Trading algorithmically a folio without stops (with IB), real $$$
- Thread starter fullautotrading
- Start date
Well, we are actively watching them. We protected the FAS situation with the CALLs, so they were not let "ride". And these are mostly 3X ETFs.
The "alpha", as I have described in the earlier posts, is produced by the past losses "recovered" during the trading process, which are slowly moved into the gain component (see previous posts).
There is also a need of balance between "load" and hedging. Stopping every second may not be going to produce smaller DDs.
Which one is concerning you most ? If you have suggestions about specific actions to carry out, please feel free to propose them: this is what this thread is for.
Hi fullautotrading,
thank you for your prompt reply and for posting additional information on the current situation of the real portfolio.
It is very informative to see what happen during live trading
and how G-bot can be used to manage many instruments at once.
Reading your latest post and trying to read the picture for current folio situation, I would like to make two suggestions:
a) perhaps displaying the same picture adopting a vertical layout
could make it easier to match the name of the instrument with its open position. I mean just rotate the picture by 90 degrees clockwise, so the instruments are listed vertically on the left and the position bars are horizontal.
b) I have found very informative to peruse both the currently open positions and the overall PNL graphs (so to see how commission are increasing, pnl direction and steepness, positive vs negative positions, etc.
So please in future updates post the two pictures.
I am eager to know how you are going to leverage G-bot to deal with the current non performing positions.
Keep us posted.
Thanks for your time.
thank you for your prompt reply and for posting additional information on the current situation of the real portfolio.
It is very informative to see what happen during live trading
and how G-bot can be used to manage many instruments at once.
Reading your latest post and trying to read the picture for current folio situation, I would like to make two suggestions:
a) perhaps displaying the same picture adopting a vertical layout
could make it easier to match the name of the instrument with its open position. I mean just rotate the picture by 90 degrees clockwise, so the instruments are listed vertically on the left and the position bars are horizontal.
b) I have found very informative to peruse both the currently open positions and the overall PNL graphs (so to see how commission are increasing, pnl direction and steepness, positive vs negative positions, etc.
So please in future updates post the two pictures.
I am eager to know how you are going to leverage G-bot to deal with the current non performing positions.
Keep us posted.
Thanks for your time.
Hi fullautotrading,
so when you are saying:
>As a general consideration, it turns out
> the 3X (like DRN, ZSL, DGLD, ...) are not much comfortable to trade,
do you mean that G-bot strategy is not suitable for these type of instruments?
Could you explain why more in details?
Would you use 3X Etfs again in future portfolios or not?
Again could you please share your reasoning?
Thanks!
so when you are saying:
>As a general consideration, it turns out
> the 3X (like DRN, ZSL, DGLD, ...) are not much comfortable to trade,
do you mean that G-bot strategy is not suitable for these type of instruments?
Could you explain why more in details?
Would you use 3X Etfs again in future portfolios or not?
Again could you please share your reasoning?
Thanks!
> states that short term volatility is greater than long term volatility
Hi jb514. No that is not the point, as that statement is obviously not true. I think I already explained what is the reason of the long term edge of this algorithmic procedure, but you may have missed some of the previous posts. Let me retry with a perhaps less accurate, but hopefully more intuitive explanation.
Assume you had a number of N couples of traders all together using the same account. Each pair does the following. One guys open a position at price P and his idea is to take profit when the prices moves favorably at level P + D. No stops.
The other guy, of the happy couple, watches his reckless partner, playing without stops, and, to protect him, decides to open an opposite position when sees him losing, say at price P - D (in practice his order could be seen as a "stop loss" for the reckless partner).
However this second guy too wants to close <b>only</b> with a profit, just as the first guy. Everybody want to close <b>only</b> if he sees a profit, <b>relative to</b> his own open price.
So if one looked at the account and believed you had only one trader either taking profit or stopping, say, at a distance D, you would have, statistically speaking, say 50% losses and 50% wins, in the long term. However, since here each ("virtual") trader is holding memory of its open price and closes only at a profit, what you get is:
Wins = 50% + R
Losses = 50% - R
where R > 0 is the fraction of trades "looking like" they were stopped, but that could be "recovered" as positive trades because of the price movements within its range (cfr., the first picture on page 4 of this thread). So the whole point of the architecture and the algorithms used on each layer ("scalping/hedging game") is to maximize the percentage R.
The idea that short term volatility is greater than long term vol is the thesis behind market making and it sounds like that is the alpha you are trying to capture in your strategy. If I'm understanding your idea, it sounds like you are basically layering bids down and taking profit on each individual order even though the position is "underwater".
Say a stock is trading at 90. You layer bids down to 80 and put a stop there. So if the price the goes against you down to and fluctuates between 83-86, you are technically down, but as it fluctuates in that range you'll profit from scalping even though you're showing a loss on your position. If over time the stock continues to trade in that range, you'll start to lower your cost basis to the point of showing a profit.
Are we starting to talk about the same thing now? I do like the thread. I actually saw your old thread a while back
That's right. Being "underwater" or above the water
is irrelevant to the purpose of the algorithm.That is an aspect of it, and it's correct.
However, the reason of the edge of the procedure discussed here (which btw is new, even respect to previous threads) is essentially the fact that we maintain past trading information. Our "long term drift" is created by such information (condition <b>necessary</b> but not sufficient), in combination with appropriate scalping/hedging games.
Hi AItrader,
thank you for your suggestions and interest.
I have just sketched a vertical layout: see picture below.
I think that I made it horizontal initially because perhaps it seemed easier to move around (remember you can move and resize it using the mouse). But a "vertical layout" is also quite interesting. Beyond 50 layers (now there are 79) it becomes pretty crowded, but anyway still useful to get the whole situation "at a glance".
<img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3909902" />
As to today the metals do not represent a serious threat yet: so I can leave them alone and see how they scalp.
Today I have loaded a new game that incorporates and automatize a lot of observations and manual interventions I made in the past, and it seems to be a significant improvement.
(If you are in the software distribution list you have them incorporated in the last update. I still need to refine the second example game, originally intended mostly to deal with inverse ETFs).
I googled it, but nothing came up... Can you post the URL of your blog?
Thanks.