Fully automated futures trading

[blink18]

I'm on https://www.toptradersunplugged.com/podcasts/systematic-investor/ this Saturday. Any questions?

Rob

Q: If you're from EU, what to do with uninvested cash (EUR) in your trading account if you systematically trade futures in USA? Interest rates are negative in EUR, US&EUR bonds are in freefall, USD is up, portfolio is invested in futures where only margin has FX risk. I was thinking about maintaining neutral cash position - splitting 50% in USD and EUR with regular rebalance. Now I have around 90% in EUR and the rest in USD.
Any better idea? Thanks.

 

[blink18]

Since end of March I was running IB Gateway non-stop with daily restarts and weekly logout/login on Sunday. Today my system sent me an urgent message that portfolio is not visible anymore, and this is the second time that this happened in one month! It happens after IB Gateway's "Restart" feature takes place (my setting is at noon - 12:00 CET) and the error mesage is "No internet connection", "Connecting to server trying for another 30s...".
My system is running on AWS cloud and internet connection is working non-stop, so that is not an issue. If I close IB Gateway and start it again, it connects to IB API server and everything works fine.

Has anyone else had similar problems? Does it matter if IB Gateway's "Restart" happens before or after IB API daily server reset (05:00-06:00 CET)?

 

[Kernfusion]

Since end of March I was running IB Gateway non-stop with daily restarts and weekly logout/login on Sunday. Today my system sent me an urgent message that portfolio is not visible anymore, and this is the second time that this happened in one month! It happens after IB Gateway's "Restart" feature takes place (my setting is at noon - 12:00 CET) and the error mesage is "No internet connection", "Connecting to server trying for another 30s...".
My system is running on AWS cloud and internet connection is working non-stop, so that is not an issue. If I close IB Gateway and start it again, it connects to IB API server and everything works fine.

Has anyone else had similar problems? Does it matter if IB Gateway's "Restart" happens before or after IB API daily server reset (05:00-06:00 CET)?

I haven't experienced this particular problem, but I'm fully shutting down Gateway every day for about an hour while automatic backups and other maintenance tasks are running. Even doing that I had cases when it couldn't properly start and was hanging with some kind of error. Since I upgraded my hard-drive to a fast pci-e SSD a couple of months ago this hasn't happened, so maybe it was related to the general slowness of my system. Still that Gateway is a bloated old java application so..

 

[blink18]

I haven't experienced this particular problem, but I'm fully shutting down Gateway every day for about an hour while automatic backups and other maintenance tasks are running. Even doing that I had cases when it couldn't properly start and was hanging with some kind of error. Since I upgraded my hard-drive to a fast pci-e SSD a couple of months ago this hasn't happened, so maybe it was related to the general slowness of my system. Still that Gateway is a bloated old java application so..

In the past I was running IBC Alpha (daily open/close & login of IB Gateway) on AWS cloud and I had no problems. This has something to do with IB API server daily reset and IB Gateway's "internal" problems...

 

[Kernfusion]

So I finally started implementing trading ICE markets without a data subscription. Sugar #11 is the first one.
I was able to setup EOD price collection pretty easily., IB doesn't give historical data as well w\o subscription, but it does give last-close for the previous day (tick 75) and 15min delayed-last (tick 68), so I wrote a job that runs every evening (Mon-Fr+Sun) and collects\updates previous EOD and inserts last-close as today's EOD, which will be updated to the real one the day after. That works pretty-well.

The other thing is submitting orders without RT-data., I'm trying to use "Snap to Midpoint" IB algo-order type (https://www.interactivebrokers.ca/en/index.php?f=5934) with 15min time-to-live, which, as I understood, submits a limit-order at the current mid-point and that limit never moves with the price if it changes. So I should get a fill at mid or better price or the order will be cancelled after 15min.
But that's not what I'm seeing sometimes. I'm currently trying this order-type in my paper system and products for which I do have RT-subscription, so I should see a pretty-accurate bid-ask at the time of order submission, and I can see that my fills sometimes happen at worse than mid-price, which should not happen with this order type:

Does anyone have an idea why this might be happening?

I see 2 possible reasons:
one is that I misunderstood how "Snap to Midpoint" order behaves, and it can actually move towards a worse price from mid-point under certain conditions (that's my main concern).
And the second is that the price simply moves between the time when I receive the bid or ask and when the order hits the exchange, so the actual mid-point at that time is already different from what I've assumed.

