Case study: Algorithmic trading with Go • Hacker News Board

Re: Case study: Algorithmic trading with Go

#211

Post by mhh__ » Sat, Jul 01, 2023, 12:58 AM UTC

IshKebab wrote: post #121

RomanPushkin wrote: post #103

Go itself, or any other language won't give you too much of an advantage. What gives you advantage is trading algo, which is always hard to find. I've spent months on figuring out the best parameters for trading. Ended up this working only on historical data, while in reality it was totally different. I could use Visual Basic, and it would be better than Go, Rust, or whatever it is out there, given the algo and strat…

You overfitted. That can happen due to human optimisation as well as computer optimisation. After you have experimented on some historical data for a while it becomes less useful.

Or the model was crap.

Re: Case study: Algorithmic trading with Go

#212

Post by _zkyx » Sat, Jul 01, 2023, 12:59 AM UTC

jjice wrote: post #182

As a layman boglehead, what are the returns of doing trades like this as opposed to an index fund? I assume it either works or there enough money to hope it works, but really curious what the situation is.

This whole project started when I was looking for an investment adviser and didn't know anything. I was looking at all these funds that had like 4% return per year and then looked at stocks like Tesla that were up like 45% return and was like this doesn't make sense. When you dig in you see that these index funds are super super low risk. However, I like extremely high risk so I was willing to explore this a little. I don't really want to say you can get better returns, because you could end up losing all your money, extremely quickly betting on single stocks, it's basically a sure thing. So, that's why I turned to building a tool that basically bets for you. So, it's more about risk. If you put everything you had into TSLA you'd be up like 50% just in the past month but that's risky as hell. So, it's more of a risk question than a return question.

Re: Case study: Algorithmic trading with Go

#213

Post by _zkyx » Sat, Jul 01, 2023, 1:01 AM UTC

lost_tourist wrote: post #209

Does anyone know if he beat the market with this and became a multimillionaire?

Not yet :)

Re: Case study: Algorithmic trading with Go

#214

Post by TheAlchemist » Sat, Jul 01, 2023, 1:01 AM UTC

maest wrote: post #163

finexplained wrote: post #150

Earlier quoted context omitted.

I work at an HFT firm. Most fun I've ever had.

But is that because of the excellent WLB and pay or because of the social impact?

The jobs in algo trading are very interesting for technically - mathéamtically inclined people. It’s really one of those fields where you have a direct impact on the results of your - measurable in additional dollars made.

There is 0 social impact. That’s the downside of course - but hey, how many jobs out there are really having any kind of positive social impact ? Not 0, but close to it.

Re: Case study: Algorithmic trading with Go

#215

Post by cratermoon » Sat, Jul 01, 2023, 1:03 AM UTC

justinhj wrote: post #195

cratermoon wrote: post #185

Earlier quoted context omitted.

Rising levels of inequality would suggest this is false.

What would rising levels of being lifted out of poverty suggest? The equality gap doesn’t count for much if you starve or freeze to death, and poverty is declining world wide.

> poverty is declining world wide

That depends on how you measure it.

Re: Case study: Algorithmic trading with Go

#216

Post by semilattice » Sat, Jul 01, 2023, 1:05 AM UTC

nvarsj wrote: post #161

skummetmaelk wrote: post #142

Earlier quoted context omitted.

Seems more pointless than crypto to be honest.

It's quite a statement. You're almost saying capitalism and efficient markets are pointless. Maybe they are, but I think it's nothing like crypto. In the old days before HFT, you weren't sure you'd get the best price. You'd have to rely on a broker to make sure that happens, but as a retail trader you generally got a worse price/out of date price. Nowadays with HFT you can get pretty much the best price anywhere. Tho…

>You're almost saying capitalism and efficient markets are pointless. Maybe they are, but I think it's nothing like crypto.

