Fully automated futures trading

In case you guys didn't see it on twitter
In case you didn't know this already: Twitter has recently changed its settings. Nowadays only people who have a Twitter account can see tweets by others. People who don't use Twitter can't see anything at all. It used to be such that people who don't have a Twitter account can read tweets.
 
In case you didn't know this already: Twitter has recently changed its settings. Nowadays only people who have a Twitter account can see tweets by others. People who don't use Twitter can't see anything at all. It used to be such that people who don't have a Twitter account can read tweets.
Interesting so forwarding a link may no longer work
 
Nearly a year after @Kernfusion first posted about it, I've finally got round to implementing instruments with no market data. Am happily collecting EOD data, and I've done a test trade with the SNAP order type (I won't do any real trades until I have been sampling them for a while and get round to putting them in my strategy, probably not for a few months). Just need to check I can collect delayed bid/ask samples which I use for slippage data (if not will have to enter these manually). As a result, I'm pushing another batch of instruments into my dataset which will be at 250+ when I have finished.

Pleasingly, the very first of these was the FTSE100 which I used to hold as a hedge against my long only positions back when I had a 'dirty' portfolio rather than the clean futures + cash I have now. Since I couldn't auto-trade it, I had to do a manual trade via the web interface every roll. You can see I was short two contracts of H15 way back on the very first post on this thread...

Rob

Just discovered Snap to Primary orders because of this post. They are a godsend!

Goodbye market data fees!
 
Good morning. At this time following the system in the book leveraged trading. Continuos Trading, the performance is:

Profit&Loss posiciones
P&L in SPY is 173.91
P&L in GLD is -3.42
P&L in TLT is -2.05

El capital allocated para cada instrumento es de: 3033.00

Capital broker in SPY is 3221.68
Capital broker in GLD is 3008.34
Capital broker in TLT is 3017.49

Rentabilidad in SPY is 6.22 %
Rentabilidad in GLD is -0.81 %
Rentabilidad in TLT is -0.51 %
Rentabilidad total del sistema 4.90%

This is for the firts system. The other one is:

Profit&Loss posiciones

P&L in AAL is -15.92
P&L in GOLD is -1.48
P&L in MRO is 8.54
P&L in BAC is 9.45
P&L in NIO is 46.06

Capital broker in AAL is 2500.69
Capital broker in GOLD is 2542.08
Capital broker in MRO is 2576.70
Capital broker in BAC is 2569.50
Capital broker in NIO is 2621.06

Rentabilidad in AAL is -2.89 %
Rentabilidad in GOLD is -1.28 %
Rentabilidad in MRO is 0.07 %
Rentabilidad in BAC is -0.21 %
Rentabilidad in NIO is 1.79 %
Rentabilidad total del sistema -2.52 %

I think if you are lucky to get a trend in one of this stocks (NVDA) you can make money, but at this time the second system is loosing money.

Any suggestion?

Thank to everyone.
 
Hi all, a question about data service provider regarding strategies in ATFS. I'm new to pysystemtrade and setting up the system for relatively small account size (not in millions..). My idea was of course to use dynamic optimisation and trade as many cheap enough futures as possible. I've set up IB and Barchart, and noticed that Barchart gives me around 80 futures, while instrumentconfig has a lot more of them. Given the constraints of limited capital and experience so far, what would be the best way to go with data to get a good enough system up and running? Should I go with Barchart for a start and use what they have, or get more data elsewhere? I guess I would prefer to get the system running and gain experience, and add more data later.
 
Would anyone here include a MR type rule (which I have constructed) in their system which has a SR of .3? This is based on a backtest using pooled returns of 40 instruments. I am tempted to throw it out but....
 
Last edited:
Hi all, a question about data service provider regarding strategies in ATFS. I'm new to pysystemtrade and setting up the system for relatively small account size (not in millions..). My idea was of course to use dynamic optimisation and trade as many cheap enough futures as possible. I've set up IB and Barchart, and noticed that Barchart gives me around 80 futures, while instrumentconfig has a lot more of them. Given the constraints of limited capital and experience so far, what would be the best way to go with data to get a good enough system up and running? Should I go with Barchart for a start and use what they have, or get more data elsewhere? I guess I would prefer to get the system running and gain experience, and add more data later.
If you have "IB set up", why would you not use it for getting data on all instruments?
 
So a different kind of question, more on the investment side of things (perhaps should've asked it before the last Rob's TTU show :) ):
Now with the rates up quite a bit, it's possible to get a risk-free return on cash of 5.5% or so., and taking into account the general opinion that stocks will not be as profitable in the next 10-20? years because of how overvalued they are after the zero-rate bonanza, does cash have a place in a long-term portfolio now?
 
Why is the raw futures-derived carry signal so wild for many gov bonds?

When compared to the YTM-derived carry (using data from Treasury.gov, Bank of Italy etc.), the futures-derived carry looks wildly off for many instruments I investigated (except ZN, ZT, & the JGB).

I'm not referring to the predictable seasonality of German bonds, I'm talking about unpredictable swings quarter to quarter.

Here are some examples:

upload_2023-8-7_14-33-54.png


upload_2023-8-7_14-34-0.png


As you can see, futures carry can give some pretty extreme readings whilst the YTM-derived carry cruises around slowly.

Ideally, the futures carry and the YTM-derived carry would match fairly closely, like they do for ZN:

upload_2023-8-7_14-34-42.png


Anyone got an idea what's going on?
 
Back
Top