Wouldn't it be nice... (the LazyWhale algorithm)

[Wandererfromthenorth]

It's a nice method, it actually menages to keep the huge advantages of trading but without the constant attention at the market required in trading.

There is one thing though that I don't get. You seem to be using a different strategy now in order to decide if it's time to buy or sell.
Look at the picture below, it seems that currently saying that if we go in the $330-$340 the current USD position must be kept.
But how's that, those targets would represent just a natural correction to the bounce after the last $275 flash crash...
The same kinda scenario happened in April-May when we bounced off the $340 flash crash, it seems that the bounce corrented itself and you actually bought in the LazyWhale algo.
It looks to me that you used the opposite strategy that you are using it now...

[oda.krell]

It's a nice method, it actually menages to keep the huge advantages of trading but without the constant attention at the market required in trading.

It's a nice method in backtesting so far

There is one thing though that I don't get. You seem to be using a different strategy now in order to decide if it's time to buy or sell.
Look at the picture below, it seems that currently saying that if we go in the $330-$340 the current USD position must be kept.
But how's that, those targets would represent just a natural correction to the bounce after the last $275 flash crash...
The same kinda scenario happened in April-May when we bounced off the $340 flash crash, it seems that the bounce corrented itself and you actually bought in the LazyWhale algo.
It looks to me that you used the opposite strategy that you are using it now...
http://i60.tinypic.com/2l9jn0w.png

I can see the confusion, but no, I (or rather, the algorithm) doesn't use a different strategy now than for the April/May signal. There's more than one variable at play in the signal, so, while the two situations look similar superficially perhaps, they are not identical for purposes of a signal.

That said, as I said above, the method isn't static. For example, some parameters are updated as new historic data is available. However, it's a relatively minor difference, and not the reason for the difference you noted. Think of it as different context producing different output for the same input (EDIT: or, if you want to think of the context as input as well, simply a different input)

[FUR11]

A very nice avatar from the Perry Bible Fellowship. That's all I need to be convinced this is a great algorithm and the set of rules applied is solid. May you profit big time from this

[oda.krell]

A very nice avatar from the Perry Bible Fellowship. That's all I need to be convinced this is a great algorithm and the set of rules applied is solid. May you profit big time from this :)

Chosen on purpose. I only pretend to know what I'm doing. However, for most practical purposes (fucking hot female unicorns, technical trading), pretending to know and really knowing are virtually indistinguishable, or so I like to tell myself :P

EDIT: http://pbfcomics.com/253/

[TonyT]

Chosen on purpose. I only pretend to know what I'm doing. However, for most practical purposes (fucking hot female unicorns, technical trading), pretending to know and really knowing are virtually indistinguishable, or so I like to tell myself

EDIT: http://pbfcomics.com/253/

Indeed.  Here is a PBF comic more indicative of your ability with mathematics:

http://pbfcomics.com/237/

[windpath]

Saw this and made me think of this thread...

[fewcoins]

400 is a resistance level not a buy signal loll

[oda.krell]

Quick update, more of a  non-update actually...

Nothing changed, recommended position still USD. But, like I said over and over again in here, the method is slow and purely reactive, so I'd argue it's a good thing there's no 'buy' signal yet, considering we're not really going decisively up, for the time being.




Then again, maybe NotLambchop is right when he says:

It's junk because it failed to yield any results.  The utility of your algorithm appears to be exactly zero--neither positive nor negative.  Without knowing the actual ruleset (which you chose not to share), I can only assume that it's junk

Whether you agree or not is for you to decide. I just publish the signal (or lack thereof), like I said I would.

[Tzupy]

I don't think it's junk, but it's probably less useful than some expected. And I doubt it will turn green soon, if it works correctly.

[windpath]

Looking forward to the green line

[TonyT]

Here's my "algo":  take a ruler, print out the chart of BTC price vs time for 2014, line up one end of the ruler at the high peak of the chart and the other end at today's price, and notice how well the intermediate points fall in between (except for the volatile month of October).

Prediction:  Bitcoin will continue to decay.  The 'resistance' seems to be early 2013 or 2012 prices.  That's < 10% of today's price.

Advice: get out. 

You are welcome.

[mmitech]

I don't know why you call it "algorithm", do you really have an actual code written to monitor the change ? because as far as I can tell anyone could see the buy signal only by watching the 1W chart, a reversal would clearly appear there without the need to write any kind of code, just observing would be enough to do the job.


2 weeks ago people where excited when we went from 275 to 417, they called reversal and the moon, while it was only a good but very risky trading opportunity... everything shows that we are still heading down.

[oda.krell]

I don't know why you call it "algorithm", do you really have an actual code written to monitor the change ? because as far as I can tell anyone could see the buy signal only by watching the 1W chart, a reversal would clearly appear there without the need to write any kind of code, just observing would be enough to do the job.


2 weeks ago people where excited when we went from 275 to 417, they called reversal and the moon, while it was only a good but very risky trading opportunity... everything shows that we are still heading down.

https://i.imgur.com/C5x2jBh.png

1) Yes. When I say algorithm I mean it's a piece of code.

2) Glad to hear you can do this by eye. If you would have read the original post, you would have seen that I already said that this method isn't intended to outperform a good discretionary trader. So, if you are one of those, good on you.

How about this: the point of this "public experiment" is to keep track of exact targets (i.e. the price at which buy and sell signal will come in). Let us know the exact targets of your trade pairs and we can compare the results in a year from now.

[sumantso]

So Mr. Krell, your algorithm hasn't signaled a buy yet?

The upward movement seems to be losing steam. I guess if it can survive till next week without falling we might get a signal.

[oda.krell]

Nope. No change in signal yet. I should probably be careful speculating about the likely change regions... last time I did that, my inner bull got the better of me. I'll do it anyway :P

From ~$360 on, a buy becomes quite likely. Price at least staying above $320 doesn't "destroy" the current setup of a possible buy. Falling below $320 and a buy signal moves further away. The above regions are only valid for some time, as they are slightly changing each day.

[Dalmar]

Did the recent mini pump create a buy signal?

[oda.krell]

Not yet, but I'm watching it like a hawk

Next calculation period closes in an hour or two, will update if there's a signal change surviving the close.

[oda.krell]

Here we are. First 'live' signal change surviving a period close.

Important reminder: read the disclaimers and explanations at the end of the post before doing anything stupid.

LW v3.10 signals BUY at 372 USD, from previous SELL at 623 USD on 2014-07-21 (back-test only).





Disclaimer 1: None of the above is financial advice. It's a public "experiment" of a highly speculative algorithmic trading method. I'm posting the results for your entertainment, and my own. Anything else is your own decision and risk.

Disclaimer 2: The signal has been quite profitable in the back-test, but the total low number of signals (by design), and few corresponding trades, means there is very little statistical power behind the results. If you feel there is a substantial risk the results so far are due to chance, you are probably correct.

Disclaimer 3: Even though all BUY signals have been profitable so far in the back-test, in more than one case price dropped briefly after a BUY signal, while the method continued holding BTC, before it went up again.

This is to say, Disclaimer 2 warns you that there is a risk the signal is not profitable at all, while Disclaimer 3 warns you that the signal might be profitable in the longer run (a month or two), but could incur a loss in USD terms at first.

[h3speros]

And one more important notice, even if this would mean that THE bottom is in, it does not mean that it would be good to buy NOW, could be that we are short term overpriced and better time to buy comes soon.

EDIT. didnt read the disclaimer 3 very carefully...

[Tzupy]

In other words, oda.krell may have built a better bull trap that I could.