People are saying this because, HFT sounds similar to 'crypto mining'. That's people with best infrastructure, the 'big-guys' -- win. While leaving out the retail investors as broiler chicken, pumped with 'drugs' (by influencers) to spend more on imaginary assets, so that they can be used for 'food' by these 'big-guys'.

There are different influencers for retail investors vs crypto. In retail investing there are promises of 'retirement paradise', actual tax deductions, the Jim Cramer-like people (at least what I heard in US)

For crypto investing the influencer are different, the geography is wider. A promise to participate in markets if you do not live the country that has adopted US/UK-based financial services.

- - - By the way, I think the markets will still have liquidity if there is a rule to wait, for say, 30 min before a stock that was just recently bought -- can be sold (unless by a clear fat finger mistake)

This rule will cause the HFTs to stop existing in the current form.

Re: Case study: Algorithmic trading with Go

#217

Post by TheAlchemist » Sat, Jul 01, 2023, 1:06 AM UTC

jimmysnuka wrote: post #173

I caught the algotrading bug recently and I've been wanting to backtest a couple of strategies involving ETFs. Right now I'm only looking at the open, close, daily high/low. The problem however is that you have to account for dividends and splits, which I'm not sure how to do (Admittedly, I haven't looked too deeply into it). Any advice on how to do that? Let's say I want to backtest some strategy involving UPRO; wou…

Harsh answer - if you have to ask this question here, then stop it now.

Thing is, you will be fooled by bad data - you will find strategies that works in a backtest but not in production. And the reason will be because you missed some important finance concept (like taxes, dividends, stock splits etc).

In this field 99% of success in my opinion is knowing what you don’t know. And only model that small part that you know you know and are fairly sure about it.

Ps. By stop it now - I mean stop algo simulations and learn about those concepts, make sure you understand perfectly what data you are putting into your backtest.

Re: Case study: Algorithmic trading with Go

#218

Post by TheAlchemist » Sat, Jul 01, 2023, 1:06 AM UTC

quickthrower2 wrote: post #177

TheAlchemist wrote: post #135

Earlier quoted context omitted.

What do you mean « moved to another exchange » ? Shares are not held at an exchange.

No idea. I don’t know technically what is meant by this.

Well, technically it’s false, so nothing :)

Re: Case study: Algorithmic trading with Go

#219

Post by _zkyx » Sat, Jul 01, 2023, 1:10 AM UTC

Fripplebubby wrote: post #119

_zkyx wrote: post #42

Earlier quoted context omitted.

Yeah, right now I'm only set up to buy stocks. I haven't tried to short anything yet. I want to get to this eventually since it would be nice to make money when the market goes down too.

I'm sure you know this, but for others reading this who are novices at finance / trading, like I am - the gotcha here is that while the strategy may be symmetric, the risk is not - when buying a stock, there is a floor to how much money you can lose (the price you paid for the stock), while with short-selling, there is no such floor, since the price can rise to any amount and increase your losses to infinite. I belie…

Yeah, I haven't explored this yet. The unlimited risk thing bothers me thought. Which probably sounds funny since this whole things is risky as hell. But, I was more looking into options or something to hedge but honestly everything takes so much time to learn about, test, and then do. So, I just wanted to focus on this and then expand.

Re: Case study: Algorithmic trading with Go

#220

Post by 1ark » Sat, Jul 01, 2023, 1:21 AM UTC

jimmysnuka wrote: post #173

I caught the algotrading bug recently and I've been wanting to backtest a couple of strategies involving ETFs. Right now I'm only looking at the open, close, daily high/low. The problem however is that you have to account for dividends and splits, which I'm not sure how to do (Admittedly, I haven't looked too deeply into it). Any advice on how to do that? Let's say I want to backtest some strategy involving UPRO; wou…

Depending on your provider, but there could be an "Adjusted Close" (Yahoo has it) which includes stock splits and dividends. I tried a few backtesting libraries, thought backtesting.py[1] was easy to work with and straight to the point.

1. https://github.com/kernc/backtesting.py