Fully automated futures trading

[Kernfusion]

This subject made me go to ebay and waste 3 hours browsing for used hardware
And looks like I discovered another option: to buy a whole server i.e. an actual server, something like this one (these people seem to be selling whole crates of them ):

The only problem with these is the horrendous noise they produce (2U\4U units should be better because they have larger, slower-spinning fans..)
But if there's a free space in a dark far-away basement corner, where no one can hear it, I think it's an awesome option! (or convert it to water-cooling..)
Maybe I'll actually try that as a small project in the new year, as I do need a backup for my trading computer..

 

[sef88]

Using an ideas-first approach, I would argue that this is wrong, even though it seems to work OK on crude oil. (I suppose because the term structure is usually monotonic.)

I think it is wrong because you're not realizing the carry from X to Z. You're realizing the carry from 2020 Z to 2021 Z.

Like I said, probably not too big a deal for crude, but it is a big problem for other agricultural contracts where the carry from X to Z (or whatever) may not even have the same sign as the year-over-year carry!

Personally, I do not trade carry on agricultural contracts for this reason.

@globalarbtrader @test_user

I came across Ernest Chan material on carry/roll returns as a predictor. Instead of using adjacent contract, he carried out regression (on daily basis) across term structure to extract the slope.

Would this be a better and more robust approach?

Jr

 

[sef88]

This subject made me go to ebay and waste 3 hours browsing for used hardware
And looks like I discovered another option: to buy a whole server i.e. an actual server, something like this one (these people seem to be selling whole crates of them ):

View attachment 247056

The only problem with these is the horrendous noise they produce (2U\4U units should be better because they have larger, slower-spinning fans..)
But if there's a free space in a dark far-away basement corner, where no one can hear it, I think it's an awesome option! (or convert it to water-cooling..)
Maybe I'll actually try that as a small project in the new year, as I do need a backup for my trading computer..

This backup seems like an overkill. haha!

 

[Kernfusion]

This backup seems like an overkill. haha!

It totally is! but I just want an excuse to play with some cool hardware also the price is hard to beat..

 

[sef88]

It totally is! but I just want an excuse to play with some cool hardware also the price is hard to beat..

You can consider using it to collect tick by tick options chain data. Those cost a bomb when you purchase it from vendors.

 

[sef88]

@globalarbtrader

Could I check with you - when you develop your trend following strategy in futures. Do you consciously compare your results against the Time Series Momentum (TSMOM) 1 year lookback factor developed by AQR (https://www.aqr.com/Insights/Datasets/Time-Series-Momentum-Factors-Monthly)?

I'm still in the midst of developing a trend following strategy using futures (loosely based on your frameworks) and I find the TSMOM benchmark 'hard to beat' even in backtests (with optimized weights from bootsrapping or even equal weights amongst the rules).

Keen to hear your thoughts on this.

Jr

 

[d08]

Regarding the computers: just like you I also prefer to have my own hardware instead of using the cloud. However, different from you I prefer to use a NUC style computer as the automated trading rig because of its low power requirements. I have my NUC powered via a UPS so that in case of a power outage my computer does not crash immediately. With the UPS I have the NUC can stay alive for another couple of hours. A desktop would consume much more power and the UPS would run out in minutes rather than hours.
I don't run a backtest every night, but wouldn't mind having the NUC spend a lot of time on it: it was made to compute, not to sit idle.

For live trading cloud makes much more sense. Because unless you have two or three different connections (Fiber/ADSL, 4G/5G, satellite?), connectivity is still an issue. Location is another benefit, lower latency - less chances of anything going wrong (SE Asia -> NJ vs. NY -> NJ in my case). There have been instances where whole countries/continents have been cut off for hours.
I don't remember having any unplanned downtime with my cloud, it's been great and I'd never go back to using my own hardware for live trading.

 

[HobbyTrading]

For live trading cloud makes much more sense. Because unless you have two or three different connections (Fiber/ADSL, 4G/5G, satellite?), connectivity is still an issue. Location is another benefit, lower latency - less chances of anything going wrong (SE Asia -> NJ vs. NY -> NJ in my case). There have been instances where whole countries/continents have been cut off for hours.
I don't remember having any unplanned downtime with my cloud, it's been great and I'd never go back to using my own hardware for live trading.

I believe that the financial risk associated with a connection outage depends on your trading strategy. I trade with a very low frequency: once per day a trade or rebalancing of investments. If this gets skipped one day due to connectivity issues at that moment (or a power outage) the impact on my financial results are rather limited. The intended trade will get delayed by one day. So it is a risk I am willing to take. Low latency is thus also not of importance in my way of trading.
Paying the extra money to have it running from the cloud does not compensate for this risk. If your trading strategy is different (much faster paced, having a higher financial risk in case of a few hours of outage) then it might indeed be a prudent decision to put it in the cloud instead of hardware at home.

 

[d08]

Paying the extra money to have it running from the cloud does not compensate for this risk.

It's very cheap nowadays though. I pay $10 a month for a single core 2GB instance. This is plenty for IB Gateway and my own Python/IBAPI solution as there's no windowing system, just the terminal.

 

[HobbyTrading]

It's very cheap nowadays though. I pay $10 a month for a single core 2GB instance. This is plenty for IB Gateway and my own Python/IBAPI solution as there's no windowing system, just the terminal.

Thanks. I did not know that the price had come down that much. I was still thinking that it would be a few hundred per month. 2 GB would also be plenty memory for the system I'm running. Anyway, I have my own hardware running and am happy with how it works, so I'm not considering changing it.