How hard would it be to make a trading algorithm that produces

>>1671254
would be ruined by market instability and political turmoil

i am also interested in that. my conclusion is that whatever safe algorythm you develop will still need daily maintenance at best.

Impossible
/thread

File: 740x-1.png (233KB, 740x616px) Image search: [Google]

233KB, 740x616px

>>1671262
>>1671263
What is pic related?

>>1671254
Very difficult

>>1671274
an algorythm being tweaked daily :^)

noticed the first 2 years. still, interesting rates overall

File: guacaveli.png (488KB, 600x450px) Image search: [Google]

488KB, 600x450px

>>1671254

Took me about four years to do it by myself with a decade of programming experience, but I knew nothing about trading when I started.

Just like discretionary trading, algorithmic trading requires constant tweaking. Most change on their own as they are based on real life variables and not math equations like technical indicators. There are tons of trading institutions that use data mining in trading algorithms. Like they found a specific pattern in different words trump uses in his tweets and they act upon them. They use terabytes of data analysis to make a decision. It isn't just if the RSI goes below 30 buy type of algorithms you think of. They hire highly skilled people with talent to see patterns in anything. They hire astronomy majors history majors Latin majors etc. Its easier to teach these people finance than it is to teach a finance major a talent of spotting things.

>>1671254
You'd probably need to find an algortihm for a non linear stochastic programming problem constrained to a minimum expected profit or minimum median profit of 20% or to define a confidence interval for the profit that goes in your desired range. Academically speaking it would be awesome to see somehing like that.

>>1671274
You know that medallion returns are due to tax evasion schemes right?

https://www.bloomberg.com/view/articles/2014-07-22/senate-literary-critics-don-t-like-fictional-derivatives

File: lolz.jpg (31KB, 573x609px) Image search: [Google]

31KB, 573x609px

>>1671443
>They hire astronomy majors history majors Latin majors etc. Its easier to teach these people finance than it is to teach a finance major a talent of spotting things.

and i thought i have seen it all at /biz/ already

>>1671254
I have a Poloniex Bot I built from scratch that works perfectly but I have stopped using it because I simply can not come up with a profitable trading algorithm. All it does is lose money.

I could easily make an adaptive algorithm that changes itself based on current market conditions. But, I simply can't make it profitable. The price movement seems to be just random.

>>1672426
>Poloniex
try equities. it's easier. trust me.

>>1671987
>can confirm
Amazon likes to hire people with doctorates in shit like economics and stick them into a spot doing data mining on their product recommendation engine. Geniuses are just easy to introduce to topics like machine learning and data mining.

>>1671479
why do you suppose it needs to be non-linear?

in actual fact, it might surprise you, but rather simple linear models tend to perform best at high frequency

>>1671443
>They hire astronomy majors history majors Latin majors etc. Its easier to teach these people finance than it is to teach a finance major a talent of spotting things.

try postgrads in STEM subjects...

>They use terabytes of data analysis to make a decision

nope perhaps to build a model but not to make a decision - decisions are based on the very recent past

>>1671963
You're confusing yourself there - the profit already exited, those trades relate to evading tax on that existing profit

>>1671254
>How hard would it be to make a trading algorithm that produces 20% in profit per year? What about 50-100+%?

Depends on your background, time frame you hold the trades for, asset class etc..etc..

answers can range from easy to impossible... for both 20% and 100%

some strategies scale well others don't but can produce very high returns... for example why do you think that so much high frequency trading is conducted by prop firms rather than hedge funds...

File: 2CtltIdTJx-ptn.gif (3MB, 317x178px) Image search: [Google]

3MB, 317x178px

guys when will dow jones crash?

Been working in AlgoTrading for the past 8 years through good times and bad times all and all its a great challenge and a great way to pay the bills

>>1672851
January 21, the day after trump takes office

>>1671254

Wasn't there a tripfag here who claimed to have made one?

File: Oj5dfB.gif (4MB, 400x168px) Image search: [Google]

4MB, 400x168px

>>1672867
Alright lets do this buddy boy.

>>1671987
https://en.wikipedia.org/wiki/Renaissance_Technologies#.22Quants.22_with_non-financial_background

Interesting, really.

I wrote a trading bot over the summer that has gained ~18% since June with a Sharpe ratio of 1.6. As long you don't have too much capital I think it can be done with a little proper knowledge, although the work required might make in not worthwhile.

>>1673637
STEM post grads not 'History and Latin' majors :D

>>1672867
Retard, it will be March 1st.

Brexit

>>1674334
i dont think you have understood how financial markets work

>>1673667
With a sharpe like that, you should be trading as frequently as possible to get dat smooth equity curve. How many trades/day does your bot do?

