Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Neural Net Side Project Makes $3500/mo Trading Stocks

indiehackers.com

Post reply on HN
Print view View original

101–110 of 122 posts

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#101

Post by ddlutz »

billconan wrote: post #12

What api is available to get realtime stock data? And what api can be used for trading? Are they expensive?

I'm fairly certain this is in the article: For trading I recommend Kite, mainly for their stable Connect APIs and the low bandwidth. Their limitation is 3 requests per second, and this was more than enough for my new strategy. Getting solid historical financial data isn't cheap, and with so many people hitting the providers to scrape and download data, I don't blame them for limiting the offered information. Intrinio is a good provider for real-time stock quotes at very inexpensive prices. However, getting access to more in-depth data would always yield better results.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#102

Post by alva »

csallen wrote: post #99
alva wrote: post #97

Earlier quoted context omitted.

You are strongly defending him. What proof do you have of his claims? If this turns out to be a dodgy submission for self promotion it is mildly embarrassing for IndieHackers if they admit error in this instance. If it turns out you have seen no solid proof for these claims and amidst the heavy skepticism express by multiple people in this thread you do not investigate and demand proof, than that calls into question…

You are consistently ignoring the logic and reasoning I provide in my responses, and I'm not sure why.

This is simple.

What proof do you have of his claims?

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#103

Post by csallen »

alva wrote: post #102
csallen wrote: post #99

Earlier quoted context omitted.

You are consistently ignoring the logic and reasoning I provide in my responses, and I'm not sure why.

This is simple. What proof do you have of his claims?

I sincerely don't understand what you want. You claimed it was impossible for him to find an exchange, and then he revealed his exchange. You claimed it was impossible for him to find data sources, but then he revealed his data sources. You claim that his system can't possibly be as good as it appears, but even Sebastian admits that he had to manually intervene and has also gotten lucky.

What extraordinary claim is being made that requires proof?

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#104

Post by alva »

csallen wrote: post #103
alva wrote: post #102

Earlier quoted context omitted.

This is simple. What proof do you have of his claims?

I sincerely don't understand what you want. You claimed it was impossible for him to find an exchange, and then he revealed his exchange. You claimed it was impossible for him to find data sources, but then he revealed his data sources. You claim that his system can't possibly be as good as it appears, but even Sebastian admits that he had to manually intervene and has also gotten lucky. What extraordinary claim is b…

> You claimed it was impossible for him to find an exchange, and then he revealed his exchange

Wrong.

"1. How did someone outside the financial world get access to an exchange that would allow this" "First bit of proof I think you should get. 1. Agreement with exchange" "1. I was not even referring to "top" exchanges. even just a single exchange. ask him which one he uses"

>You claimed it was impossible for him to find data sources,

Wrong.

"2. He states that he trained the model on financial data. This data is Extremely expensive. If he is using public data like Yahoo Finance, the level of detail that is provided is so low that it is incredibly unlikely any automated strategy could work off it" "[provide] 3. Data sources for AI training" "2. It is very unlikely a succesfull strategy could be based on this [public data]

> What extraordinary claim is being made that requires proof?

The claim that someone with 5 months of financial self learning has created an AI strategy that makes a profit 95% of the time. That is an extraordinary claim.

I have asked him elsewhere in the thread and still have not had a response.

Post historical and current positions, open and closed.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#105

Post by csallen »

morgante wrote: post #100
csallen wrote: post #95

Earlier quoted context omitted.

I find these criticisms quite unfair, especially given that he admits in the article that (a) he is new to trading, (b) his results cover a short period of time in a bull market and could easily be 100% luck at this point, and (c) he's had to manually step in to avoid ruin. You're excoriating Sebastian for not hedging claims that he has repeatedly hedged. I also disagree that there aren't any valuable or inspiring le…

What are your standards for including interviews on Indie Hackers though? Honestly, my criticisms are less of Sebastian than of you. He admits that his strategy is unlikely to be profitable in the long run, but what exactly is the merit of including stories about an amateur trader on a site about technical side projects? I, and likely many others in this thread, have tried our hand at trading. I've had months where I…

I'm also trying not to be too defensive, but I strongly disagree with your interpretation of this interview. Most of the IH audience will read this, care very little about the "secrets" behind the trading strategy, and instead focus on what you call "trite lessons".

In this particular case, the amount of time and effort an 18 year old put into researching and educating himself about a difficult field is quite inspiring. As is his dedication to doing some non-trivial programming work and actually launching. That might not be inspiring to you or to professional traders on HN, but I can assure you that most IH readers will not care if/when his trading bot ultimately fails.

> What, exactly is the point of the interview then? You could just as well cover random sites that have launched with 0 profit or revenue. Heck, if you're in that business, I've got a ton of side projects I'd love to share.

I've featured lots of very small projects. My minimum cutoff nowadays is $100/mo, but in the past I've done interviews where the revenue was $0 or not shared. Although it's fun celebrating big revenue wins, that's not the point of the site. (If you don't believe me, I can point to numerous places where people have actually asked me to feature $0 failure stories on the site.) The point is motivation, inspiration, and education for getting around the most common obstacles that stand in the way of creating an online business.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#106

Post by csallen »

alva wrote: post #104
csallen wrote: post #103

Earlier quoted context omitted.

