Hour, Field.Volume), 
                                                algorithm.SMA(self.symbol, 720, Resolution.Hour, Field.Close))
        self.wilr = algorithm.WILR(self.symbol, 504, Resolution.Hour)
        self.minimumVolume = minimumVolume
    
    def Investable(self):
        return (self.volume.Current.Value > self.minimumVolume)

edit / update:
it seems as there are some things to consider when live trading on crypto exchanges (in this case bitfinex).
The algorithm from above now shows a negative btc balance while ignoring an existing xmr balance. Something that is simply impossible on a spot trading account. 
Maybe somebody has experience with live instances on bitfinex and would like to give advice?
qc live instance:

bitfinex:

Hi Andreas Dekrout,
For issues regarding live trading, please get in touch with support@quantconnect.com.
Best,
Varad Kabade 

Hi

I changed my username since last time (Ya1). Please do you know if QC allows trading crypto CFD on OANDA or FXCM like any other CFD on these mainstream forex exchanges?

Thanks

Hi,

is there a easy possibility for this algorithm to plot the RSI Indicator for all symbols in one chart? 

Thanks and best regards

Marcel

Hi again :)

Attached you find my backtest for now. 

So how can I can I plot the rsi, rsiema, band1 and band2 for the 3 tickers in one chart?

The second question may be a bit easier, but how can I log the self.pairs values, which should contain the 3 tickers? 

Thanks for your help everyone :) 

Best regards

Marcel 

Hi Marcel,

concerning the plotting you might want to loop over the symbols in onData just like in Louis code above (for pair in self.pairs: ….). The same goes for Logging those values.But practically you will very soon run into the 400 max datapoints limit for plotting and/or the max 10 series per backtest limit. Also there are limits on logging, depending on your subscription.best regards,Andreas.

Hi Andreas, Thanks for your feedback. Since I'm still relatively at the beginning, I wanted to check with the graphics whether I got the correct values ​​that I would like to have. I can now draw the graphic, thanks for your help! Unfortunately I can only display it individually, do you have an idea how to put everything into one graphic for the 3 coins BTC, LTC and ETH? I deliberately only selected 3 coins, here I should be able to work with them without the limitations, if necessary I'll only use 2 or 1 coin :) Thanks! best regards Marcel

Hi AlMoJo,
We do not support crypto cfds.
Best,
Varad Kabade

Hi Marcel!

When Plotting the first argument is the chart you are plotting to, the second argument the name of the series and the third argument is the value to plot. 

In your case I would suggest using something like 

self.Plot("My awsome chart", symbol + “ Price”, price)
self.Plot("My awsome chart", symbol + “ BB”, bb)
...

Hi Andreas,

Thanks, that really helped me!

Marcel

Hi,

Do you know if there is any plan to add more crypto pairs to CoinBase? Currently only 9 pairs are supported.

Thanks

Hi Francesco!For a full list of pairs available on coinbase you might want to take a look at the discussion at: 

 and the full list of supported pairs for gdax (aka coinbase) in the csv posted in that discussion (seems, I cant post the link to the csv here).Anyways, there should be 101 pairs available to trade via qc.

Thanks a lot Andreas! 
I was not aware of this post and I kept referring to the documentation. Glad I asked!

Some caveats for live trading:

• use the CashBook for querying amounts
• use a Custom Liquidation function that also uses the CashBook, you need to be aware of lot_size (allowed precision), stepsize (derived from lot size, but might be different for some coins) and  min_notional (currently 10 USD min trade)

def LiquidateCrypto(algo, symbol):
    crypto = algo.Securities[symbol]
    coin = crypto.BaseCurrencySymbol
    quantity = algo.Portfolio.CashBook[coin].Amount
    lot_size = crypto.SymbolProperties.LotSize
    quantity -= quantity%lot_size
    min_notional = crypto.SymbolProperties.MinimumOrderSize
    if abs(quantity) * crypto.Price > min_notional:
        algo.MarketOrder(symbol, -quantity, asynchronous=False)

On an algorithm like this would there be away to set a moving stop lose incase of market crash  such as 

StopLoss = StopMarketOrder(Symbol, -quantity, currentPrice * (1m - StopLossPercent));

then update the stop loss if price is increasing such as

if (currentPrice > highestPrice)
 {
     highestPrice = currentPrice;
     StopLoss.Update(new UpdateOrderFields{ StopPrice = currentPrice * (1m - StopLossPercent) });
}

would you have to hold an array of invested symbols with stoploss and current price or is there a built in function to do this? I was thinking in the crypto space putting a stop loss at 5% might catch some of the volatility that comes with crypto.

also at what resolution is this running? I notice we are not setting one.  

Hi Brian Colliflower,
We can store the stop-loss tickets in a dictionary keyed by the Symbol object. We recommend going through the following bootcamp lesson for more information. Please note that if we do not specify the resolution it is set to the default which is Minute.
Best,
Varad Kabade

Santa24 : thanks for sharing the code on custom liquidation. I've just run into this rounding error in my live algo. 

I tried the code you shared above, and in my backtests, it's not quite working as expected – it leaves tiny amounts in my cashbook (eg like 0.00001), which messes up my logic (my algo still sees that we are ‘invested’ so it doesn't take any new positions).

I resolved it by removing any checks for lot size and notional amount. 

Did you run into any similar issues? Where this custom liquidation doesn't completely empty the account of the specified symbol?