Thanks a lot for your input, the code looks nice, I will pin your comment for other viewers in case they need the improved version.

hi im newbie for python. would you help me where input this code? thanks much!

can you help for me how i can replace these code? I think I place the wrong place and it crashed

@@tamaspete8299 # Determine VWAP signals for row in range(backcandles, len(df)): upt = 1 dnt = 1 for i in range(row - backcandles, row + 1): if max(df.Open.iloc[i], df.Close.iloc[i]) >= df.VWAP.iloc[i]: dnt = 0 if min(df.Open.iloc[i], df.Close.iloc[i])

Hi, thank you for your comment. Indeed I didn't want to make scalping video at first because it's challenging the data has too much noise, but this was requested so many times, and I think the VWAP is really good and also the BBands help define a good entry point. Good luck to you! (actually it has been a while I haven't looked into lower timeframes maybe I will revisit scalping soon).

Thank you for your supportive comment, it's nice to know and very encouraging! Good luck!

Happy you liked it, I like the vwap recently it's a nice indicator

Probably one of the most rewarding/satisfying comments! Glad I could help! Good luck!

@@thinketh2408, I see. So, first and foremost, the trading platform I am using should be allowing (providing api). Only then I can automate the process via building/using a trading bot.

Yes it depends on the platform they usually provide a python interface that you can add to your program and automate the trading process, I have an old video on this using oanda s API.

@@CodeTradingCafe , thx for the reply. I will definitely check it out!

@@dxtrmst1618 I think it is being rejected when spread is high

@@CodeTradingCafe suggest

Thank you for your feedback, just be careful with the drawdown keep it on your radar.

👌 Did you add technical indicators or just adjust the values for the given tis?

​@@troygomez4089 just adjusted the parameters: rsi length: 16 bbands length: 15 bbands std: 2.0 back_candles: 26 total signal buy below rsi: 46 total signal sell above rsi: 59 atr index: 7 slatr: 1.2 tpsl_ratio: 1.89 close long above/equal rsi: 90 close short below/equal rsi: 10

Thank you for sharing

@@troygomez4089 also had a typo ... used BBL instead of BBU to confirm the total sell signal... but correcting it decreased the success in return and win rate ... I guess the other parameters would have to be optimized again - although still 1300% return

Thank you for your support and encouragement 🙂

Thank you for your support! You're right it is worth trying an optimization see what it gives, I am not sure how much improvement there will be.

Thank you for your support 😊

Thank you for your support!

Thank you for your comment, harmonics are not reliable, just my opinion (and my experience) but I might be wrong.

Support and resistance levels are crucial, I agree and the trading time wasn't included yet so trades are opened anytime the market is opened this can be improved definitely.

​@@CodeTradingCafe I use volume data to confirm signal above or below vwap. I want to see heavy buying or selling volume moving the price action in relation to time. I use at least 2 or 3 time scales on the chart such as (1, 5, 10, 15, daily). I only focus on trading 1 stock/product at a time and get to know how it moves and try to understand the time frames that larger investors make their trades. It's always interesting to see how light volume moves the price after these heavier volume periods. Everything you discussed in this video is pretty similar to the strategy I use to scalp with.

Thanks for watching! and for your support!

Thank you and good luck with your coding!

Thank you much appreciated!

Hey, thank you for your constant support, glad the vid is helpful

Hey, thank you for your comment, I will try weighted moving average see how it goes. Good luck!

Glad it was helpful! and thank you for your support. Please keep in mind this backtest doesn't include commissions so the trade management part need to be human controlled. On the other hand scalping is way more difficult to automate than higher timeframes... in my opinion but I might be wrong. Good luck!

@@CodeTradingCafe how much can you realistically get in return using longer timeframes?

Around 8-20 % per year all costs included (my experience, but maybe other algo traders do better)

@@CodeTradingCafe damn not as much as I thought. I thought it would be like 2-5% per month at least. Thanks for your insight tho!

If you have time to do hybrid trading, meaning an algorithm will detect patterns and send you notifications but then the trading is done manually you might reach 5% monthly with good trade and risk management, but this requires a lot of waiting time as well.

Thank you for your supportive comment. Sentiment analysis is on my list at least to show how it can be implemented, the only challenge is that you never know when the sentiment will be reflected on the market in a day, a week or more, but I think this is something we can discuss when I make the video. Have great new year!

