>Not including brokerage fees
Paying $7 a trade is going to fuck you really quickly, amigo.

It has to be that easy since my almost literally retarded cousin has made a 300% profit over the course of the past 3 years on his initial investment... This is all self reported though verbally and havent actually checked his portfolio

But it seems like very dense people can make money trading so long as you dont listen to platitudes

>>1246571
good luck in a real account

>>1246582
>Assuming you know my backtesting methodology
I'm not doing HFT with multiple trades daily, I'm just automating a simple-ass mean reversion. Timeout is 20 days and I am currently shooting for a target return of 4.5 each time I go in.

>>1246590
Of course I'm nervous about involving real money but I'm probably just going to set it up with a live simulation for a while until I'm actually confident enough to let it go live. I am a bit afraid of overfitting to my backtests and I think seeing it operate in real time will help to ease that concern.

>>1246571
How do you decide the trigger and the target?

Surely it can't just be a simple mathematical equatino, right? If it was then how would this be possible with all the competition?

>not using quantopian to backtest strategies

immediately take out as many loans as possible and implement this in a real acc op

go quickly now before someone steals your methodology!

report back here with results in 2 weeks :^)

>>1246613
I was using Quantopian for a little while, I actually used it to test this strategy at first. I just decided I wanted more control (and didn't want to use their API) so I just found a simple Yahoo Finance module for Python.

>>1246610
Right now it's observation and looking at historical data. I plan to make it a bit more involved soon with more data points. It needs to be fitted to each stock; I find GOOG does well with -2 while MSFT needs a -3 to be consistent, for example. I hope to automate that tuning as well.

>>1246617
YES! I was hoping something like this was possible! When it works I will give you 50% cut, friendo :^)

>>1246613
Oh, and does Quantopian have a console? The main reason I wanted to roll my own platform was that I wanted to run some methods without needing to resort to a full backtest. I guess I could switch back if I was more comfortable with the tools.

>>1246571
Backtesting is misleading and generally very shitty. I began programming trading algorithms around 4 years ago and had backtests showing over 2000% monthly gains with no leverage. When you actually implement it you'll realize that the majority of backtest engines are filled with bugs and shittery. Live small capital test or forward test it, anything else may be useless.

You picked two stocks that consistently went up over the last two years and are surprised that you figured out a strategy that made money trading those two stocks?

have you accounted for slippage?

>>1246693
this

or
>>1246700
this
>>1246582
Especially this

>>1246700
They went up, sure, but I'm still coming out a considerable amount ahead of a buy+hold of those two companies.

Wow you made simulated money with back testing on data from literally the largest bull market in the history of the US exchanges.

You're surely a financial genius and a l33t cs wizard. Please start more threads bragging about how you've accomplished fuck all.

>>1246711
Right. If you know the outcome of a roulette spin you make more money betting on the specific number rather than red/black or odds/evens.

Is it possible for a small retail investors to instantly buy and sell stocks? Forex I know you can.

Also OP are you considering monthly broker fees and trading fees?

>>1246571
Where'd you get your historical data from?

msft and goog have basically never dipped below their 2014 prices. a very long term bull market in the entire SPX. test your shit on the nikkei.

i'm sim trading live data manually. have been doing so for about a year. went in for real money. lost like 2-3 grand. about to go in for real money again. you'll feel like hot shit when the market trends. it's when things get choppy your strat will show its worth.

>mfw made 80% in 3 weeks
>lose 70% of that gain in 2 days
>trade again the next week
>so far still positive at +50% 5 weeks running.

my best trading strat so far. if i'm consistent enough, and have enough capital i'll learn programming to implement this. would probably need at least $50K for a nice buffer in case the algorithm goes haywire.

>>1246731
provided certain liquidity thresholds, yes.

>>1246571
>simulation

>>1246571

It actually is that easy.

A few of us hangout in this discord. Come introduce yourself. Ask for Guac.

discord gg/0r6y9uFQmX4TlwF6

File: 37191603.jpg (81 KB, 200x291) Image search: [Google]

81 KB, 200x291

Ctrl-F commission

no results

uh... OP how are you gonna trade for free?
you retarded brah?

>>1246571
Finally someone makes a thread for this.

I'm using the Yahoo Finance Python library to get my historical data, however I am encountering some difficulty in creating a backtesting platform.

The reason for this is that I created my own MS SQL database to store all the historical data for all the symbols for my market (ASX), and that whenever I use pyodbc to retrieve them, I couldn't figure out a way to multiply series and dataframes effectively without doing a stupid for-loop.

I guess what i'm saying is, I'm not at your level yet OP, if you have some advice on this, it would be much appreciated.

>>1247143
So you're not quite up to the level of retard?

OP hasn't even accounted for the most obvious reason his scheme will fail - trading costs.

>>1247146
Hey there, i would assume that trading cost can be added in as a fixed variable, and the number of trades and the size of each trade can be adjusted such that a strategy works.

Granted that this is all backward-looking, and that no strategy should take comfort in historical numbers, it still helps give a rational basis for your entries and exits.