I do not find people I would expect for some topics and the same people come up for several topics. Is this also factoring in karma? If so, results will be skewed.
It does not use karma at all, only comments and the context of said comments. Basically, it ingests comments and from that identifies experts, creates a search graph, etc.
The accuracy would be higher, but there are limited number of comments on HN.
The more comments from a user, the more accurate the system. In the case of HN, many people make less than 10 comments and the majority are shirt responses (i.e. contains minimal data).
If you try "postgresql" or something, there are enough people who have discussed it to provide good results. If you type "hnprofile", likely only these comments will show up.
I like this idea as a startup. It could use multiple data sources, link identities across websites (medium, reddit, hn, quora, irc, github). And be used as a new product for finding potential hires. I really like that product idea.
Which uses this data to find insiders in companies and suggests short term trades on that info (we also have a long term algorithm I use, up 100% yoy).
There's actually a ton of applications, our focus is Enterprise search at the moment.
However, it's setup so what ever you send to the ingestion end point is captured (so you could send all customer engagements, or Reddit, or medium). Basically, using the system how you want.
Does this really work? Do you have a backtests? I trade primarily options and I came across another service that tracks google trends compared to the stock price as well. There was an employee of TSLA who tweeted they hit the numbers goal before, and if you had gotten in on short term optionos, you would have gotten destroyed, even though he was correct. A lot of the times, the news is FUD when it comes to the markets.
I've seen many people post they scrape employee lists from LinkedIn, then follow them on social networks and compile a list of posts based on companies and then bet on that. I haven't seen trade results though.
Edit: Just went through a video and you mentioned prediction of earnings at 80%. Will have to investigate this.
I've personally used it as a tool, and the automated algorithms do very well. With the addition of my insights I've been doing 100%+ yoy. Something like 300%+ 2017 and 250%+ so far 2018. The website itself doesn't publish the algorithm I'm using as testing it requires years.
Yes everything is back tested from 2008 forward, unfortunately although the earnings appear high - it's still not fool proof. If something changed in 2018 that broke my method.. then it doesn't matter how much back testing was done.
Combination, I typically hold 5 assets at once for my long term investments. Rebalancing every 12 months. Last year I was holding Litecoin, Ethereum, Amazon, AMD (been holding that for 3 years now..), and TMobile.
This year it is Amazon, AMD, Discover Financial, PayPal, and Heritage-Crystal Clean (just because it trends oil, but also has assets, algorithm wasn't specific).
I've successful made over 100% yoy, since 2013 though. Again, take that with a grain of salt. Who the knows if it'll work next year.
This topic hasn't been discussed on Hacker news, we missed it, or you found a bug!
So, I'm wondering how you are determining what gets discussed here.
Also, although you say you don't use karma at all, I am guessing you use something else that closely approximates karma, like frequency or number of comments including the term because my old handle, which has 25k karma, comes up in the #2 position for the term "health" though I haven't posted under it in 9 months and, so far, I haven't seen my current handle.
What it does is compare your "expertise" (loosely defined as how often you discuss a topic with technical aptitude), against other users "expertise".
It doesn't matter at all how much or often you discuss a topic only whether or not your aptitude in discussing it is high; when discussing it.
There is a limiting factor, however, if you discuss various topics too often; it'll drop your lower scoring topics you have "expertise" in. This is a storage saving mechanism.
Regarding the handles, I just have the criteria it has to have posted within the last 1 year. Time does not degrade the expertise, only discussing other topics does (I believe there
are 25 top topics the system tracks per user).
16 comments
[ 2.0 ms ] story [ 35.0 ms ] threadWish I could see the which comments caused the mood change(s).
The accuracy would be higher, but there are limited number of comments on HN.
Could you elaborate on that?
If you try "postgresql" or something, there are enough people who have discussed it to provide good results. If you type "hnprofile", likely only these comments will show up.
Although different idea slightly..
Theres also our website: https://projectpiglet.com
Which uses this data to find insiders in companies and suggests short term trades on that info (we also have a long term algorithm I use, up 100% yoy).
There's actually a ton of applications, our focus is Enterprise search at the moment.
However, it's setup so what ever you send to the ingestion end point is captured (so you could send all customer engagements, or Reddit, or medium). Basically, using the system how you want.
I've seen many people post they scrape employee lists from LinkedIn, then follow them on social networks and compile a list of posts based on companies and then bet on that. I haven't seen trade results though.
Edit: Just went through a video and you mentioned prediction of earnings at 80%. Will have to investigate this.
Yes everything is back tested from 2008 forward, unfortunately although the earnings appear high - it's still not fool proof. If something changed in 2018 that broke my method.. then it doesn't matter how much back testing was done.
This year it is Amazon, AMD, Discover Financial, PayPal, and Heritage-Crystal Clean (just because it trends oil, but also has assets, algorithm wasn't specific).
I've successful made over 100% yoy, since 2013 though. Again, take that with a grain of salt. Who the knows if it'll work next year.
So, I'm wondering how you are determining what gets discussed here.
Also, although you say you don't use karma at all, I am guessing you use something else that closely approximates karma, like frequency or number of comments including the term because my old handle, which has 25k karma, comes up in the #2 position for the term "health" though I haven't posted under it in 9 months and, so far, I haven't seen my current handle.
What it does is compare your "expertise" (loosely defined as how often you discuss a topic with technical aptitude), against other users "expertise".
It doesn't matter at all how much or often you discuss a topic only whether or not your aptitude in discussing it is high; when discussing it.
There is a limiting factor, however, if you discuss various topics too often; it'll drop your lower scoring topics you have "expertise" in. This is a storage saving mechanism.
Regarding the handles, I just have the criteria it has to have posted within the last 1 year. Time does not degrade the expertise, only discussing other topics does (I believe there are 25 top topics the system tracks per user).