Thank you, yes I was thinking m15 might be a good candidate... There only one way to find out

Thank you for your support 😊

Thank you, to use a strategy live we can link it to a broker API, check this one: kzfaq.info/get/bejne/jcmWfsSLsZraoHU.html

Thank you for your support ! Glad you enjoy the learning process.

You can see this videos, a very elegant way to use the BB. 👌👌 kzfaq.info/get/bejne/hsmGe7epmq69p5c.html

Thank you for your support and encouragement, good luck!

Thank you for your support 😊

Thank you for your support, good luck!

Thank you for your support, your idea is something on my list I always vouch for a simple approach. Good luck!

Hi thank you, I will look into this for the coming year.

Hi, thank you, I haven't tested the robustness of this strategy yet, I must say it's easy to get lost in the numerous ways one can trade and always trying to choose the best.

Great that you liked these videos. Yes the spread can be included as 2/1000 commission (depending on broker and asset) it's not very dynamic but the approximation is ok. And yes you can apply hedging using backtesting I did it in one of my recent videos on trading the grid system where we buy and sell 2 positions at the same time.

Thank you for your comment, don't forget that fees were not included here so initial capital might not be very accurate

@@CodeTradingCafe Thanks for replying. Will you consider making more videos on intraday and scalping strategy

I will try but lower timeframes are more challenging for coding it's much easier on daily timeframes, but yes I will keep this ongoing see if we can sort a nice strategy

Hi, thank you for your support, my data is rom yfinance or dukascopy (free data), you can also download from your broker. I will make a video maybe this week showing how to download data, I am getting this question a lot, so probably worth it :)

Thank you for your support!

Thank you for your support, I usually only buy what the signal indicates, I don't know if this answers your question

Yes robustness must be tested as well to find the best parameters set, it's a start base as it seems. Usually scalping strategies are hard to tune I was happy with this one.

Hi thank you, at some point yes I will make a demo trading live but I am still tuning strategies...

Hi, thank you for sharing, but the ATR is always below the price value, it's just the range, can you be more specific?

Thank you for your support!

You're welcome! Thank you for your support.

Hi thank you for your support, there is only one way to find out ... Backtesting, and it might need some parameters tuning depending on the traded asset. Good luck!

Thank you for your support!

Hi thank you for your comment, this is python notebooks so you need jupyter notebook installed, try installing anaconda (see my very first video of this channel)

@@CodeTradingCafe great bro, you just gained a subscriber and python student. where I enter the broker's data to do the tests, thank you very much. Greetings from Venezuela

Thank you 🙂

Hi, thank you for your comment, I agree the optimize function is important to show I will have to make a video about it soon.

Thanks for the help. Waiting for the same

Thank you for your support! yes it works on any chart because it purely technical, but the results might be different, and you definitely have to tweak the parameters.

Hi thank you for your support, which divergence?

RSI regular divergences. Thank you so much.

Thank you, glad you liked it!

@@CodeTradingCafe It's a treasure ❤️.

Let us know how it goes, thank you

Hi, I have to recheck how backtesting library is calculating the commission fees when we use for example 0.005 (the parameter in the backtesting function we are using), so far I didn't include the fees because I wanted to show indicators effects and compare among different indicators combinations.

Hi thank you, yfinance or dukascopy both provide data.

Hi, we used the moving average in other videos, this one was for vwap and it wasn't as bad.

Hi, actually it is in days, and yes it's a lot but the test was on 3 years I think. So the system might work for a year then stop working for a while or so.

actually my code would only look at 1 vwap value for each window, but surely there is another way to do it right

Hi thank you for sharing this, actually yes we can vectorize most of the computed values. But just for the sake of this discussion, sometimes if you are streaming for a live trading bot it's easier to have a function to call for each candle (can't explain clearly in the comment only in some particular cases of live trading it works better this way, but definitely slower).

​ @CodeTrading Ah yeah I understand, you get a live feed candle by candle, and you want to pass to a function only the last n candles to calculate just the very latest vwap signal. Anyways, for anyone that wants the vectorized form for doing parameter optimization, I finally came up with the solution: downtrend_mask = df[["Open", "Close"]].max(axis=1).lt(df["VWAP"]).rolling(backcandles+1).sum().eq(backcandles+1) uptrend_mask = df[["Open", "Close"]].min(axis=1).gt(df["VWAP"]).rolling(backcandles+1).sum().eq(backcandles+1) df['VWAPSignal'] = 0 df.loc[downtrend_mask, "VWAPSignal"] = 1 df.loc[uptrend_mask, "VWAPSignal"] = 2 --- The +1 after backcandles is to count the current row AND n backcandles (same effect as what is done in the iterative version done in the video/jupyter)

@@CodeTradingCafe Today I learned that you can just add the decorator @njit from the numba library to your function and will probably run faster than my complicated vectorized stuff. Might need to pass the pandas columns as numpy arrays but @njit takes the iterative loop, does just-in-time compilation of it and efficiently executes it, and can do it in a vectorized way

@@lautaa33 Hey thanks, I didn't know this, have to check it out!

