I was hoping to share my Buy criteria and my Sell criteria to get some insight into whether it needs other consideration.
Looking for Stock Market Data Engineer
- Thread starter TheBulldog
- Start date
Stock selection criteria:
- Gain 10% or more from previous day AND
- Price between $2 - $20 per share AND
- 2M+ Accumulated Volume AND
- Some sort of News catalyst AND
- high price above vwap AND
- RSI <= 50 AND
- current minute volume > 1800
- Positive MACDHistogram AND
- 50 < RSI > 98 AND
- Price Action move of +5% from previous two minutes
- MACDHistogram goes negative OR
- RSI drops below 75 OR
- Price drops below 5% from previous two minutes OR .10 above Bought Price
That looks like Ross Cameron's, aka Warrior Trader from YouTube, COVID strategy. He's not trading that any more because post-COVID markets lost volatility.
Just by eyeballing I would say that you are entering way too late and leaving most of the money on a table, like +10% is very big. Did you try to optimise and see where your parameters create some fitness function stability pools?
Other significant inputs that you need are size of float, that is particularly important for bellow $20 stocks that you are targeting. Float is important but I forgot in which way. Too small float will cause stock to jump as a hell, but there would be no liquidity to exit when you want. Ross Cameron explained it in one of his videos.
Another thing that looks suboptimal is that that type of system works well only within the first 30 min. after the opening, if news event was pre-market. So, I guess you need to include the time that passed from publishing the motivational news item.
As well, you want to know the percentage of institutional ownership in the stock, if that percentage is too high their algos are going to kick in and start adjusting their portfolios.
All this info is readily available all over the place.
Would you mind sharing what data resolution were you collecting from Poligon.io?
Just by eyeballing I would say that you are entering way too late and leaving most of the money on a table, like +10% is very big. Did you try to optimise and see where your parameters create some fitness function stability pools?
Other significant inputs that you need are size of float, that is particularly important for bellow $20 stocks that you are targeting. Float is important but I forgot in which way. Too small float will cause stock to jump as a hell, but there would be no liquidity to exit when you want. Ross Cameron explained it in one of his videos.
Another thing that looks suboptimal is that that type of system works well only within the first 30 min. after the opening, if news event was pre-market. So, I guess you need to include the time that passed from publishing the motivational news item.
As well, you want to know the percentage of institutional ownership in the stock, if that percentage is too high their algos are going to kick in and start adjusting their portfolios.
All this info is readily available all over the place.
Would you mind sharing what data resolution were you collecting from Poligon.io?
Last edited:
Yeah, my post editing time expired.
I forgot to say, Ross Cameroon had several times shown his broker statements, and one glaring thing that is immediately obvious if you freeze few frames is that 99% of all of his trades are executed within just x30 min after the open. from the opening. So, for argument sake, entering such a trade x3 hours after close would be meaninglsess.
You can easily do a histogram and vary entries, say every x15 min., and histogram will tell you what is optimal, without too much over-optimisation.
I forgot to say, Ross Cameroon had several times shown his broker statements, and one glaring thing that is immediately obvious if you freeze few frames is that 99% of all of his trades are executed within just x30 min after the open. from the opening. So, for argument sake, entering such a trade x3 hours after close would be meaninglsess.
You can easily do a histogram and vary entries, say every x15 min., and histogram will tell you what is optimal, without too much over-optimisation.
Thank you for responding.
Whether some of my strategy resembles Ross Cameron's or not, doesn't really matter here. And, while statistically the best trading time is the first three hours of the trading day is true, there are still opportunities for great trades throughout the entire day (Pre-market\Market\Post). My algorithm doesn't care what time it is. If the conditions are ripe, it's going to trade it. I would prefer to stay on topic of my Algo and not debate over these kinds of things please. tyvm
I stream real-time Trades, Quotes, and OHLC from Polygon.io. And I simultaneously perform API calls for RSI and MACD from them as well. Note that while I can call for the RSI and MACD values, the values don't actually change until after the end of the minute. So, a full minute goes-by before I know whether the values have changed. Which leads me to your comment "...you are entering way too late and leaving most of the money on a table, like +10% is very big." I'm confused here. I agree that I'm entering late. Nowhere in my criteria did I say that I wait for +10% before Buying. I do, however, specify "Price Action move of +5% from previous two minutes."
Next, I have absolutely no idea what you mean by "fitness function stability pools." Sorry.
I agree on the float. I will work on that.
As far as the percentage of institutional ownership is concerned, I don't see where that will be of any concern to my algo. If the trade conditions are right, then my algo will buy and sell the same way whether there's high or low percentage of institutional ownership or not. Am I right? There is the possibility that I'm missing something here.
Whether some of my strategy resembles Ross Cameron's or not, doesn't really matter here. And, while statistically the best trading time is the first three hours of the trading day is true, there are still opportunities for great trades throughout the entire day (Pre-market\Market\Post). My algorithm doesn't care what time it is. If the conditions are ripe, it's going to trade it. I would prefer to stay on topic of my Algo and not debate over these kinds of things please. tyvm
I stream real-time Trades, Quotes, and OHLC from Polygon.io. And I simultaneously perform API calls for RSI and MACD from them as well. Note that while I can call for the RSI and MACD values, the values don't actually change until after the end of the minute. So, a full minute goes-by before I know whether the values have changed. Which leads me to your comment "...you are entering way too late and leaving most of the money on a table, like +10% is very big." I'm confused here. I agree that I'm entering late. Nowhere in my criteria did I say that I wait for +10% before Buying. I do, however, specify "Price Action move of +5% from previous two minutes."
Next, I have absolutely no idea what you mean by "fitness function stability pools." Sorry.
I agree on the float. I will work on that.
As far as the percentage of institutional ownership is concerned, I don't see where that will be of any concern to my algo. If the trade conditions are right, then my algo will buy and sell the same way whether there's high or low percentage of institutional ownership or not. Am I right? There is the possibility that I'm missing something here.
"fitness function stability pool" is, say your fitness criteria is profit, but for some group of values of parameters after a small change profit changes very little, but for other group of parameter values after a small change profit changes a lot. So, the first group of parameters would be considered stable, and the other parameter group would be considered unstable.
I am not fully familiar with your setup, but if you are getting RSI and MACD from Polygon, maybe they don't let you vary parameters. These x2 functions are very easy to calculate, and if you calculate them yourself you will be able to see where the "fitness function stability pools" are located in your parameter space.
Unless you are coming from finance background there are lots of surprises waiting hidden for you in markets. So, its better to check things even if they don't make sense for you now. That way you will make big and hopefully profitable discoveries.
I am not fully familiar with your setup, but if you are getting RSI and MACD from Polygon, maybe they don't let you vary parameters. These x2 functions are very easy to calculate, and if you calculate them yourself you will be able to see where the "fitness function stability pools" are located in your parameter space.
Unless you are coming from finance background there are lots of surprises waiting hidden for you in markets. So, its better to check things even if they don't make sense for you now. That way you will make big and hopefully profitable discoveries.
I’m an algo trader myself and use python for my own trading but I trade in different market . But I’m sure I can do what you need , all sorts of backtest with different parameters (Keeping aside some chunk of data ) to validate that strategy is not curve fitted. once that’s done then you can automate it and run but it . It’s not going to be easy or a 45 degree curve though that only happens in dreams . But if your strategy works I can test and deploy it for you .
Hey joker, I have a working system, only makes a $million a day, can you test it?
I can test it sure , but I’m sure it doesn’t work . Give me your criteria etc