Equities analysis school project - need help
I am designing an algorithm for classifying stocks. i have math and CS background, not finance, so I need help defining the problem. I will let you guys know how I am thinking, and which are the points where I need help. Thanks in advance, this is very important stuff. Please don't get bugged in details, I just need some values of X and Y that make certain sense, the project is just an exercise anyway, not to be used in practice(you will see what X and Y are)
GOAL: Given a set of stocks, select winners, i.e. stocks expected to go up.
QUESTIONS: How do I define winners?
DETAILS: Ideally, I would like to select stock expected to go up for at least amount X after Y days, long them today, and sell them after Y days for a profit. The problem is:
-
What value of X should I use? I emphasize, this value is human input in the algorithm as parameter. I just need some threshold of X to overcome transaction costs.
-
Same for Y. Given the nature of stocks, how many days are enough for them to fluctuate for some measurable profit to occur?
Please give me some reasonable values of X and Y that make sense.
I'm a computer engineer major (mostly CS) so I understand where you're coming from. I've had projects for my internship where I had no idea what was going on in the big picture but could program what was needed based on equations, etc.
As an active equities/options trader, I have a good basic understanding of investing and this is my take...
Firstly, if there was a general algorithm that predicted stocks that are most likely to go up most of the time, investing would be too easy. Instead, there's several ways people often go about picking stocks.
Technical traders look for stocks based on past prices (charts) and use things such as Bollinger Bands, moving averages, momentum, and MACD. These may be sample 'formulas' you could implement. Google "technical trading"
There's also trading on the fundamentals where you could basically create a stock screener to find a stock with low P/E, high growth rate, etc. Many screeners already exist on the internet (google has one) so this will also not easily get you rich. Most fundamental investors dive far deep into a companies balance sheet and so on.
The most algorithmic and technical trading is quantitative finance (quants) and might be a too complex for the scope of your project.
Hope this pointed you in the right direction..
Can you please tell me how much a stock price should grow so that buying it and selling it will make a profit, even with transaction costs? Just a rough estimate.
Or perhaps let me know how large are transaction costs and how they are expressed(per lot / per share)?
its entirely dependent on the stock price and how many shares you own. if you buy 1 share in citi, with 12.00 transaction costs going in and going out, youll never profit. if you have 1 share of BRK-A, then it will have to go up VERY little in terms of % to make a profit....
come on now, you can make a profit with any stock, depending on how many shares/where its trading. you can't figure out profit????
I am just looking for a number, Can you give me an example of transaction cost value, which is hopefully somewhat representative of the average rather than the extremes?
I would say an average cost for the retail investor on a site like TD Ameritrade, eTrade, etc. is about $10 per trade.
So, 10 dollars to by X shares, 10 dollars to sell X shares.
Yes, thanks, that's what I need. This is only an exercise in machine learning, so I am trying to use some numbers that roughly make sense, not make a profitable algorithm. Also, if anyone can recommend a source of suitable data for this sort of project(other than yahoo/google finance) that would be greatly appreciated.
For $10, that is, with a single trade, can I trade multiple companies, or a single trade is defined to involve the shares of a single company?
Est vel iusto hic. Qui eos incidunt aut quasi. Est aut amet enim nihil sunt dicta qui. Minima consequuntur perferendis hic provident quibusdam enim et.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...