5 min trade bar

Fred Painchaud  HI can u look into this code , i want to set this bot such that when  the price is low  or equal to the historic low price it will buy , but  i also want to add that when people put a lot of sell order the bot should know that it's time to buy and sell when people are buying more.Can u tell me how can i do it ?

Hi Blackpanther (✊🏾),

Sure. Please let me confirm your specs first, mind you?

1- “when  the price is low  or equal to the historic low price it will buy”: you want to use the lowest low over which period of time? It looks like it is over the last 5 days? Is that what you want? Or do you think/want that your algo looks over the last 5 hours?

2- “also want to add that when people put a lot of sell order the bot should know that it's time to buy and sell when people are buying more”: by putting a lot of sell orders, you probably mean that the volume increases but the price goes down, exact? If so, the fundamental issue with that is that you are trading a stablecoin which is not traded on a single market, or at least on one major market where a very very high % of the total world volume is to be found. So volume on FTX is going to be just that, volume on FTX. Volume can go up or down on FTX but in another direction at the same time on Binance, on Coinbase, etc, etc, etc. Moreover, even with traditional equities like AAPL:NASDAQ where volume is somewhat significative, “sell volume” and “buy volume” is a big approximation in lieu of having access to level 2 data with some deep penetration. But well, now that you know the limitations, you can still try it, but informed. So, are you saying you would like the bot to buy when volume is up and price is going down (selling more) and sell when volume is up and price is going up (buy more)?

Fred

Oh and btw, you are targeting some major league here. USDTUSD is pure scalping. Backtest vs live can be very, very different.

Fred

Fred Painchaud Hi , i thought i had set the historical price at 5min  , and yes u are right usdt doesn't get traded on big volumes but that's what is good in it the coin falls and rises up again and again . 

the price fluctuation is stable so what i wanted that when the price is low same as it was before 5min or more i will execute my orders. I know the market is different in outside world but the pattern is same buy low sell high and the usdt is stable most of the time , it is very much predictable than any other coins.

Hi,

Ok. So 5 min period. Your current period is set to 5 days since the first parameter of timedelta is days so timedelta(5) is 5 days. So self.History(self.crypto , timedelta(BAR), Resolution.Hour) returns the hour bars over the last 5 days. I'll change all that.

I didn't say USDTUSD was traded on small volumes, I was saying it is not centrally traded, it is decentrally traded and as such, volume on one market does not mean much. USDTUSD daily trades at a volume > 70B, it's not small to me.

Yeah, buy low sell high applies, as it always does, the issue is to make it happen. Anyway…

You're writing something different now: “when the price is low same as it was before 5min”. Do you want to compare the current price to the price 5 min ago or do you want to compare it to the minimum price over the last 5 mins? I guess it is to the minimum price over the last 5 minutes but confirming since I will work on that in a couple of hours anyway so there's time.

Fred

Hi,

Attached backtest is one version using the previous 1m close to gauge selling high volume and buying high volume.

I also illustrated with the sma of the close over the last 5m as a gauge. That other version is here:

# region imports
from AlgorithmImports import *
# endregion

class CryptoConsolidator(QCAlgorithm):
    def Initialize(self):
        self.SetStartDate(2021, 4, 1)
        self.SetEndDate(2022,  4, 20)
        self.SetCash(1000) 
        
        self.crypto        = self.AddCrypto("USDTUSD", Resolution.Minute, Market.FTX).Symbol
        
        self.min           = self.MIN(self.crypto, 5)
        self.max           = self.MAX(self.crypto, 5)
        self.price_sma     = self.SMA(self.crypto, 5)
        self.volume_sma    = SimpleMovingAverage(5)
        self.volume_std    = StandardDeviation(5)

        self.nextEntryTime = self.Time

        self.SetWarmup(6)

    def OnData(self, data):
        volume = self.Securities[self.crypto].Volume
        self.volume_sma.Update(self.Time, volume)
        self.volume_std.Update(self.Time, volume)

        if not all([indie.IsReady for indie in [self.min, self.max, self.price_sma, self.volume_sma, self.volume_std]]):
            return

        if self.Time < self.nextEntryTime:
            return

        self.nextEntryTime += timedelta(minutes=5)

        price = self.Securities[self.crypto].Close

        if not self.Portfolio[self.crypto].Invested and\
            (
                (price < self.min.Current.Value) or
                (
                    (volume > (self.volume_sma.Current.Value + (2 * self.volume_std.Current.Value))) and
                    (price <= self.price_sma.Current.Value)
                )
            ):
                self.SetHoldings(self.crypto, 1)
        elif self.Portfolio[self.crypto].Invested and\
            (
                (price >= self.max.Current.Value) or
                (
                    (volume > (self.volume_sma.Current.Value + (2 * self.volume_std.Current.Value))) and
                    (price >= self.price_sma.Current.Value)
                )
            ):
                self.Liquidate(self.crypto)

