Fully automated futures trading
- Thread starter globalarbtrader
- Start date
Glad to be of service....
Monthly update: Last one was November 4th
Up 12% of capital, or around £48K
Last month:
Longer term:
Well up to yesterday I was looking forward to telling you what a f**** genius I was, and actually posting at a HWM level for a change. Then Mr Draghi opened his big mouth and I dropped 4% very quickly indeed. Movements in european bonds were about 6-7 daily standard deviations (the much discussed Euro move was a mere 4). The system I use to manually alert me so I can check and filter large price movements triggered half a dozen times in a matter of minutes; normally it's a couple of months before each trigger.
It would have been even nastier but my hedged equity portfolio added a bit. Talking to friends last night in the CTA industry losses of 4% were fairly common (and I run at a higher vol target than most).
Still I'm still up for the week, and the month, so shouldn't really complain. You can barely see yesterdays move on the second set of charts above.
The real lesson here is that diversification and risk control are there exactly for this kind of situation. Which meant I could be relatively sanguine. Even I however find it hard not to be emotionally attached to day to day p&l. The difference is that it makes no difference whatsoever to the way my system trades, since I don't let my system know how bad or good I'm feeling
Drawdown: 6.2% (off yesterday morning's new HWM)
Gainers:
Copper 8200
Platinum 8125
Gold 6700
Gas 6000
Crude 4500
Pallad 3700
Livecow 2600
Losers:
Eurodollars -2000
NASDAQ -2400
SP500 -2500
Another vote for diversification. Over the last few years the biggest money making sectors in trend following have been financials - equities, bonds, rates and vol; especially bonds. Last year in particular european bonds were an absolute barn stormer of a trade. In contrast many commodity hedge funds have shut their doors recently.
But over the last 30 days I've made nothing in bonds (actually small losses in European bonds, not big enough to trouble the figures above) but commodities have been wonderful. I hope the guy who posted on this thread saying 'why don't you just pick the best markets' is reading this.
(There is also a link here to the blog post I made on equity curve trading - again we're seeing that the returns of trend following systems tend to be negatively autocorrelated - bad follows good and vice versa, rather than the other way round...)
Positions:
Risk:
(there is some discreprancy in positions because I've closed long BTP, OAT and short EUR positions this morning; after my nightly risk report ran but before I generated the list of positions above)
Trades
Expected slippage £549; actual £499
Lots of rolls (its that time of year).
I said last time that I was putting off refactoring my code; well I've decided to put it off no longer and I've begun to rewrite everything in python 3, nicely factored, properly tested and documented, the plan being to release it open souce as I do it. I'm starting with the backtesting engine, since that is the part that is dodgiest, and also because with a good backtesting engine I can be more productive with my research.
I've also just got a contract to write a more accessible book, about portfolio construction and asset allocation; and I've agreed to contribute a chapter to a book on trading rules. So it will be a busy new year.
If you're in London next week I'm speaking at traders expo. If not, and if I'm not in touch on here again before then, merry christmas (or generic seasonal greetings to those of other faiths), happy new year (and again, if you're on a different calendar, many happy wishes for whatever significant dates fall between now and January 2nd on my calendar).
GAT
Up 12% of capital, or around £48K
Last month:
Longer term:
Well up to yesterday I was looking forward to telling you what a f**** genius I was, and actually posting at a HWM level for a change. Then Mr Draghi opened his big mouth and I dropped 4% very quickly indeed. Movements in european bonds were about 6-7 daily standard deviations (the much discussed Euro move was a mere 4). The system I use to manually alert me so I can check and filter large price movements triggered half a dozen times in a matter of minutes; normally it's a couple of months before each trigger.
It would have been even nastier but my hedged equity portfolio added a bit. Talking to friends last night in the CTA industry losses of 4% were fairly common (and I run at a higher vol target than most).
Still I'm still up for the week, and the month, so shouldn't really complain. You can barely see yesterdays move on the second set of charts above.
