Algo trader here. Current job: sell side algo trading

write code to buy low sell high

Do you deal with HF trading? Is it an integral part of algo trading?

maths. studied a mix of pure / applied. my university was / is quite theoretical and so i learn to code only during research and once i started 9-5 work.

avg % a day?

I presume you mean hft shops like optiver, virtu, kcg, imc etc. banks can't and dont compete in this field here in euroland. not sure about usa/asia.

kinda. there is sub millisecond requirements but they're typically not risk on like they are for the groups mentioned above.

enough to entice smart money

When they say fake it till you make it, they probably don't mean larping on anonymous imageboards.

Taking interns? I'd work for free just to learn

for trading/financial models, what do you think is smarter to learn? Python or R?

yes. but not from here. just go ask to meet trading desks and shout them a coffee (and do not work for the desks that don't offer to pay for you).

thanks for your enlightening presence here

i would suggest either. if you want to be valuable on the engineering/production side java / c++ will be useful too

>just go ask to meet trading desks and shout them a coffee
firm handshake tier advice

>Python or R?
Apples and oranges

>java / c++
Apples and oranges

Great thread keep it up OP

So you run your own fund? Assume its long/short? Wasnt clear if you do crypto only or other securities. Indication of AUM? And... indication of fund % irr last year ;) . Sell-side IBD here.

so apples or oranges?

Oops just re-read - sellside. US BB bank?

How much math do I need to know if I'm looking for an internship? I'm a programmer but I'm a brainlet in math. But algo trading is really interesting. Also how would I go about finding a trading desk that trades crypto?

On the sell side... refer to OP who will know more but think you might struggle... also how can you do programming but be Math brainlet? Math is pretty important for anythign in finance. Indeed, almsot every intership application at a bank requiress a numerical reasoning test, so suggest you focus on boosting that.

OP how many bittcorn do I need to make it?

>hey guys, cfo of a Fortune 500 company here. We are about to buy a ton of bitcoin and start a mining operation. AMA!

Not a total brainlet but get C's in my math courses. Pretty much a bit below average when it comes to math.

For your needs: R easymode, Python hardmode, but you'll learn a programming language that's useful beyond math/statistics. If you'll end up wanting to implement your model in the real world, knowing Python will let you alter existing bots like Zenbot(crypto) that's written in Python.

to be honest - self funded. through contacts funding/scale isnt an issue, but for the moment it's just a baby. average is astronomical compared to anything we're doing in a day jobs =]

euro bank here. question for you: do you enjoy what you're doing + banking in u.s? i dont see too much a future sell side...

i'm biased. i did math, and i like those that also did math. i also realise i'm not the best programmer, and so i feel like we're 'a man down' when a programmer isn't around. if you're deep already in programming you'll be valuable. as to finding desks that trade crypto? i only know of 1 or 2 and it's very quiet (deliberately i suppose).

i keep my work and my own crypto completely seperate.

any resources on beginning learning? complete coding brainlet

there are many. a quick google "python/r for finance" should yield a lot.

>Python hardmode
>ever

Attached: brainlet.jpg (1024x853, 118K)

python is a good all round choice.

i would suggest not using any existing bots/libraries unless you know what they're doing.

What type of strategy do you use during downtrends?

Compared to R for "trading/financial models" it is

Can you NOT tell brainlets about algos? Thanks

strategies that don't depend on the price relative to fiat

why?

That's vague

I'm working with a simple bot and was just looking for ideas besides sell and wait

Do cryptos you trade have particular requirements? (e.g. you only trade low/high mcaps, volume has to be sufficient, etc.)

Attached: 1515860966358.png (226x223, 6K)

yes! you'll need enough depth in the market if you want to trade volume.

there is a lot of suspicious activity in crypto space - the recent medium post (don't have link handy) showing the chinese volume / wash trading hapenning.

pick your battles

Do you know of any mainstream institutions / hedge funds doing algo trading on crypto?
I feel like they could absolutely crush the opposition, for a while at least.

What is bitcoin going to do next?

what sorts of returns are you getting, and seeing as it's arbitrage based (statistical arbitrage based, by the sounds of it), do you believe you'll realistically be able to maintain that level of performance as time goes on?

So is your strategy directional and not non directional market making, or is your strategy spreads?

I'm this user quantitative approaches aren't usually about any sort of market making, that gets more into high frequency trading

they're often more based around looking for extremely strong relationships, monitoring for any deviations from that relationship, and then taking positions to profit from the return to a normal state for that strong relationship

look into things like pairs trading/statistical arbitrage, and congregation (example for bitcoin:megatrader.org/en/poleznaja-informacija-en/19-bitcoin-statistical-arbitrage)

where speed becomes an important factor in quantitative approaches is that other humans eventually find out about the same relationship, and it becomes extremely self-fulling (because everyone ends up doing it), other people will eventually find ways to profit off of the self-fulfilment faster than you - and so the edge inevitably fades away

like all arbitrage opportunities, it's a race to the bottom, given enough time

>congregation
*cointegration

Have you ever tried evaluating risks for crypto using VaR? Have you tried operating portfolios?
What can you suggest studying in terms of maths (Markov chains... or anything deeper)

why do you apply anything more complex than symbolic algebra and linear algebra to the simple subset of information: bid/ask volume open buy/sell, and the order book? isn't that all it would take?

what language do you use? what's your typical gains? how much historical data do you feel is important? do you take into account relationships between all coins, or just a few?

How would value at risk calculations benefit a self-funded trader? Isn't the only purpose of VaR to determine whether you have sufficient capital reserves in place to cover losses? If you're self funded it never comes into question.

Great thread, thanks for posting. Wish you'd take my email to discuss interning but I don't blame you for not wanting folks from here...they are mostly poison. I'm still not sure if the amount of learning from here is worth it compared to the oh-so-hated Reddit.

Anyways, I studied math, programming, and engineering in college -- I think I have a good technical background for this kind of thing, but I am sadly lacking on the "trading" side of what a bot would do.

Could you link me to some reputable sources that you think do a great job of discussing trading indicators? Specifically, the trading indicators that you yourself use in your bots. There is so much misinformation and quackery out there in the TA realm that I don't know where to turn.

I'd definitely have to get up to speed with the relevant Python libraries and whatnot but the trading side is nearly a blank slate to me. I've learned a good deal about personal finance since being out of college, and I've learned a lot about market basics from crypto and stocks. But when it comes to deeper trading patterns, reliable indicators, and the true economic "science" to a bot I'm lost.

Cheers, and thanks for posting

Credit Suisse, UBS, BNP, Deutsche
Who are you and what is your ADV%?

second this, I'm 90%of the way there but the last 10% takes the longest. Essentially, you don't know what you don't know and short of guessing, paying larping masterminds, or mining the whole internet for answers, some non-larp direction would be great.

Attached: 1506892221498.png (459x650, 832K)

it's me againthis is a good thread, but seems OP left :(

I believe what you're referring to is "Value at Risk", right?

that other user is talking about something known as Vector Autoregression, it's something different

Sigh. Hopefully OP comes back

obviously i can't know exactly what OP's talking about, and you are unlikely to be revealed an edge from someone else

but if you look at the comment made here >kinda. there is sub millisecond requirements but they're typically not risk on like they are for the groups mentioned above.

OP seems to be playing statistical arbitrage on very small timescales, faster than humans can pick up on