Topic: (Closed) Butter Bot!: Premier Bitstamp, and BTC-E EMA Trading Platform (Closed) - page 107. (Read 274784 times)

Hi PuertoLibre ,
   I have added the ADX strategy you suggest to our internal discussion board. When we advance development on our new strategy modules we will be back testing this and other suggested strategies against the Bitcoin market to find the most effective ones. Please be sure to post any other suggestions you may have!



Best,
  Pablo.

ADX?

http://www.youtube.com/watch?v=Csq7gDnbapI#at=46

During a Gold rush, the people that end up making the most are the ones that sell the shovels..     

That is great news Pablo that you are looking into implementing other methods besides EMA..    I am pretty new to Technical Analysis and EMA is all I know but I know there are a lot of other methods..
 

Hi Jubalix ,
   It's not really "if" the bot works, the math behind the bot is solid and EMA is a well known algorithm. We have based the bots logic behind "Gomboo's Thread" which is one of the most respected trading discussions of EMA as applied to Bitcoin anywhere by one of the most respected members of the Bitcoin community. The thread has a complete explanation of the math behind the bot and why it is profitable over time.

Please take a look here:  https://bitcointalksearch.org/topic/goomboos-journal-60501

As to why we sell it: We are a service provider. Why does ASICMINER sell shares when they are clearly profitable on their own? To raise capital for continued development. We use our own product, one which we continually upgrade and work on. Like ASICMINER, we would make money by trading on our own but providing the B-Bot service allows as to develop the bot beyond what we otherwise could for our benefit and the benefit of all the members of our community.

Please let me know if I may be of service in the meantime .

Pablo.

Hi Dresden ,
  As Kuroth points out, this is all well and good in theory, but Back Testing continually shows that trading at crossovers is way less profitable than trading at threshold 0.25, 1 hour interval settings. I recall someone posting back testing data of this last week (again, I think it was Kuroth). Gomboo's thread goes into great detail as to why this is is and as to why EMA trading requires longer intervals of time to measure success than say, day trading (where P&L is immediate).

We are working very hard to push out the Back Testing feature for the bot and we are well advanced on it. This feature should take care of this entire debate as you will be able to back test your strategy yourself against all historical data. For now, please believe that the settings we recommend, which are Gomboo's settings, are the best mathematically profitable and proven EMA thresholds out there.

Regarding your point about flash crashes and the need for earlier signal: EMA is designed to be profitable over a longer period of tiem accounting for a few larger wins to profit over a couple of smallish losses hence taking a small loss on a trade is part of the cycle and current settings should make these losses comparatively small to your wins. Again, it's part of the cycle. Altering trigger mechanisms would anull the mathematics behind the bots logic.

Just as a general FYI, we will be looking into incorporating alternative strategy modules (Besides EMA) in the V2-V3 life cycle so that everyone can use a strategy suitable to their risk profile (as EMA is mathematically provable but takes some patience). If anyone has any ideas as to what strategies they would like to see, please let us know .

Pablo.

Also call me crazy but I dont plan to measure the success of my Bots trading(settings) until about 1 year of running it 24/7 365 days a year..


I think most want to measure their success in 3 or 4 weeks...       EMA dont work that way and if that is what your after, you need to find some other method.


For example..  In the last 2+ weeks of using this method and Bot, I have had 4 losses(small) in a row and just now early this morning my first win(small).

I am not sure most people have the patiance for this Long term method(as stated over and over and over in GoomBoss thread)   

Guys..  All good points and theories BUT having read GoomBoo's thread and back tested myself, it dont prove out...  I was amazed in my back testing how time and time again the 10, 21, 0.025, ,0.025 1H ALWAYS had the best results..   Now I know back testing results might not always have the same results in the future but its still pretty accurate..

GoomBoo did at one point go to 1D instead of 1H...  But he was trading manually and not using any Bot(That we know of)


HIGHLY recommend you read all of GoomBoo's Thread

I've been thinking along the same lines.

The problem with going with "trends" in one hour increments owes to how volatile the BTC market is. Unlike the stock market, where crashes are protracted over days and weeks, crashes on MtGox and BTCE are protracted over minutes and hours.