I sincerely don't understand what you want. You claimed it was impossible for him to find an exchange, and then he revealed his exchange. You claimed it was impossible for him to find data sources, but then he revealed his data sources. You claim that his system can't possibly be as good as it appears, but even Sebastian admits that he had to manually intervene and has also gotten lucky. What extraordinary claim is b…

> You claimed it was impossible for him to find an exchange, and then he revealed his exchange Wrong. "1. How did someone outside the financial world get access to an exchange that would allow this" "First bit of proof I think you should get. 1. Agreement with exchange" "1. I was not even referring to "top" exchanges. even just a single exchange. ask him which one he uses" >You claimed it was impossible for him to fi…

> It is very unlikely a successful strategy could be based on this [public data]... The claim that someone with 5 months of financial self learning has created an AI strategy that makes a profit 95% of the time.

I've responded to this many times now, and you've ignored me, which is why I don't believe you actually want an answer, so this will be the last time I respond:

Sebastian has not claimed that his strategy will work 95% of the time. That would be crazy. He hasn't even claimed that it has worked 95% of the time –- he specifically stated that if he hadn't manually intervened, it would have lost all his money.

He merely reported that over a limited window of time (5 months) with a limited number of trades (non-HFT) in a bull market, 95% of those trades were profitable. If you break down the math correctly (which you did incorrectly in our Twitter convo), winning 19 out of 20 coin tosses = 1 in 52k. That has surely been done many millions of times, assuming hundreds of millions of trades are made daily.

I'm sorry, but the extreme claim that you're looking for proof of was never made. You are battling a strawman.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#107

Post by kasey_junk »

hmate9 wrote: post #47
brianwawok wrote: post #27

Earlier quoted context omitted.

Backtesting does not mean much. It's easy to make an algo do well yesterday. Very hard to make it do well tomorrow.

I completely disagree. Backtesting matters, especially in HFT. What would you rather have: an algorithm that has performed well on 5 years worth of historical data or one that has performed well in the past 2 weeks.

In my experience? The later. Regime change is a killer in trading algos.

Further, accurate back testing is very difficult for lots of reasons, the most obvious being your own actions aren't reflected.

Most algo firms use historical data to develop a hypothesis & then many different ways to validate that hypothesis. Test trading being the one with the most weight.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#108

Post by brianwawok »

samfisher83 wrote: post #69
brianwawok wrote: post #26

Earlier quoted context omitted.

You can make more than 100% going to the slot machine at the casino. The difference between gambling and trading is not that big. Trading has more ways where you can get an edge over the house, but it is not that different from poker or slot machines.

If you make a million spins of $1 each you will end with $800000 or whatever the state law is regarding payout. In general an index fund will go up since most economic policies are geared toward inflation.

HFT is not an index funds. No relation.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#109

Post by brianwawok »

tunesmith wrote: post #58
brianwawok wrote: post #27

Earlier quoted context omitted.

Backtesting does not mean much. It's easy to make an algo do well yesterday. Very hard to make it do well tomorrow.

Does backtesting work better if you, say, start with seven years of data, use the first five to judge the strategy, and then test it against the last two? Is that a good way to avoid fitting the data?

No.

When you trade in a live market, your actions cause responses. When backtesting your actions get no responses.

Re: Neural Net Side Project Makes $3500/mo Trading Stocks

#110

Post by morgante »

csallen wrote: post #105
morgante wrote: post #100

Earlier quoted context omitted.

What are your standards for including interviews on Indie Hackers though? Honestly, my criticisms are less of Sebastian than of you. He admits that his strategy is unlikely to be profitable in the long run, but what exactly is the merit of including stories about an amateur trader on a site about technical side projects? I, and likely many others in this thread, have tried our hand at trading. I've had months where I…

I'm also trying not to be too defensive, but I strongly disagree with your interpretation of this interview. Most of the IH audience will read this, care very little about the "secrets" behind the trading strategy, and instead focus on what you call "trite lessons". In this particular case, the amount of time and effort an 18 year old put into researching and educating himself about a difficult field is quite inspiri…

> Most of the IH audience will read this, care very little about the "secrets" behind the trading strategy, and instead focus on what you call "trite lessons".

I'd like to think I'm part of the core IH audience. I come for interesting stories of people building software from scratch and generating revenue from it.

Someone who got lucky gambling their money for a few months is just not an interesting story. It's way outside the mold of what you usually publish.

> My minimum cutoff nowadays is $100/mo, but in the past I've done interviews where the revenue was $0 or not shared. Although it's fun celebrating big revenue wins, that's not the point of the site.

It's good that you have a cutoff, but I think the issue is that this project doesn't necessarily meet that. With your typical interviews, you know they're making revenue from actual customers and that revenue is a meaningful metric. He happens to have generated profit so far, but I also happen to have generated profit on my gambling trip: it's not a meaningful metric, and in fact, might as well be 0.

He could easily have all his revenue wiped out next month (and, in fact, almost did). I'm really not sure what we have to learn from someone who has successfully gambled for a few months.

Post reply on HN