If it's the second I'm fine with it, I'm just worried that I misunderstood how this order-type works..

 

[globalarbtrader]

So I finally started implementing trading ICE markets without a data subscription. Sugar #11 is the first one.
I was able to setup EOD price collection pretty easily., IB doesn't give historical data as well w\o subscription, but it does give last-close for the previous day (tick 75) and 15min delayed-last (tick 68), so I wrote a job that runs every evening (Mon-Fr+Sun) and collects\updates previous EOD and inserts last-close as today's EOD, which will be updated to the real one the day after. That works pretty-well.

The other thing is submitting orders without RT-data., I'm trying to use "Snap to Midpoint" IB algo-order type (https://www.interactivebrokers.ca/en/index.php?f=5934) with 15min time-to-live, which, as I understood, submits a limit-order at the current mid-point and that limit never moves with the price if it changes. So I should get a fill at mid or better price or the order will be cancelled after 15min.
But that's not what I'm seeing sometimes. I'm currently trying this order-type in my paper system and products for which I do have RT-subscription, so I should see a pretty-accurate bid-ask at the time of order submission, and I can see that my fills sometimes happen at worse than mid-price, which should not happen with this order type:
View attachment 284157

Does anyone have an idea why this might be happening?

I see 2 possible reasons:
one is that I misunderstood how "Snap to Midpoint" order behaves, and it can actually move towards a worse price from mid-point under certain conditions (that's my main concern).
And the second is that the price simply moves between the time when I receive the bid or ask and when the order hits the exchange, so the actual mid-point at that time is already different from what I've assumed.

If it's the second I'm fine with it, I'm just worried that I misunderstood how this order-type works..

I guess one way of ruling out (2) is if it's a 'slow' market when the price doesn't change very much.

Have you tried Snap to Market? That's one I had considered myself for this scenario, as it automatically submits an order at the current best bid or offer without needing to know what that is.

Rob

 

[globalarbtrader]

No hedge versus stocks today or yesterday...

I hit a HWM on Thursday night, but then gave back 1% yesterday and 4% so far today. Ouch.

On the upside book is going well. As some of you may have seen on twitter, I found a fast mean reversion straegy (the logic of which I've touched on before eg here) with a SR over 2! So clearly overfitted, but great click bait none the less.

Rob

 

[Kernfusion]

I guess one way of ruling out (2) is if it's a 'slow' market when the price doesn't change very much.

Have you tried Snap to Market? That's one I had considered myself for this scenario, as it automatically submits an order at the current best bid or offer without needing to know what that is.

Rob

I haven't tried Snap to Market, from the description they look very similar, the reason I'm trying snap to mid is that I hope it will give me better executions on average. I.e. most times I'll get the mid price and occasionally will have to re-submit. Or there's really no hope there and I should just use "Snap to Market"?
The one thing is I'm deleting lots of intermediate prices from the db at the end of day, but I can run a report before I do that to see if the price moved between the order submission and execution. I suspect that in reality this behavior is cause by the fast price movement..

My system also lost some money in the last several days..

 

[Kernfusion]

As some of you may have seen on twitter, I found a fast mean reversion straegy

"mean reversion strategy with a momentum overlay and volatility attenuation" - that's a cool name for a strategy, my inner geek is already drooling

 

[Gambit]

Basically mean reversion, initially on individual futures and then at some point on spreads / flys. Holding period will be a few days.

The key point here is the execution, essentially you place limit orders eithier side of the 'fair value' so you're always filled passively (unless you're using stop losses). That requires a different kind of backtest to my current 'assume you always pay the spread and get filled at the close the day after the signal is generated'. Instead you need to work out whether the market moved through your limit during a particular time segment (in which case you assume you get filled at the limit price), and obviously with more frequent intraday data you can avoid missing the situations when the market touched your price and then moved away again.

I will probably use stop losses for which the logic should be something like (conservatively) assume if we traded through the stop that it was executed at the worst level [although it's possible again that the price touched an even worse level in the intra- time period between samples], or (aggresively) exactly at the stop level plus the spread, or somewhere in between.

(Yes I should have OHLC data. But going back in time to get this would be a huge job and not worth the effort to get a slightly better idea of what my slippage is likely to be)

GAT

You’re doing god’s work. Thank you!