The real lesson here is that diversification and risk control are there exactly for this kind of situation. Which meant I could be relatively sanguine. Even I however find it hard not to be emotionally attached to day to day p&l. The difference is that it makes no difference whatsoever to the way my system trades, since I don't let my system know how bad or good I'm feeling
Drawdown: 6.2% (off yesterday morning's new HWM)
Gainers:
Copper 8200
Platinum 8125
Gold 6700
Gas 6000
Crude 4500
Pallad 3700
Livecow 2600
Losers:
Eurodollars -2000
NASDAQ -2400
SP500 -2500
Another vote for diversification. Over the last few years the biggest money making sectors in trend following have been financials - equities, bonds, rates and vol; especially bonds. Last year in particular european bonds were an absolute barn stormer of a trade. In contrast many commodity hedge funds have shut their doors recently.
But over the last 30 days I've made nothing in bonds (actually small losses in European bonds, not big enough to trouble the figures above) but commodities have been wonderful. I hope the guy who posted on this thread saying 'why don't you just pick the best markets' is reading this.
(There is also a link here to the blog post I made on equity curve trading - again we're seeing that the returns of trend following systems tend to be negatively autocorrelated - bad follows good and vice versa, rather than the other way round...)
Positions:
Code:
code contractid positions Lock WrongContract InFwdNotRoll
1 AUD 201512 1 False False False
19 BTP 201603 2 False False False
3 COPPER 201603 -3 False False False
18 CORN 201612 -5 False False False
15 CRUDE_W 201612 -3 False False False
5 EDOLLAR 201906 1 False False False
16 EDOLLAR 201903 7 False False False
23 EUROSTX 201512 -13 False False False
22 GAS_US 201602 -4 False False False
10 GBP 201512 -1 False False False
0 GOLD 201602 -2 False False False
12 JPY 201512 -2 False False False
13 KR3 201512 7 False False False
2 LEANHOG 201606 -1 False False False
20 LIVECOW 201610 -2 False False False
9 MXP 201512 -2 False False False
7 OAT 201603 1 False False False
17 PALLAD 201603 -1 False False False
25 PLAT 201601 -3 False False False
6 SOYBEAN 201611 1 False False False
11 SP500 201512 1 False False False
4 US2 201603 3 False False False
21 US5 201603 1 False False False
24 V2X 201601 2 False False False
14 VIX 201601 -1 False False False
8 WHEAT 201612 -3 False False FalseRisk:
Code:
Expected annual risk more than GBP6400 per year, GBP400 per day
code multisignal expected_annual_risk expected_annual_risk_per_contract position expected_annual_risk_rounded_pos
Shorts:
27 JPY -12.7 8441 4071 -2 8141
17 VIX -10.5 6982 8252 -1 8252
0 CORN -12.9 8602 1960 -5 9801
26 GBP -13.4 8910 5591 -2 11182
4 WHEAT -17.7 11795 3839 -3 11516
2 LIVECOW -13.8 9167 5905 -2 11811
25 EUR -13.1 8712 13394 -1 13394
32 PALLAD -27.7 18387 14223 -1 14223
33 PLAT -26.8 17799 6169 -3 18508
35 GAS_US -26.8 17799 4668 -4 18673
31 GOLD -31.7 21045 9617 -2 19234
30 COPPER -26.8 17799 7400 -3 22199
34 CRUDE_W -31.7 21045 8483 -3 25448
Longs:
10 OAT 19.9 13229 5015 2 10031
36 EDOLLAR 16.2 10751 1543 8 12341
8 BTP 23.5 15603 5042 3 15125(there is some discreprancy in positions because I've closed long BTP, OAT and short EUR positions this morning; after my nightly risk report ran but before I generated the list of positions above)
Trades
Code:
code contractid filled_datetime filledtrade filledprice
7363 AEX 201511 2015-11-10 07:30:04 -1 465.600000
7309 AUD 201512 2015-11-06 13:58:23 -1 0.704900
7483 AUD 201512 2015-11-19 06:47:22 1 0.715200
7609 AUD 201512 2015-12-02 02:54:38 1 0.731200
7576 BOBL 201512 2015-12-01 08:20:06 -2 130.000000
7579 BOBL 201603 2015-12-01 08:20:06 2 131.580000
7672 BOBL 201603 2015-12-04 08:39:34 -2 130.670000
7297 BTP 201512 2015-11-06 07:36:20 -1 137.840000
7426 BTP 201512 2015-11-16 07:03:27 1 139.160000
7489 BTP 201512 2015-11-19 08:03:17 1 139.910000
7588 BTP 201512 2015-12-01 09:01:29 -3 140.590000
7591 BTP 201603 2015-12-01 09:01:29 3 138.970000
7675 BTP 201603 2015-12-04 08:43:09 -1 137.010000
7411 CAC 201511 2015-11-13 07:31:44 -1 4831.500000
7501 COPPER 201512 2015-11-23 12:12:21 -1 2.023500
7612 COPPER 201512 2015-12-02 12:02:38 2 2.063500
7615 COPPER 201603 2015-12-02 12:02:38 -2 2.067000
7627 COPPER 201512 2015-12-02 12:09:35 1 2.062500
7630 COPPER 201603 2015-12-02 12:09:35 -1 2.066000
7384 CORN 201612 2015-11-11 13:38:56 -1 391.250000
7471 CORN 201612 2015-11-17 13:33:00 -1 390.250000
7603 CRUDE_W 201612 2015-12-01 14:35:40 -1 48.340000
7312 EDOLLAR 201812 2015-11-06 14:03:04 -1 97.865000
7324 EDOLLAR 201812 2015-11-06 15:49:03 -1 97.830000
7327 EDOLLAR 201812 2015-11-06 17:52:36 -1 97.825000
7390 EDOLLAR 201812 2015-11-11 16:05:20 -1 97.825000
7453 EDOLLAR 201812 2015-11-17 11:25:55 -5 97.900000
7456 EDOLLAR 201903 2015-11-17 11:56:49 5 97.825000
7564 EDOLLAR 201906 2015-11-27 12:43:32 1 97.875000
7657 EDOLLAR 201903 2015-12-03 17:53:41 -1 97.835000
7645 EUR 201512 2015-12-03 14:23:41 1 1.079700
7666 EUR 201512 2015-12-04 03:09:40 1 1.092500
7300 GAS_US 201601 2015-11-06 12:02:15 1 2.517000
7504 GAS_US 201601 2015-11-23 12:23:50 1 2.249000
7507 GAS_US 201602 2015-11-23 12:23:50 -1 2.293000
7510 GAS_US 201601 2015-11-23 12:29:31 2 2.254000
7513 GAS_US 201602 2015-11-23 12:29:31 -2 2.298000
7516 GAS_US 201601 2015-11-23 12:35:04 1 2.255000
7519 GAS_US 201602 2015-11-23 12:35:04 -1 2.299000
7291 GBP 201512 2015-11-05 13:42:30 -1 1.525000
7294 GBP 201512 2015-11-06 06:16:28 -1 1.528600
7480 GBP 201512 2015-11-19 01:50:36 1 1.528400
7552 GBP 201512 2015-11-25 03:36:27 -1 1.510000
7669 GBP 201512 2015-12-04 03:24:32 1 1.512300
7288 GOLD 201512 2015-11-05 12:14:42 -1 1107.600000
7465 GOLD 201512 2015-11-17 11:35:05 -1 1078.800000
7495 GOLD 201512 2015-11-19 17:03:29 1 1083.000000
7555 GOLD 201512 2015-11-25 12:22:44 -1 1073.000000
7633 GOLD 201512 2015-12-02 12:15:19 3 1067.100000
7636 GOLD 201602 2015-12-02 12:15:19 -3 1066.700000
7648 GOLD 201602 2015-12-03 14:41:32 1 1056.700000
7441 JPY 201512 2015-11-17 01:31:53 -1 0.008112
7492 JPY 201512 2015-11-19 16:57:33 1 0.008150
7663 JPY 201512 2015-12-03 19:42:51 1 0.008167
