Daily Candlestick Strategy Performance

[MatHatter]

Hi,

So within the past 3 days I was on a mission to recreate my Reversal Candlestick Strategy I had made for TD Ameritrade API with python to connect to the API and deal with the if statements (Windows MS SQL for selecting the correct orders for something) and then used Windows Task Manager execute the script. I was logging my results with a stop in there, and it seemed great, but when I ran it on here, I quickly realized I may have been wrong about something. Doesn't matter. I used the backbone of the strategy and was working on building a strategy on /ES funny thing is it works better on /NQ.

The thing about this strategy is that I have multiple buy and sell conditions and when I tweak one thing, it significantly tweaks the results of everything else so it is kinda like luck, although my intention behind it is to capture the moves I see, but so many things are effected based on one simple change.

Anyways, I really don't know what are good results, but I know this is better than before because I would get a lot of red sometimes based on the time frame, so I extended my timeframe to capture as much as possible.

I understand the importance of not overfitting, but even so it is not even possible because one little thing effects it so much. The funny things is I was actually trying to get the code to buy on market open without knowing the mechanics of NinjaTrader not allowing that unless you enable Tick Replay and all the query I went through as you can see in my other posts.

My question is are the /NQ results good enough you think? Notice how my stop is at 0.014, which is relatively conservative I believe as this seems to come out to 1.4%. Moving forward, I hope that I can improve /ES and that will translate into /NQ. I can clearly see there needs to be work somehow done on the short side, but again everything is so sensitive. Both either way I want to get that drawdown to at least 30k, $20k - $10k more comfortable for me.

These results don't par with /CL as I find the candlestick behavior must be way different as it intrinsically has a lot more volatility, but I did see an improvement in performance when I increased the stop loss. Nonetheless, I would like to try it on /RTY or stocks, but I haven't tried Russel yet and the stocks don't see to work, which I don't care because I only want to do futures.

/NQ/ES with one of my sells removed problem is there is a big dip in the chart so I am still going to plow forward (as I have been stuck on this buy market open tick replay stuff in the past few days)





From,

MatHatter

edit: simply by changing > to >= in a buy I got this for /NQ $363,090

now with changing that in a sell as well

[MatHatter]

HEY!

turn out my effort to get current market price trade on lead me to strange results that was giving me actually what seemed like strong results, but when printing to the output I could see the current day kept printing the same strange 2600 number over and over again.

this ===> I completely removed from my code now:

// Get the bar index for the current trading day
DateTime currentDate = Time[0].Date;
Print("Current Date: " + currentDate);
int currentDayBarIndex = Bars.GetBar(currentDate);
Print("Current Day Bar Index: " + currentDayBarIndex);
int firstBarOfDayIndex = CurrentBar - (Bars.GetBar(Time[0]) - currentDayBarIndex);
// Get the open price of the current trading day
double currentDayOpen = Open[firstBarOfDayIndex];
Print("Current Day Open: " + currentDayOpen);


and I just use follwingCandleClose (previous day). I prefer less profit but I know what is going on and you can see that the number of trades went down by half!

/nq



/es



I have incorporated RSI, MACD, and EMA's RSI and MACD did give good results, but my strategy is better without them. I really was just trying to focus on price action in this strategy anyways.

Cheers!

MatHatter

now:

/nq

8/16/2024
drawdown down profit up # of trades not too bad

[NinjaTrader_Gaby]

Hello MatHatter,

Thank you for your post.

While our support team can't comment on the performance of your strategy, this thread will remain open for any members of the community who may want to provide their thoughts, feedback, or suggestions.

[MatHatter]

Hello NT Community,

This is where I am leaving it off and going to just run it. I have ran it on minute charts and I profit well or not well on some days. I can see I need some new buy algorithms for the minute chart. For my Daily Reversal Candlestick Strategy I ended up with 5 Buy algorithms and 9 Sell algorithms. I tried to weed out some bad sells but it effected way too much. I even implemented a new buy right after these bad sells, but it still didn't show overall high performance, so I just kept it. I am pretty content with the results though, pretty stoked, so I don't care. Ideally, I don't want it overfit, but honestly, it can never really be too overfit unless if you make the buys and sells so specific and I was really trying to just capture the essence of the candlestick patterns. I can't wait to run it in real life and modify it now for the min charts! It is going to be my first main algorithm for my futures fund, let me know if you are interested in investing. Take it or leave it. I am by no means pressuring anyone to do this and this is not my intent on being on the forum. I am just saying that option is available to you as I am securities licensed and of course there are no guarantees of profit - you invest at your own risk - I would have to offer you a disclosure document.

I hope to make this algorithm go > $1M when I work on it again within this time frame. There is a sweet spot with this algorithm as one little thing effects so many other things. I am grateful I even got it to $900,000 and I actually was working off it with $887,000 because I believe I needed to ensure there was safety in one part, so I weighted the safety more than the profit, but I was able to tweak it to get it back to the NET CUMULATIVE PROFITS high, so I was happy.

I will be working on an intraday algorithm with an old saved version of this script that performs better than this one and hope to run both everyday!

After working on the intraday algorithm, I want to rehabilitate a double Bollinger band TD Ameritrade script I made a long time ago, it looked really promising, interesting, and fun.


Naming this Dragon. Unfortunately, when applied on /ES it is not as great as it was before, so I would have to use the old saved code and maybe add some newer code, but adjust a few numbers, because as we know it /NQ has a volatility greater than /ES as we know it margin is always more from my experience which is calculated by the exchanges. The results never pared well with Russell, because I think different indexes behave differently as I mentioned about /CL.

From,

MatHatter

Update 8/19/2024

so I was working diligently on my intraday algorithm and I noticed the first buy called buy was no good when I was getting a succession of smaller candle bodies leading up to the previous day. I put in the negation statement and now:




Cheers!