|
|
Mon, Apr. 28th, 2008, 12:48 pm
All work and no play makes Homer something something
Over on Bruce Schneier's blog, he links to a study that boring jobs dull the mind (to the task at hand). That particular study was in reference to airport security scanners, and I was under the impression that specifically for that reason, the software would occasionally show false positives on the screen that were not in the bags, and the screener still had to flag them or they were pulled off (although I would imagine the false positive examples were not exhaustive and you could probably grow to recognize them as they came up, so you are less recognizing threats as you are learning to beat the test). It reminds of me some software that my company was trying to sell to law firms to help with the e-discovery process, so that the many people reading over documents wouldn't glaze over and miss a document, and also speed the process up. It was not well received, as law firms make money hourly, so speeding things up is not a terribly large incentive for them, and the trust levels on software are still quite low. At a hedge fund conference, I heard someone from the legal team of the SEC note that they have many documents to review, and not much of a budget nor staff to work on things (compared to funds they are going after who have many more resources - if you are a top notch lawyer, would you rather work for a hedge fund, or against them - if the money to be made were massively larger on the fund side, many will choose that of course). So I figured perhaps we didn't market this properly and should have tried for government contracts - except that I would rather torture than have to go through all of the government hoops for securing contracts - lucrative or not, they are not a sure thing to get and can cost a lot to secure - but once you have them... good stuff. Train of thought posting, it is what I do. Mon, Apr. 28th, 2008 05:42 pm (UTC)
![[info]](http://p-stat.livejournal.com/img/userinfo.gif) freelikebeer: Gov't contracts
The small business administration and your local, friendly congressperson, can usually find grant money for you to jump through the hoops of contractor qualification if you have something that the government wants to buy. It takes a little marketing, but mostly time and effort, not materials. The law firms are your enemy, not your friend. They need their economic [gag] paradigm [hack, plooey, did I just use the 'p' word?] broken. The big ones charge $300/hr for someone who you are probably smarter than to write monotonous motions and other paperwork. They get that rate because the ABA has formed an ad-hoc cartel, and they have the human-resources mentality about the fact that they control confidential information. Despite their professional oaths, it's a fact with consequences in the negotiating ether that decides market prices. [I almost never go see my doctor without already knowing what is wrong; Dr. Google is teh best. The only thing that my GP really does is take on the legal liability for dispensing meds]. You ever thought about writing a lexicon analyzer? It would be interesting to see what you came up with if you mined the set of all legal briefs; something a la lexis-nexis, but with a little more meat [L-N still relies on people at the other end to draw conclusions about what's in the articles]. By the by, how much did you seed your b'roll with for your return to online pkr? If two trains of thought depart separate cities ... Mon, Apr. 28th, 2008 05:51 pm (UTC)
stenz: Re: Gov't contracts
My company is in the middle of something right now where we are essentially being pulled off the market (being absorbed by a client), so all side projects are largely null and void at this point. (and also hence my return to poker, as I no longer spend all free time at home trying to come up with something for the company and instead just do mindless repetitive tasks for hours on end, that for some reason bring me great pleasure) As for the lexicon analyzer, I am fairly certain that there are many working on their PhDs doing that sort of thing, and my access to data like that for free is fairly limited - plus it totally gets in the way of my poker playing :) As for bankroll, in terms of what I used strictly to stick into the site, I didn't use "enough" - but bankroll is really more what you put aside and are willing to lose, not necessarily just shove into the site. Shoved into the site is probably $1k in PS and $500 into FTP. At just those figures, I am under-rolled for the level of PLO that I am playing ($0.25), but the fact that I could put more in were I so inclined is really the main point to consider. I am probably going to have at least one table of limit Omaha Hi/Lo open shortly, just for the joy of the lower variance and the nearly zero effort strategy required to still make money there - use that for (more) quickly clearing the 100% signup bonus at FTP. Part of my poker joy is putting others on tilt, and I cannot do that at FTP unless I single table, as the games are both much faster, and the interface is "Teh Suk" for multi-table play (compared to PS). Mon, Apr. 28th, 2008 06:05 pm (UTC)
freelikebeer: I hope that
being absorbed by a clientmeans that congratulations are in order. Are you going to be able to stay on, and stay in Cambridge? Is this going to be career-ish for you? I am fairly certain that there are many working on their PhDs doing that sort of thingUnless you know those particular Ph.D.s, I wouldn't be so certain. Working around it maybe, but directly working on creating, uhhhh [my heads going to explode from the buzzing of words], value, or really just destroying the value of those hardworking, severely indebted law students, I'm not so sure. Don't make no never mind, though. Ramble brain, that's all. Mon, Apr. 28th, 2008 06:15 pm (UTC)
stenz: Re: I hope that
It is a congrats in that we won't go out of business, but nothing like being bought out by Google or something like that. I will be staying on right here, doing largely all of the same things, and we won't have to fire anyone, so that is good. As for the summary of words things, I read a few blogs (and many books) about NLP and a few of them have talked about similar things, but you have a point, not sure how accurate/successful/targeted they are in that area. Mon, Apr. 28th, 2008 07:15 pm (UTC)
freelikebeer: Blathering blatherskite
I took NLP for my master's stuff [I know OR != NLP, but hey, it's fun]. Professor Hirschberg taught the class. Maybe it was because she has a comprehensive understanding of the field that I got the general-study-vibe from the class. Either way, if people are studying at the edge between NLP systems and particular knowledge areas, they don't talk much about it. Mike Kearns wrote a blurb about using NLP systems to get edges on the market. That's pretty much all that I have read. [The punchline, he doesn't think that the edges that you'll get are very big.] [Train departing the station: it's interesting to me that the Kelly Criterion says bet about the same % of your bankroll as your edge, where other (poker?) conventional wisdom says that you have to bet big to make money on small margins; reconciling those two seems, hard.] Edges in the market are pretty easy to define, I'm curious to hear how people might define edges in areas that are not as readily quantified. Mon, Apr. 28th, 2008 07:29 pm (UTC)
stenz: Re: Blathering blatherskite
The Kelly Criterion is technically if you have inside information so that you know your exact edge, then it is optimal. In reality, you rarely know your exact edge, and therefore many (most?) in the markets will use a partial Kelly style bet (half or even quarter). This of course reduces the gains, but they are still nearly maximized and it greatly reduces your risk of going bust. In fact, this is one of the largest issues in market analysis stuff, using math in one area that works against the markets, assuming that it gives the same information so that it can be used reliably - but it does something much worse - it gives you data that *looks* like it is accurate, but technically can be hugely off in some cases. That is essentially what NNT's Black Swan is about, the inability to properly account for the edges or lack thereof, and therefore improperly taking on more risk than one is aware of and therefore going bust in the unforeseen/unaccounted areas. Technically if poker were to use the Kelly Criterion, you would never bet, as you rarely have an exact known edge (there are cases where you have negative expectation until you take into account implied odds and/or fold equity - so it depends on how you argue that I guess - those figures are largely subjective). As for NLP on the market, the most immediate use for that which I can think of would be to track news data and equity/currency/commodities/etc movement against it in the same way that you would with pairs trading (arguably that is the same as pairs trading, but with additional information), assuming you can determine positive/negative impact of the information and its date. Tue, Apr. 29th, 2008 12:59 am (UTC)
freelikebeer
the inability to properly account for the edges or lack thereofHold that thought. I have some other things to dump from my brain, but I have an idea for you to try on. |