7360 KOSPI 201512 2015-11-10 01:14:46 -1 248.100000
7336 KR10 201512 2015-11-09 01:12:40 -1 124.380000
7396 KR10 201512 2015-11-12 01:23:10 -1 123.900000
7348 KR3 201512 2015-11-09 02:10:32 -1 109.120000
7357 KR3 201512 2015-11-10 00:31:57 -1 109.130000
7402 KR3 201512 2015-11-12 01:26:09 -1 109.070000
7351 LEANHOG 201606 2015-11-09 15:48:20 -1 72.875000
7474 LEANHOG 201606 2015-11-17 14:03:03 -1 71.100000
7606 LEANHOG 201606 2015-12-01 16:16:39 1 75.075000
7420 LIVECOW 201610 2015-11-13 15:08:15 -1 124.575000
7321 MXP 201512 2015-11-06 15:36:13 1 0.059230
7405 NASDAQ 201512 2015-11-12 13:42:52 -1 4619.250000
7639 NASDAQ 201512 2015-12-02 14:29:52 1 4719.000000
7660 NASDAQ 201512 2015-12-03 18:00:40 -1 4621.000000
7477 OAT 201512 2015-11-18 07:03:30 1 153.830000
7486 OAT 201512 2015-11-19 07:05:39 1 154.150000
7582 OAT 201512 2015-12-01 08:31:38 -2 153.910000
7585 OAT 201603 2015-12-01 08:31:38 2 151.810000
7678 OAT 201603 2015-12-04 08:43:59 -1 150.000000
7378 PALLAD 201512 2015-11-11 11:46:34 -1 596.250000
7618 PALLAD 201512 2015-12-02 12:04:00 1 540.200000
7621 PALLAD 201603 2015-12-02 12:04:00 -1 540.450000
7303 PLAT 201601 2015-11-06 12:17:00 -1 954.300000
7414 PLAT 201601 2015-11-13 13:44:35 -1 868.600000
7366 SMI 201512 2015-11-10 07:33:43 -1 8900.000000
7570 SMI 201512 2015-11-30 13:57:39 1 9026.000000
7651 SMI 201512 2015-12-03 15:34:19 -1 8837.000000
7306 SOYBEAN 201611 2015-11-06 13:07:16 -1 875.250000
7369 SOYBEAN 201611 2015-11-11 11:34:12 -1 870.000000
7549 SOYBEAN 201611 2015-11-24 12:37:55 1 884.500000
7558 SOYBEAN 201611 2015-11-25 15:26:12 1 886.000000
7567 SOYBEAN 201611 2015-11-27 14:47:41 1 895.250000
7600 SOYBEAN 201611 2015-12-01 14:30:00 1 902.250000
7624 SOYBEAN 201611 2015-12-02 12:07:59 1 909.000000
7408 SP500 201512 2015-11-12 14:46:14 -1 2055.500000
7561 SP500 201512 2015-11-26 15:36:29 1 2095.000000
7318 US2 201512 2015-11-06 14:12:16 -1 109.007812
7498 US2 201512 2015-11-20 15:43:37 1 109.078125
7522 US2 201512 2015-11-23 14:04:12 -3 109.015625
7525 US2 201603 2015-11-23 14:04:12 3 108.726562
7315 US5 201512 2015-11-06 14:05:16 -1 118.664062
7573 US5 201603 2015-11-30 15:10:48 1 118.687500
7429 V2X 201512 2015-11-16 08:35:19 -4 23.500000
7432 V2X 201601 2015-11-16 08:35:19 4 25.200000
7546 V2X 201601 2015-11-24 09:17:10 -1 23.800000
7642 V2X 201601 2015-12-03 09:22:41 -1 22.400000
7447 VIX 201512 2015-11-17 09:29:54 1 17.750000
7450 VIX 201601 2015-11-17 09:58:30 -1 18.350000
7459 WHEAT 201612 2015-11-17 11:29:15 -1 518.000000
7462 WHEAT 201612 2015-11-17 11:30:28 1 518.750000Expected slippage £549; actual £499
Lots of rolls (its that time of year).
I said last time that I was putting off refactoring my code; well I've decided to put it off no longer and I've begun to rewrite everything in python 3, nicely factored, properly tested and documented, the plan being to release it open souce as I do it. I'm starting with the backtesting engine, since that is the part that is dodgiest, and also because with a good backtesting engine I can be more productive with my research.
