stop loss and a take profit algorithm

goldorak · Post by **goldorak** » Thu Apr 30, 2015 6:04 pm

You misunderstood. Let's take a numerical example. Without leverage, if I do -10% and then +10%, I reach -1%. With 2x leverage, -20%, then +20%, and I reach -4%. With 6x leverage, -60% and +60%, and I reach -16%. Feel the lambda squared in the formula below?

svisstack · Post by **svisstack** » Thu Apr 30, 2015 7:33 pm

>> With 6x leverage, -60% and +60%, and I reach -16%.

-36%

goldorak · Post by **goldorak** » Thu Apr 30, 2015 7:36 pm

sorry, typo. Thx svisstack for the correction.

TSWP · Post by **TSWP** » Fri May 01, 2015 9:17 am

>With 6x leverage, -60% and +60%, and I reach -16%.

That's a bold oversimplification that assumes you have first a loss and then a gain and they are equal in size and nothing happens after that.

That's not necessarily what happens in reality.

You may have long sequences, where W is "win" and L is "lose", and % expresses how much you win or lose, so you may have:

W3%, L-1%, W5%, L-6%, W8%, L-3%, etc.

There are infinite numbers of possible combinations and some are producing positive returns in the long term, with or without leverage and returns and losses can be symmetrical, asymmetrical, equal, etc.

The difficulty lies in creating strategies with positive long-term return sequences, this is way too long to be discussed here.

goldorak · Post by **goldorak** » Fri May 01, 2015 12:49 pm

I gave that numerical sample to help you understanding the power of the formula below. Apparently it did not help. I am sorry, but here I cannot help more.

goldorak · Post by **goldorak** » Fri May 01, 2015 12:56 pm

> The difficulty lies in creating strategies with positive long-term return sequences,

I consider that being the easiest part. What is the HARD part is how to get in and out (or how to increase and decrease weighting) of a strategy with time, mainly to separate what is plain volatility around expected returns and actual change in expected returns.

TSWP · Post by **TSWP** » Tue May 05, 2015 10:25 am

> the HARD part is how to get in and out (or how to increase and decrease weighting) of a strategy with time

I couldn't agree more, that is exactly what I do (successfully) with my strategy, without excluding stops, without excluding leverage. That's even HARDER! Cool

svisstack · Post by **svisstack** » Mon May 11, 2015 1:04 am

@goldorak @TSWP:

having underlying valuations in time and my performance in time (no sampling, low latency track),

why this is hardest part? I missing something? I looks like normal research task.

Probably some example will end my wondering about it.

sorry for off-topic ${L_SMILIES_WINK}$

goldorak · Post by **goldorak** » Mon May 11, 2015 8:27 am

Well, take a simple strategy (and let's forget about all details on the strategy itself).

long @Close(t) if Close(t) > Close(t-1) and short @Close(t) if Close(t)

Finding the strategy was very easy. Finding how to time the strategy itself is another story.

TSWP · Post by **TSWP** » Tue May 12, 2015 8:43 am

svisstack,

what I was trying to say is that the timing of your investment decisions cannot be based simply on research on past data, you need "something" that can give you the right timing and usually that something is very complex and hard to build/embed into your strategy.

It's not something you simply "find" in the data through research and slap onto an existing strategy.