I'm about to hook into FXCM's FIX API myself and let rip

Easy. The problem is that it's cost-prohibitive. Having your equipment physically close to the NYSE, NASDAQ, etc. means your trades execute faster, which is critical. High-frequency traders are paying millions of dollars for that luxury. Unless you can compete with that, you can't get into automated algorithmic trading.

Coming up with an algorithm that works for low-frequency trading is of course extremely difficult.

>>1676082
Maybe around 2-5 per day, with a holding period usually no more than a few hours. I trade lower-liquidity stocks, so the number of trades I can do depends on the amount of activity in the stocks I trade. I also picked a fairly high coefficient of risk-aversion in my utility-function, since I rather have very small drawdown than maximize the long term return.

I don't actually think individuals can compete in the FX markets. Too liquid and too efficient to find any edge in any time frame, let alone short-term.

>>1676125
Does geographical proximity mean that much? I mean electric signals travel about 2/3 the speed of light and fibre - well the speed of light.

>>1676125
It's not millions of dollars, you can get a slot in the exchange building for around 3000$ a month plus 1000$ for the connection

File: CIS.jpg (30KB, 467x256px) Image search: [Google]

30KB, 467x256px

>>1676125
You don't need to have microsecond rountrips and nanosecond port-to-port latency to run an algorithmic trading shop. Algorithmic trading is simply trading that does not involve manual human input in normal day-to-day operations.

>>1676241
That's cool, glad to hear you've got small DD

Liquidity is never a bad thing, and I'm not sure I agree that it's impossible to find an edge. There's probably millions of low-capacity strategies that are completely ignored by prop/institutional traders because they're using vastly different amounts of capital.

There's also highly variable liquidity and volatility across different pairs. EURUSD is highly liquid (90% of all forex trades involve EUR or USD) and not volatile at all, while GBPNZD regularly has over 1,000 pip/day swings. There's a pair for whatever your style is.

>>1676852
With too liquid and efficient I was of course referring to pair like EURUSD. I've looked into some emerging markets currencies and they seem a lot less efficient, however I'm not sure how one would go about to start trading them since any spread you would get would be very large. That's sort of my problem with the spot FX market, there is no one market to speak of since you'll always be dealing with dealers who set the price for you.

Anyway, here is an example of a FX strategy I have found that seems to be very successful.

>http://www.qtscm.com/accounts/

>>1671254
it doesnt matter what algorithm you spent your hard labor on, the cia will find a way to fuck it up because theyre jews.

>>1671443
if the algorithm is based off of tweets then america is far below economic standards compared to asia

>>1671263
/thread

10% annual is about as good as it gets, anything beyond that has me skeptical (ponzi, lucky year, skewed data, etc..)

>>1671274
Some random fund no one has ever heard of claiming 20% every single year.

Sure thing mate!

>>1676891
that's bullshit, I know for a fact that you can do a lot more

>>1676996
then when are we going to hear from your fund on forbes?

bernie ponzi offered something like 9% and managed 50 billion dollars. sure yours could manage 100 b

>>1676996
The truth hurts doesn't it

>>1671334
I'm in discord with this guy and he's a liar. Always claims he does but never has any proof and ends up just throwing out trading buzzwords when questioned about his system.

Either trade frequently enough to be part of the signal, or trade infrequently enough to be part of the noise.

Looking in the FX market trying to find a trading strategy based more on qualitative indicators, where would one begin? Ill never be able to out compete in terms of formula but competing on logic is something that should be possible to come up with some sort of momentum based strategy that allows me to make bets a few times a day/week. If all the current focus is in HFT scalping, monthly models, and discretionary trading - what are areas to be explored for edge to be found outside that? Certainly out of 70 pairings there must be short term +EV bets.

>>1677095
>where would one begin
>some sort of momentum based strategy

Ride momentum up and down and don't get greedy. If you can't build the strategy yourself beyond that then you probably should not be pouring time, money, and energy into this. Not being a dick but trading isn't for the faint of heart and if you can't be bothered to do your own research and development then you probably won't make a profit.

This thread tickles my sensitive G-spot. We should band together and try to develop one. Who's with me?

If you're in, which is the best medium to communicate with? Slack, discord, IRC? I'm genuinely serious. Lets do this shit.

It doesn't have to be perfect, but using our knowledge in both programming and the stock market, we might be able to create a nice algorithm. Over time we can refine it to make it better.

>>1677307
anyone with a working strategy will not share it.

File: BIGshortISDA.jpg (29KB, 1200x480px) Image search: [Google]

29KB, 1200x480px

>>1677307
Why would anybody teach some random dumbass their trading strategy? We don't work for the same firm, we don't know each other, we're just 4 guys shit-posting back and forth acting like hundreds of different people. Nothing more, nothing less.

Pic unrelated you goddamn newfag