Pages:
Author

Topic: Large walls moving on Gox orderbook aka Manipulator (Diagram) (Read 6484 times)

legendary
Activity: 1708
Merit: 1020
there was a small bug in the automatic updating process that prevented the image on bitcoinX to update - should be fixed now.

@mushoz: the file is a couple hundred of megz compressed.  where would you like me to upload it?  no experience with all these shady upload services..
hero member
Activity: 686
Merit: 500
Bitbuy
Quick question: Do you happen to have your data in raw format as a download as well? Would love to play around with the data and create some of my own graphs. Thanks! Smiley
hero member
Activity: 686
Merit: 500
Bitbuy
Now live updates every 17 minutes on http://bitcoinX.com/charts (at the bottom)

Thanks! Amazing new feature! Much appreciated Smiley
legendary
Activity: 1708
Merit: 1020
Now live updates every 17 minutes on http://bitcoinX.com/charts (at the bottom)
legendary
Activity: 1708
Merit: 1020
I never said I could read much from the graphs. But the huge rapid swings are very interesting. also the amount of money in the books is a signal of speculation interest.

also there is another point: whenever you put an order in the book somebody else can take it. that means if someone has five hundred thousand coins (and may be a handful of people who do) they can sell them for a million dollars in the blink of an eye.

whatever orders you have in the books is at risk of being filled.
legendary
Activity: 1264
Merit: 1008
I've said it before and I'll say it again. As long as anyone can have programs that put in bids or asks they never intend to fill, and instantly pull them out before they would get filled, no one should take the Mt Gox market depth data seriously.

+1 

There's no guarantee that the "depth data" includes all asks/bids or that the asks/bids it does include will be acted upon. 
Seems to me an intelligent bot would ignore that stuff completely and work with real trade data.  Hopefully some of that can be trusted but again you never know what people are doing off the books. 
hero member
Activity: 675
Merit: 502
I've said it before and I'll say it again. As long as anyone can have programs that put in bids or asks they never intend to fill, and instantly pull them out before they would get filled, no one should take the Mt Gox market depth data seriously.
legendary
Activity: 1708
Merit: 1020
updated... hmm...
legendary
Activity: 1708
Merit: 1020
Quote
the manipulator

Nope, it's just PayPal or MasterCard with another wallet stealer trojan  Roll Eyes

Edit: Link: http://www.eweek.com/c/a/Security/Waledac-Botnet-Reappears-as-New-Password-Stealing-Variant-882729/

only steals conf --> client with password is protected
legendary
Activity: 1554
Merit: 1021
Quote
the manipulator

Nope, it's just PayPal or MasterCard with another wallet stealer trojan  Roll Eyes

Edit: Link: http://www.eweek.com/c/a/Security/Waledac-Botnet-Reappears-as-New-Password-Stealing-Variant-882729/
hero member
Activity: 756
Merit: 501
There is more to Bitcoin than bitcoins.
updated once more....    on the weekend I lost some data because my dns stopped resolving mtgox  Shocked

This is excellent.
legendary
Activity: 1708
Merit: 1020
updated once more....    on the weekend I lost some data because my dns stopped resolving mtgox  Shocked
vip
Activity: 448
Merit: 252
Yeah thanks,

it was an amazing race... fucking manipulators .... since few weeks
legendary
Activity: 1708
Merit: 1020
updated.... another nice move
hero member
Activity: 686
Merit: 500
Bitbuy
so what do you say, is it useful / interesting, should this go on bitcoinX ?

Definitely! Thanks for posting these!
legendary
Activity: 1708
Merit: 1020
Posting mostly to watch this thread, but I was curious as to the big drop last week... my thoughts were, after looking at it for awhile, was a glitch in a bot and not an intentional dump.  Does your data support that potential theory?  

I have data points only every 17 minutes or so but it looks like an intentional dump to me. Like the manipulator let a bulging bitcoinica burst: https://bitcointalksearch.org/topic/comic-the-sharks-and-the-trampoline-52336  Wink


The asks were reduced by a crazy amount right before the crash --> single entity.
edit: actually that statement was wrong for the last crash but seems to be true for today's


Maybe somebody has more detailed order book data for the time of the crash - would be interesting.
legendary
Activity: 1260
Merit: 1000
Posting mostly to watch this thread, but I was curious as to the big drop last week... my thoughts were, after looking at it for awhile, was a glitch in a bot and not an intentional dump.  Does your data support that potential theory? 
legendary
Activity: 1708
Merit: 1020


Pseudocode:

exp = 10

loop over time:
  loop over bids:
    weightedBid += bAmount * ((bPrice / price) ** exp) * bPrice

  loop over asks:
    weightedAsk += aAmount * ((price / aPrice) ** exp) * price * 1.0

second_subplot = np.array(lWeightedBid) - np.array(lWeightedAsk)


I have played a bit with these formulas last weekend. Checked the correlation to the price for a bunch of strategies but did not get a good result. The closer you go to the middle of the melting pot (higher exp) the more noise you get.


note how there is much more noise on the price chart starting with the new year (jump). Traders suddenly more active? New bot?



hero member
Activity: 756
Merit: 501
There is more to Bitcoin than bitcoins.
Just thinking... Asks and bids which are far from the current price are "less serious" and should be weighted down in this kind of analysis. What matters most is what happens close to the current price. Can you consider adding weighing coefficients which decay as you go further from the price at any given moment?

+1, that would be very useful

that was among the first things I tried. I will post how it looks tomorrow. The problem is that it is hard to grasp / describe what is going on behind the scenes.

Great! I already find your diagrams quite interesting.
legendary
Activity: 1708
Merit: 1020
Just thinking... Asks and bids which are far from the current price are "less serious" and should be weighted down in this kind of analysis. What matters most is what happens close to the current price. Can you consider adding weighing coefficients which decay as you go further from the price at any given moment?

+1, that would be very useful

that was among the first things I tried. I will post how it looks tomorrow. The problem is that it is hard to grasp / describe what is going on behind the scenes.
Pages:
Jump to: