Being replaced by lab rats

jslade · Post by **jslade** » Fri Sep 26, 2014 5:38 am

http://www.artmarcovici.com/rat-traders

Nonius · Post by **Nonius** » Fri Sep 26, 2014 10:13 pm

dude, when I come down from the V-Rs I'm going to watch this and take notes.

ricko · Post by **ricko** » Sat Sep 27, 2014 11:35 am

Does this guy fit the definition of a crank well?

"when they took the wrong button, they received a minor electric shock."

He is also a cruel

This guy must be told that hit rate is irrelevant in trading unless it gets close to 100%. One simply can't compare trading results based only on hit rates.

"STEP 4 MAKING A BUSINESS

I am now working on a software generating real time ticker tracks in order to put the system to the ultimate test (the real life market conditions). If the performance is good, I will keep the experiment going and either setup an own hedge fund managed by my rats or train rats to work for hedge funds and banks...

(further images and sound files will be uploaded in due time)"

This guy thinks that hedge funds and banks trade the ticker tape. Someone must inform him of what they do...

Smells a rat. If this is not just a joke, then we are at the final stages of entropy increase in the universe...

jslade · Post by **jslade** » Sat Sep 27, 2014 10:59 pm

I didn't read much, as it seemed like fun baloney, but if it were me, I'd have started with fruit flies and, depending on how I feel about traders at the moment, maybe a bug zapper. Sugar futures of course.

FDAXHunter · Post by **FDAXHunter** » Sun Sep 28, 2014 5:12 am

I don't think this research is nearly as far fetched as people make it out to be. Yes, may seem comical, but I don't think the result looks much different if you used humans and more sophisticated forms of "analysis" and/or trading.

goduranus · Post by **goduranus** » Sun Sep 28, 2014 9:17 am

Hell, it may seem funny at first, but the underlying concept, of successively combining successfully trading algos to generate new algos is basically a biological version of the "Genetically evolving technical strategy generator" that people were developing at Wharton.

http://finance.wharton.upenn.edu/~allenf/download/Vita/genetic.pdf

ricko · Post by **ricko** » Sun Sep 28, 2014 12:11 pm

From the Wharton paper:

"Of course, using building blocks similar to trading rules studied in the past poses the danger of data snooping. We try to mitigate this problem by restricting the building blocks to simple rules that have been used in practice for several decades."

Using rules studied in the past poses danger of data snooping so they decided to use simple rules that have been used in practice for several years. Hilarious.

Also, this paper finds no evidence of excess returns from these rules even when results are not properly adjusted for data-mining bias with methods that were developed after the paper was published.

The conclusion is that the artist gives pain to the rats for no reason.

Nonius · Post by **Nonius** » Sun Sep 28, 2014 4:30 pm

finally watched it...tend to agree with Fdax that it's not as wild as it sounds.

nikol · Post by **nikol** » Sat Dec 06, 2014 1:49 am

A year ago over the beer in the discussion about "what can science do" I engineered a joke about frog-traders as a replacement of neural network. I stopped short because I did not know if frogs can be taught at all and how to translate pos/neg PnL into frog's utility. In company, we started to make jokes about opening the company training frogs for hedge funds. Train 1000 frogs in a farm, select performing one (with track record) and sell. Once they stop performing you eat them. ${L_SMILIES_SMILE}$

Also, look at recent article about mouses upgraded with human brain cells. Mouses v2.0 demonstrated 4 times longer (bad) memories.

Polter · Post by **Polter** » Fri Jan 02, 2015 2:46 pm

https://hbr.org/2015/01/rats-can-be-smarter-than-people

...so, they might be onto something ;]