Fully automated futures trading

If all instruments are equal weighted, the following Sharpe ratios are achieved in the backtest:

automated clustering 0.711
manual clustering 0.683
equal weights 0.667
 
globalarbtrader said:
Yes, that's definitely not a significant difference.

GAT
More...

I've done the same with automatic clustering + improved interpolation technique:
clustering_weights_comparison_2.png

The outcome was Sharpe ratio of 0.703. Basically the same as 0.711 that I got from the new recently published methodology using the SLSQP optimizer and estimate uncertainty + hacks.

I guess the conclusion is: it doesn't matter. Just pick a "good enough" option and forget perfection.
 
Hi, not sure how systematic (or dumb :) ) this question is, but at least maybe it'll humor someone.
So in March-April after everything went crashing down, many people went bargain-hunting but it wasn't very clear what was "a good deal to buy"..
So I thought to myself:
1. for my long-term investment portfolio I have a very long investment horizon, so I can easily wait years..
2. the world basically has 2 choices now: either we recover (maybe slowly, but see #1) or we just unwind this whole "society thing" with it's capital markets and all stocks go to zero, in which case my investments will be one of my minor problems :)
3. the simplest definition of a "cheap, good deal" is just "something that recently dropped in price by a lot".

So with that in mind, I basically gathered all the available cash I could find (by selling some of my bond and gold ETFs, and by using some of my available 2x Regt-T leverage in my futures trading account) and bought the following stocks with the following reasoning:
1. QQQ - historically it grew much higher than the regular SPY ETF, maybe with higher volatility, but I don't have much cash and I want as much long-term equity exposure as I can get
2. VBR - it simply dropped by a lot, + the value risk-premium..
3. Oil companies (bought probably 5-8 different ones, tried to pick larger ones, but also those which dropped more in price) - COP, IMO, FANG, CVE, ENB, SU, CNQ..
4. all 3 cruise companies :) - I thought I just wouldn't be able to forgive myself if I don't buy something that fell 80% and it recovers back. So the regret of loosing all, if these companies go to zero, would be less than seeing them bouncing back and knowing that I could've bought in. Admittedly I only spent ~2k on all of them.
5. 2-3 airlines (AAL, ALK, BA) - same reasoning as with Cruise companies
6. assortment of financial + some telecom + REITS (APLE, SBRA, HST, STOR, IRM, GEO, O, MPV, BRX..) + random companies with low debt and some other parameters
7. EU and Asia (VPL, VWO, VGK)

Just to note, the total amount I spent on this wasn't extreme for me, and because the futures system slashed it's positions after making some gains I was able to dip into that unused cash with a good cushion still available.. so none of that would ruin me really, unless the majority of the world stocks went to zero :)

So now all of this went up, much faster than I expected, as I was fully prepared to wait 2-3 years, and I'm taking some gains off the table (still holding the cruise companies though :) ).

So I guess my question is in general, how dumb (genius :) ) of a thing was to do it ? I mean essentially I betted that the world will return to normal as a whole (because I diversified quite well among ETFS and individual stocks), and my risk was that.. well, "there will be no such thing in the world as cruises, airlines and oil companies anymore"..
The "wrong" thing was that I completely ignored my tactical target asset allocation and overweighed equities a lot more than "is appropriate", but I needed cash for a long term..

Also, are there some grounds for extracting a systematic rule out of it ? something like "buy something when it drops in price and sell back when it recovers even if it takes a long time"? Of course that "something" has to be things that in general go up over time like stocks. And of course the horrific negative skew of such a rule will be a problem, and also there should be a limit to the "depth of the hole" it's allowed to dig.. Or was it all a complete gamble and there's no wisdom to extract from it ?
 
Last edited:
I can't tell you whether it was dumb or genius (I still don't know) but I did the same thing, with exact same thinking :)
I bought a basket of 10 stocks, cruise companies, airlines, hotels and car rental companies, with the exact same idea, to hold for very long. I also knew that some of them might go bankrupt, and one did, Hertz.
Even with that loss, I'm also already up on that basket, and I sure wasn't hoping to be this soon. Since it wasn't a lot of money, I'll hold mine for a bit longer, probably at least until summer which is when current estimates say vaccine will be widely available.
 
Rebalancing to your target asset allocation would capture some of this buying the dip effect. Since the equity portion would be underweight, you would need to buy more of it to get back to target. Not to say anything about picking individual stocks though
 
Kernfusion said:
Hi, not sure how systematic (or dumb :) ) this question is, but at least maybe it'll humor someone.
So in March-April after everything went crashing down, many people went bargain-hunting but it wasn't very clear what was "a good deal to buy"..
So I thought to myself:
1. for my long-term investment portfolio I have a very long investment horizon, so I can easily wait years..
2. the world basically has 2 choices now: either we recover (maybe slowly, but see #1) or we just unwind this whole "society thing" with it's capital markets and all stocks go to zero, in which case my investments will be one of my minor problems :)
3. the simplest definition of a "cheap, good deal" is just "something that recently dropped in price by a lot".

So with that in mind, I basically gathered all the available cash I could find (by selling some of my bond and gold ETFs, and by using some of my available 2x Regt-T leverage in my futures trading account) and bought the following stocks with the following reasoning:
1. QQQ - historically it grew much higher than the regular SPY ETF, maybe with higher volatility, but I don't have much cash and I want as much long-term equity exposure as I can get
2. VBR - it simply dropped by a lot, + the value risk-premium..
3. Oil companies (bought probably 5-8 different ones, tried to pick larger ones, but also those which dropped more in price) - COP, IMO, FANG, CVE, ENB, SU, CNQ..
4. all 3 cruise companies :) - I thought I just wouldn't be able to forgive myself if I don't buy something that fell 80% and it recovers back. So the regret of loosing all, if these companies go to zero, would be less than seeing them bouncing back and knowing that I could've bought in. Admittedly I only spent ~2k on all of them.
5. 2-3 airlines (AAL, ALK, BA) - same reasoning as with Cruise companies
6. assortment of financial + some telecom + REITS (APLE, SBRA, HST, STOR, IRM, GEO, O, MPV, BRX..) + random companies with low debt and some other parameters
7. EU and Asia (VPL, VWO, VGK)

Just to note, the total amount I spent on this wasn't extreme for me, and because the futures system slashed it's positions after making some gains I was able to dip into that unused cash with a good cushion still available.. so none of that would ruin me really, unless the majority of the world stocks went to zero :)

So now all of this went up, much faster than I expected, as I was fully prepared to wait 2-3 years, and I'm taking some gains off the table (still holding the cruise companies though :) ).

So I guess my question is in general, how dumb (genius :) ) of a thing was to do it ? I mean essentially I betted that the world will return to normal as a whole (because I diversified quite well among ETFS and individual stocks), and my risk was that.. well, "there will be no such thing in the world as cruises, airlines and oil companies anymore"..
The "wrong" thing was that I completely ignored my tactical target asset allocation and overweighed equities a lot more than "is appropriate", but I needed cash for a long term..
More...

I did something similar in my long only portfolio. I wasn't aggressive as you; I invested in a broadly diversified set of UK equities with deliberate sector diversification* rather than, and I bought gradually over a period of several weeks (so though I called the bottom almost exactly, my average price was higher than it should have been). Strictly speaking the selection criteria was systematic, but the timing decision was - I am ashamed to say - discretionary (there's more on my COVID trading here).

* Actually I bought a UK high yielding ETF as a placeholder, then gradually liquidated this over the following few months and bought stocks, allowing me to pick up sectors starting with the cheapest. But like you I basically went very long equity Beta on a hunch in March.

Kernfusion said:
Also, are there some grounds for extracting a systematic rule out of it ? something like "buy something when it drops in price and sell back when it recovers even if it takes a long time"? Of course that "something" has to be things that in general go up over time like stocks. And of course the horrific negative skew of such a rule will be a problem, and also there should be a limit to the "depth of the hole" it's allowed to dig.. Or was it all a complete gamble and there's no wisdom to extract from it ?
More...

Well the trite answer is "buy something when it drops in price and sell back when it recovers even if it takes a long time"?: dude you just discovered value investing...

More serious answer: If you've read 'Systematic Trading' you might recall a story I tell about investing in BARC, at what would have been within 5% of the low in 2009, selling at something like a 400% profit. That was another 'once in a lifetime' trade, or certainly once in a decade. I'm not sure such trades can be systematised; there just aren't any data points and every crisis is different.

However what can be systematised is your portfolio and risk management. My regret in the BARC trade was that I panicked and put the trade on in tiny size, about 1 tenth of what I'd originally planned to do. This is because, at the time despite working for a systematic hedge fund, I didn't have a process for systematically managing my portfolio.

Now I do have that process. That means when I did have my instinctive, once in a decade, 'this is the bottom' feeling in March I was able to size and manage the trade using my existing systematic methodology. That meant I didn't do a complete pivot and fill my boots in one day, because that would have gone against my risk management. But it did mean I could put on a meaningful sized position without bricking it.

So what you can do is combine gut feeling with systematic position and risk (what I call 'semi-automatic trading' in the books). Or you can do what I do, which is trade systematically 99% of the time, and every now and then place a bet based on gut feeling, but still use your system to size and manage the position.

GAT

PS If anyone cares, here is what the final basket looks like:

Aviva PLC (LSE:AV.)
Babcock International Group PLC (LSE:BAB)
BT Group PLC (LSE:BT.A)
Centrica PLC (LSE:CNA)
Crest Nicholson Holdings PLC (LSE:CRST)
Direct Line Insurance Group PLC (LSE:DLG)
Dixons Carphone PLC (LSE:DC.)
Go-Ahead Group (The) PLC (LSE:GOG)
Greencoat UK Wind (LSE:UKW)
Hammerson PLC (LSE:HMSO)
Imperial Brands PLC (LSE:IMB)
Investec PLC (LSE:INVP)
ITV PLC (LSE:ITV)
Lloyds Banking Group PLC (LSE:LLOY)
Marks & Spencer Group PLC (LSE:MKS)
Morgan Advanced Materials PLC (LSE:MGAM)
Morgan Sindall Group PLC (LSE:MGNS)
Morrison (Wm) Supermarkets PLC (LSE:MRW)
Royal Mail PLC (LSE:RMG)
Senior PLC (LSE:SNR)
Rathbones
Greencore
Vesuvius
 
globalarbtrader said:
So what you can do is combine gut feeling with systematic position and risk (what I call 'semi-automatic trading' in the books). Or you can do what I do, which is trade systematically 99% of the time, and every now and then place a bet based on gut feeling, but still use your system to size and manage the position.
More...

Thanks GAT, yep, that's pretty-much what I'm doing normally, my long-term portfolio is constructed hierarchically based on your "Smart portfolios" book (75%EQ, 20%Bonds, 5% alternatives and down into the sub-groups from there). It's just in this case I decided that I want to place a bet like that because I'm still not retired, so there will be more cash coming into my portfolio so I could think that I'm just pre-allocating some of the upcoming cash by borrowing a little right now..
And also, I wasn't really that aggressive, I maybe increased the cash-amount of my equities by 1.4 from what it was before, and I was buying them over a couple of weeks too.. It's easier now to say "of course cruise companies would recover", but back then it felt like people will never step foot on one of these ships, so I only bought a "symbolic amount" of them, about 600-700$ each.
And btw, yes, you called bottom almost perfectly :)

So yeah, I guess that's what risk is - often there's the "normal outcome" that is expected to happen, but there's still a chance that it's not going to be normal and the risk of that is priced-in, so by taking that opposite side there's usually a reward.. Though not always, e.g. Japan index is still down from it's 1990 high..
 
globalarbtrader said:
1) That's actually quite a hard question to answer. I'd probably use 30 years as it's the easiest and most conservative.

2) Use post cost returns for the input. That assumes that costs are known with 100% certainty (assuming costs are known with x% certainty is another wicked problem), which is close enough to the truth; leaving the method to worry about the uncertainty of the pre-cost returns.

GAT
More...

Hello Robert,

How would you go about adjusting returns for the costs?
Suppose we have cost estimates in Sharpe ratio units "c" and returns series as "series".
I would do (for weekly return data):
adjustment = stdev(series) * c / sqrt(52)
series = series - adjustment

Then these series could be fed in to the portfolio weighting process.

This adjustment does not change stdev or correlations between the series of different assets. Seems quite a logical choice. Do you see any problems with that? Do you have a better approach?
 
You must log in or register to reply here.
Back
Top