Analyzing Mt. Gox

[holy_cow]

Since I like playing with numbers, here is some data I put together using the Mt. Gox API.

I looked at the first trade of every hour and compared it to the previous hour. Then for each hour of the day I added that to the hours total.

So if everyday at 1AM CST you bought coin and sold it back at 2AM CST, over the week you would have made $10.10.

I wouldn't recommend using this to try to make money, but I'm interested in feedback.  I already realized it is a little confusing because at first glance you might think over the 2:00 hour it went up 10, but that actually happened from 1AM-2AM.

Hours in CST.  0 is midnight.  Enjoy.

Hour	Total	Max Gain	Max Loss
0	0.652200000000022	2.63217	-3.38998999999998
1	-9.59388999999996	3.70000999999999	-17.59999
2	10.1076099999999	5.79998000000001	-0.195970000000017
3	-0.420729999999963	6.56999000000002	-4.1131
4	-8.59320000000001	4.19980000000001	-13.7
5	-3.84286999999999	13.87991	-19.2
6	0.65012000000003	7.60003	-9.51000000000001
7	17.84346	13.109	-9.38951999999999
8	11.75949	8.39952	-16.759
9	-49.03894	1.90001000000001	-35.25
10	17.81302	8.68505000000002	-3.51900999999999
11	-10.94309	12.40098	-16.1
12	-12.26199	2.99000000000001	-16
13	-22.03871	10.881	-35.53171
14	-33.15812	3.8	-34.46829
15	2.65399000000001	10.99	-15
16	56.65748	54.89	-1.78644
17	-23.01694	13.4	-38.89
18	0.365310000000008	8.8	-3.41469999999998
19	22.31429	16.48159	-1.95741000000001
20	26.25344	12.41841	0
21	-12.70428	1.90000000000001	-9.83636000000001
22	1.23339999999997	5.2	-9.09711999999999
23	-8.19104000000002	7.73348999999999	-13.00001

[Chainsaw]

If I were analyzing this data, the next step I'd probably make is to perform the same analysis, but after having first separated your input data into two camps, based on the larger, encompassing trend.
For example

Bucket 1 - 24h Avg is Bullish

Hour   Total   Max Gain   Max Loss
0
1
2



Bucket 2 - 24h Avg is Bearish

Hour   Total   Max Gain   Max Loss
0
1
2

Then, if you saw any significant variance for certain hours from one camp to the hour, you may have discovered a signal. (i.e. it is more notable to see an average gain in the short term, amidst a general downtrend, than it is to see a gain within a general uptrend.)

Another way I'd probably want to split the data - the exact same analyses as below, but additionally displaying the averages amongst gains only, and the average amongst losses only. (i.e. downturns in hour 1 tend to have 6x the magnitude of upturns in hour 1)

I'll be interested to see how this turns out!

[gizmoh]

Go analyze their lag time its more complex: CURRENTLY 17 minutes and forever increasing..

[Chainsaw]

Honestly, I had been on the fence about whether the next leg was up or down, as we seemed to be terminating the symmetric wedge.

When I saw this image comparing the last big crash to the chart picture right now...

https://i.imgur.com/fZwy3xd.png

This, paired with the uncertainty, and I moved to a 50/50 position.

Then the lag spike hit.  We'll see in a few minutes...but I think we're far more likely than not to have another downturn, from here ($109).

[creativex]

Honestly, I had been on the fence about whether the next leg was up or down, as we seemed to be terminating the symmetric wedge.

When I saw this image comparing the last big crash to the chart picture right now...

https://i.imgur.com/fZwy3xd.png

This, paired with the uncertainty, and I moved to a 50/50 position.

Then the lag spike hit.  We'll see in a few minutes...but I think we're far more likely than not to have another downturn, from here ($109).

You mean based on Mt. Joke exchange rates and their currently 28m lag and low trading volume?

[Chainsaw]

Creative - yes, that is what I was referring to.

In my view, these are just the natural growing pains of moving out of bleeding edge territory on the adoption curve.  We grew too fast, people got hurt.  But traders will naturally migrate to other exchanges (and more exchanges will spring up) as a result of the pain. So in the long run, the community will be more robust, and we'll further reduce the perceived single-point-of-failure issue.

[holy_cow]

Go analyze their lag time its more complex: CURRENTLY 17 minutes and forever increasing..

AFAIK I would have to gather this data myself.  I don't think they provide historical lag data.

[creativex]

Creative - yes, that is what I was referring to.

In my view, these are just the natural growing pains of moving out of bleeding edge territory on the adoption curve.  We grew too fast, people got hurt.  But traders will naturally migrate to other exchanges (and more exchanges will spring up) as a result of the pain. So in the long run, the community will be more robust, and we'll further reduce the perceived single-point-of-failure issue.

I hope people actually do learn from this. It does seem like those trading on the other exchanges are reacting less and less to Mt. Joke collapses, which is a very healthy sign.

[holy_cow]

This is only for 7 days so it isn't that interesting.  I'll post the one for 30 days later.

	Total	Percent	Max Gain	Max Loss
Sunday	24.54784	15.3948517706949	5.44999000000001	-4.90000000000001
Monday	24.05218	12.8296068751752	7	-10.81
Tuesday	42.04	19.3084547448499	10.881	-3.87100000000001
Wednesday	-67.18999	-48.4400099159623	54.89	-38.89
Thursday	-56.70002	-49.0732739222904	13.109	-35.25
Friday	6.00001	-1.53202344963444	13.4	-17.59999
Saturday	1.24999	0.863068416275368	1.23999000000001	-0.500010000000003