Thank you for your support, you are right ideally we can fit on 2 month and trade one week for example, then slide our fitting window for a new trading period, I will include this in one of my future videos.

Hi, yes we close the trade if the rsi goes above 90 for long positions and below 10 for short positions, since the rsi will show the overbought and oversold ranges of an asset.

Hey, thank you for your nice comment. Which timeframe are you using? The vwap only works on short frames 5min or 15 min are good.

@@CodeTradingCafe I am using a 5 min timeframe, but I realize now I was only using a day's worth of data.

@eminshall did you find a solution to this problem. I am stuck at the same and I can’t use this strategy in a live account as a result. @CodeTrading are you able to share any insights on this issue? Thanks so much for taking your time to share these ideas in a very clear and concise manner.

@@bretgratt9568 hey, how is it going??

Yes it can be tuned this way but vwap is not applicable for higher timeframes so maybe we switch to different indicator.

Hi, yes if you use these algorithms just for signaling, but then you as a human trader validate or reject the entry signal, plus you manage the trade manually you will get better results (around 25% annual, only my estimation other traders might be better and provide 50% or more... not me). So basically the algorithms are only here to do the waiting on your behalf and when the pattern is detected it becomes your job to trade.

Hi thank you for the update, it might need some tuning, mainly the SL TP ratio and SL position, it should be fine for some set of parameters.

Hi, thank you for your comment, both Python and MQL need algorithmic skills, I personally prefer Python it gives more control and you can use it in different fields.

@@CodeTradingCafe Thank you sir for the feedback. Can python create MT5 EAs too so it's easy to attach in MT5 platform? Is it possible or what is the workaround for that?

yes@@wizdeasis9846

​@@wizdeasis9846you can use python to connect to mt5

Sure it's possible to add fees, backtesting library can do this but I don't know how it's counting fees in the background, or we can do it manually as well. Connection with the broker... I have to make a new video about it, in the meantime check this one kzfaq.info/get/bejne/m5eFp5d6xtaVYIE.html

Hi you can use dukascopy they have a lot of data, otherwise yfinance is faster and easier to get.

It's relatively simple but the problem you will have to deal with the names of the Bollinger columns created by pandas_ta for each set of parameter. Other than that putting everything into functions and calling an arching function to execute all the files can work well.

Hi, Dukascopy is a good source, also any broker with a demo account can have access to historical data.

This is a python code it will not run in mt5, it has to be rewritten in mql5 language.

it's on my list, but it might take me a while to reach it...

It might a variation, you can code your own as well if you need a specific VWAP. I hope this helps.

Thank you for sharing, it's nice to know how it works on different assets. Good luck!

Hello, may I kindly speak to you

Hi, no need to use tradingview it's all through Python, and this is mainly for backtesting ideas.

Hi, thank you for your support, the classes are from a package called backtesting py you can install it using pip install backtesting and their web documentation is on git kernc.github.io/backtesting.py/ I hope this helps, Good luck!

Hi, you have a good point! the stdv is computed from the moving average, but also the VWAP is used to test if consecutive candles are above or below the curve, the stdv is used only on the current candle. I hope it's a bit clearer.

My favorite is using VWAP and wait for the price to bounce on after completed retracement, 5min or 15 min timeframe (I prefer 5 min) but in general I avoid scalping it's tiring for me, I start making mistakes after 1 hour of scalping.

@@CodeTradingCafe but arent u using this trading bot??

Hi yes you can download different dataset in csv format and use it for backtesting. Usual data sources are yfinance or dukascopy.

it seems that you are trying to plot data with only 2 values and 1 index value, I don't where the mismatch is coming from, are you sure bt.run() completed successfully?

I also have the same problem, after run 'bt', it shows results, but after line bt.plot(show_legend=False) it shows error: " Data contains too many candlesticks to plot; downsampling to '1H'. See `Backtest.plot(resample=...)` ..... (many lines) ValueError: Length of values (2) does not match length of index (1)

that's one variation we can try maybe I can do it in a future video

Hi, sure, check the link in the description of the video. Good luck!

Thank you for your support, yes you are absolutely right regarding the fees, and becomes more challenging for lower timeframes, it's somehow easier on daily and 4h timeframes.

I mean, there were 2076 trades with 197 dollars of income (197%) so You can calculate that for every trade there was about 9 cents made - commision parameter is set to 0, so I think that no broker has less than 9 cents fee per transaction, so the strategy is tragicaly bad, but video and explanation is good keep it up

some brokers dont charge any fee for trades@@majkel1381

I liked this one too as a starting point, just don't forget fees are discarded here, but it's a good base strategy to develop...

Hi, I get this question a lot, short answer yes you can but is much less than people think (8-14% per year). If you want more return you can use algorithms to send you notifications and execute the trades yourself and manage opened trade manually this is more profitable but also comes with more effort. Good luck!

Hi thank you for your comment, I am not sure I understand what you mean by signal sorting.

@@CodeTradingCafe I reviewed the video carefully... The question is removed... I was a little mistaken..

Glad you worked it out, good luck!

I think yes, I haven't tried OKX since crypto crumbled, but I guess their API should be easy to code.

A full video about it just for you :) kzfaq.info/get/bejne/r9uehLdmnp6aZmg.html

@@CodeTradingCafe thanks appreciate it :)

Good point! there should be, but so many ideas came up and never got back to this one.

Hi, well it's kind of easy simply comment the lines where sell positions are opened and erase the SL TP from the function, but how will you backtest this if it never closes a position

@@CodeTradingCafe To close the position use another indicator, why i don't use SL and TP because after so many tests me and my best friend the 'AI' 😅 we find out that it's not good to use it, """ I don't encourage anyone not to use SL and TP, do your own research and testing , good luck 🤗 """, i use several vwap z-score indicators and other indicators like sma cross for example to open or close the position I try to do it with the backtest library and I don't have the same result with my back_test script.

Ah ok I understand, thank you for your reply!

Great to know 😁

Hi thank you for sharing, yes I agree with you actually I haven't introduced any proper trade management yet in these videos, I will get something ready showing the different ways of closing a trade, your percentage idea is also a good one! Good luck!

Hi you can install it using pip, pip install backtesting

Hi, yes you can use it with any coin or stock as long as you fit the parameters on the traded asset. You can use it on Binance account but you will have to replace the functions using their Python API docs.

Hi, thank you for sharing this, it has been proposed before, I think it's a good idea to link Python and MT4/5 I will add it to my list.

Hi this is a Python language code it doesn't run in metatrader, it has to be interfaced with your broker API depending on whether they are offering one. Check this video it might be of interest kzfaq.info/get/bejne/jcmWfsSLsZraoHU.html

you're welcome

Hi, really sorry but working hard all day now I I am waiting for holidays for a new video.

Hi, depending on how much risk you want to take, you can have extra safe strategy but only 1 percent return per month or very high risk and more return. All the parameters will change accordingly.

@@CodeTradingCafe thanks mate... I wanted to know you preference regarding this... Like what should be the minimum sharpe ratio and maximum draw down before you start using the strategy for real trading...

I would say Sharpe ratio minimum 2 preferably 2.5 with spread fees included is a good start

@@CodeTradingCafe thanks

Hey thank you, different ways you can use yfinance or dukascopy among others...

@@CodeTradingCafe Thanks for the quick response!

@@CodeTradingCafe yfinance let you download 5-min timeframe? Very nice strategy by the way, it will be interesting a video with a platform like mt5 and his api implementation! I wish you all the best!!!!!!!!

@@rubenghidanac Hi, thank you for your support. Yes with yfinance we can download 5 min timeframe but I think only 3 months at a time, so we can put it in a loop and concatenate the 3 months data, didn't try it but it should work.

Hi thank you, can you send me an email with a photo example? I don't get the "next candle start trading below low of signal candle" part.

Hi check my other videos where I am generating signals in both directions you can do the same here.

I like it and it's easier then another method I published here for support and resistance. Thank you

@@CodeTradingCafe Thank you, too. I will be waiting for your next video.

I agree, this is just to compare indicators performance rather than a full strategy, if you are keen to add spread into this a quick way is to apply a 2/1000 commission rate, it accounts for spread (but not for commissions that are broker dependent). Thank you for your comment! Good luck!

Hi, I tried in the past it was too much to manage, I find the comments section here easier to follow