Those 1 to 2 hours could make all the difference in a crash situation, and the way the bot currently works with thresholds could lead to you losing a lot of money if say the first threshold for a sell is reached when BTC drops from say 90 to 80, its entirely possible that price to crash all the way down to 60 in that next hour or two. So you're looking at a situation where if you had say 10 BTC and a crossover occurred, you could have sold at say 85 when the crash was happening, instead of the 65 that will happen because the bot hasn't hit its thresholds yet.

That's why I proposed the earlier feature that would allow a user to only buy and sell AT crossovers, or at least allow the mix of the two. For example I could see myself wanting to use thresholds in all my buy situations but only wanting to use strictly crossovers in all of my sell situations.

Trailing Stop Loss = A percentage value. (Got it)


TSL is 1.0 %
If Max_Price_of_BTC < TSL
Then
Sell_All_BTC_For_USD


Huh....?

< I am confused.

=================

I understand that TSL is made to tackle a very critical problem of the average price going up rather than down. In day trading you cannot honestly lose if you turned all BTC to USD and the price plummets. (except in cases where it reaches zero)

You'd always want to increase your BitCoins by following a basic principle.

A) Sell your BTC for USD At the highest peaks, B) then when the price plummets buy the coins you sold plus whatever else extra you can afford at the lower price point. C) Then, when the price goes up again, you resell the BTC for USD again and you end up with more USD than the first round.

The only way to lose in this strategy is if (Part C) never happens again OR (In part A) you sell too low and misjudge where the peak is.

So Trailing Stop Loss is there to prevent (Part A) from happening or limiting the loss (Stop Loss percentage?). Do I understand this correctly? Or is a Trailing Stop Loss different than a regular Stop Loss?

----------------------

What is catching me off guard is that you are supposed to be using TSL as a function to determine where there is a loss by the average price in a period going up (after a Sell_All_BTC gets triggered). Or did I just read that wrong?

Usually a "crash" is defined as a downward market. If this is the case wouldn't it be extremely profitable to ride the market downhill and at the lowest crossover signal to trigger a Buy_All_BTC?

Or is B-Bot's strategy completely different from what I have described? (I know, I know, "Trends" not "Day Trading")

can some on explain, if this bot worked to acquire more btc, why would you need to even bother selling it...?

Hi,
   Yes, I'm happy to explain .

Trailing stop loss allows you to set a value which is a percentage. Then the bot will sell all your BTC for USD if the max price of BTC since you turned on the bot falls below your Trailing Stop Loss setting by the percentage you have specified. It is a safeguard feature in case of a crash.

In the past, if the TSL was triggered but the market went up again, the bot could in theory rebuy incurring possible losses through fees. We have recoded it so that a TSL trigger will pause the bot until an operator has had a chance to review everything and reset it.

We believe this mode will be much more efficient.

Please let me know if you have any other questions ,

Pablo.

Perfect, no problem .

Please let me know if you need anything.

Pablo.

Nevermind, I'm an idiot, 0 is technically greater than any negative number.

I just changed the sell thresholds to -0.01 and -0.02 and it triggered a sell almost immediately.

Could you explain why this hasn't sold anything yet?




This seems pretty straightforward to me.

Before the 20:00 mark the first threshold, 0, would have been met (That's when the crossover occurred).
Exactly at the 20:00 mark, the second threshold would trigger because -0.01 is < -0.029

I can maybe imagine some lag and the sell not going off until 21:00, but it still hasn't, what the heck?

-Recoding of "Trailing Stop Loss," bot is disabled when TSL is triggered to avoid rebuy fess.


Could you explain more about this particular feature?

Looking forward to the release..   

I just want to let everyone know that we are on track for V2 release next week.

The code has been entirely rewritten from scratch and several important features added such as:

-Support for BTCE
-Server Back End to improve stability and support
-Fiat Reserve
-BTC Reserve
-Increased speed, and versatility.
-No need to reset, just set it and go.
-Recoding of "Trailing Stop Loss," bot is disabled when TSL is triggered to avoid rebuy fees.

The website and the bot interface are also going to be relaunched, we hope everyone loves the new product!

BUY NOW: There is still time to get in on the "Early Adopter" deal before we launch V2. Anyone who purchases before the V2 launch will receive a FREE 1 year subscription to V2, an HUGE value.

If anyone has any questions, please drop me a line.

Pablo.
www.btcbots.com

Definitely. I'm still in the experimentation stage. I'm not a big enough player to where I care that much if lose some BTC or USD in the process of trying to figure out what works the best.