Same performance, i.e., not very good - which is expected. You'll need much much more initial investment in infra and contacts to scalp thousandths of cents.

And another version where you buy when volume and price is up and sell when volume is up and price is down:

# region imports
from AlgorithmImports import *
# endregion

class CryptoConsolidator(QCAlgorithm):
    def Initialize(self):
        self.SetStartDate(2021, 4, 1)
        self.SetEndDate(2022, 4, 20)
        self.SetCash(1000)

        self.crypto        = self.AddCrypto("USDTUSD", Resolution.Minute, Market.FTX).Symbol

        self.min           = self.MIN(self.crypto, 5)
        self.max           = self.MAX(self.crypto, 5)
        self.prev_price    = 0.0
        self.volume_sma    = SimpleMovingAverage(5)
        self.volume_std    = StandardDeviation(5)

        self.nextEntryTime = self.Time

        self.SetWarmup(6)

    def OnData(self, data):
        price  = self.Securities[self.crypto].Close
        volume = self.Securities[self.crypto].Volume
        self.volume_sma.Update(self.Time, volume)
        self.volume_std.Update(self.Time, volume)

        if not all([indie.IsReady for indie in [self.min, self.max, self.volume_sma, self.volume_std]]):
            self.prev_price = price
            return

        if self.Time < self.nextEntryTime:
            self.prev_price = price
            return

        self.nextEntryTime += timedelta(minutes=5)

        if not self.Portfolio[self.crypto].Invested and\
            (
                (price < self.min.Current.Value) or
                (
                    (volume >= (self.volume_sma.Current.Value + (2 * self.volume_std.Current.Value))) and
                    (price > self.prev_price)
                )
            ):
                self.SetHoldings(self.crypto, 1)
        elif self.Portfolio[self.crypto].Invested and\
            (
                (price >= self.max.Current.Value) or
                (
                    (volume >= (self.volume_sma.Current.Value + (2 * self.volume_std.Current.Value))) and
                    (price < self.prev_price)
                )
            ):
                self.Liquidate(self.crypto)

This is NOT financial advice.

Fred

Fred Painchaud  hi , i read your msg just now , sorry for the delay . what i want is that when the price hits the previous low or is near to it buy and when the price increases upto  previous high or near to it then sell.

Hello,

That's what I did on top of your volume signals discussed above.

“Previous low” is the low over the last 5 minutes. “Previous high” is the high over the last 5 minutes.

High volume is modelled as two standard deviations over the mean. Sell and Buy volume are modelled in different ways (see above).

Fred

Fred Painchaud  hi Fred thanks for your reply , i was thinking if candle stick pattern could help in this , If u think they could be useful then tell me ?

Hi Blackpanther,

I'm really sorry to burst your bubble but I would recommend you to look elsewhere in terms of strategy. The issue you are facing is the necessary speed you need to have to both “buy low” and “sell high”. And the fact that the orderbook has already 8M waiting in queue to buy at 1.0000 and 19M to sell at 1.0001. And I'm not mentioning the eventual fees.

Fred

Fred Painchaud  i will search a strategy which will be perfect for my idea , thanks for all the help really appreciate it :)

Fred Painchaud  hi just last question after that I will close the discussion , how can i store price of last candle into variable which will update itself ? 

Hi Blackpanther,

Cool, good luck.

If you want the price of the last 1m candle (since you are on minute candles above), self.prev_price in my code above is already the price of the previous candle - and it updates itself every minute. However, your algo was trading only once every 5 minutes so it still only trades once every 5 minutes.

To keep only one previous price, you simply need a variable and you update it in OnData, after using it (of course, as if you update it before using it, you won't be using the previous price but the current one).

Fred