I've also just got a contract to write a more accessible book, about portfolio construction and asset allocation; and I've agreed to contribute a chapter to a book on trading rules. So it will be a busy new year.
If you're in London next week I'm speaking at traders expo. If not, and if I'm not in touch on here again before then, merry christmas (or generic seasonal greetings to those of other faiths), happy new year (and again, if you're on a different calendar, many happy wishes for whatever significant dates fall between now and January 2nd on my calendar).
GAT
Nice update, thank you.
If you get a chance, perhaps next time you update it would be good to know your thoughts on the recent AHL team co-authored paper (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2695101).
What's the response function stuff about in the momentum section of it?
Cheers,
Q1
If you get a chance, perhaps next time you update it would be good to know your thoughts on the recent AHL team co-authored paper (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2695101).
What's the response function stuff about in the momentum section of it?
Cheers,
Q1
Extreme values of momentum don't translate into larger trends (think - dead cat bounce)
I use a simple cap to get round this. However AHL use something where larger signals are actually downweighted.
The blue line shows the original signal; the red line applying the response function.
Note for very large signals (bigger than 1.5- AHL scale their forecasts so they have a standard deviation of 1.0) the forecast after applying a response is actually getting smaller as we get more confident.
There is some statistical evidence that this makes sense, but it is quite weak and doesn't seem to be around in every asset class or speed of momentum. Also it leads to weird behaviour - when a trend strengthens you start closing your position; then when it weakens again you open your position up before closing it again. Which is why I didn't keep this feature when I built my own system.
(Technical issue: you also need to make sure the distribution of the forecast after applying a response function is correct)
GAT
Hi, short question: is that a typo at the end of your trades list:
You are selling the same contract (Dec 2016) and buy it back one minute later? Or is it a roll and one of them is the Dec 2015 contract?
Another question: Do you decide on trades using the previous day settle/close price or the current price?
If latter, how do you avoid that the position signal is, e.g. 0.51 -> rounded to 1 so you buy one contract and soon after it drops to 0.49 again, forcing you to sell one contract?
Or do you only trade each contract once per day, ignoring all latter signals?
Thank you for your time answering questions (and your book, really good read).
Michael
Code:
7459 WHEAT 201612 2015-11-17 11:29:15 -1 518.000000
7462 WHEAT 201612 2015-11-17 11:30:28 1 518.750000You are selling the same contract (Dec 2016) and buy it back one minute later? Or is it a roll and one of them is the Dec 2015 contract?
Another question: Do you decide on trades using the previous day settle/close price or the current price?
If latter, how do you avoid that the position signal is, e.g. 0.51 -> rounded to 1 so you buy one contract and soon after it drops to 0.49 again, forcing you to sell one contract?
Or do you only trade each contract once per day, ignoring all latter signals?
Thank you for your time answering questions (and your book, really good read).
Michael
No it's really a sell and a buy.
In chapter 11 I describe how you wouldn't change your position unless the rounded position has moved by more than 10%. I use something similar to that, which has a similar effect.
I trade based on daily prices, but using the last price in day. Since I sample those more than once per day, it means there can potentially be edge cases where I will buy and sell in the same day. I can wait quite a long time for a fill to happen (hours if needed); long enough so that another price has come in and I change my mind about what position I want on within a minute.
All this is probably too complicated and I plan to switch to just using the previous days close in the next few months.
GAT
That's another thing that I did not fully understand: Only trading when the rounded positions differ more than 10% only makes a difference when the positions are greater than 10, I would think? Otherwise any difference is more than 10%.
Trading too often by mistake is one of the things that I try to avoid. My nightmare is a run away algo trading program that keeps buying and selling the same thing every minute or so, spending comissions and spread.
Trading too often by mistake is one of the things that I try to avoid. My nightmare is a run away algo trading program that keeps buying and selling the same thing every minute or so, spending comissions and spread.
You're right. That's why I use a slightly more complex method; but it is a bit involved to include in the book. I'll blog about it some day.
Agree with the nightmare. I have a lot of controls to limit the damage; for example I can't trade more than 5 wheat contracts a day.
GAT