Fully automated futures trading
- Thread starter globalarbtrader
- Start date
Interesting question. But this is a bad idea.... There isn't enough data in the backtest (at most 35 years, but on average about 15 years, and in some cases just 2 years) to prove statistically that instrument A is "probably" better than instrument B.
Number of instruments with data, by year. From http://qoppac.blogspot.co.uk/2015/03/simulating-my-futures-system.html
And this means one year of live trading data is wholly inadequate. This is because the differences in performance aren't sufficiently large to find a difference.
From table 6 of my forthcoming book given an average correlation between instruments of 0.5 I'd only need 8 years of data if one instrument had a Sharpe Ratio (SR) that was 1.0 higher than another (say a SR of 1.5 compared to one of 0.5). However there just aren't any instruments with that kind of SR difference over 8 years or more. More commonly the difference in performance is perhaps 0.5 or 0.25 SR units. With that kind of 'advantage' I'd need 25 years and 40 years of data respectively. I just don't have it.
GAT
Ha! Yes forgot that one. But one of the nice advantages of being an independent trader is to go on holiday whenever you feel like it.....
GAT
But even 1 year data is like 250 trading days, that is a lots of data and statistically signifficant number. Also markets change over time, so just because the backtest worked in let's say 2006, that doesn't mean the strategy still works today with the same instrument.
I am curious about the breakdown of the 40 instruments as P/L goes? Let's say 10-15 very well performing, 15-20 so-so, and 10-15 losers. What is the point of carrying the loser?
Another question: How about correlated markets? Even if it is 2 futures, if they are strongly correlated, they can be treated as one. This is including inversely correlated markets too. Dollar/gold is such an example. The point here is that just because you have 40 instruments, you might only trade 25 uncorrelated markets or even less....
I am curious about the breakdown of the 40 instruments as P/L goes? Let's say 10-15 very well performing, 15-20 so-so, and 10-15 losers. What is the point of carrying the loser?
Another question: How about correlated markets? Even if it is 2 futures, if they are strongly correlated, they can be treated as one. This is including inversely correlated markets too. Dollar/gold is such an example. The point here is that just because you have 40 instruments, you might only trade 25 uncorrelated markets or even less....
The point of 'carrying the loser' is that you don't know in future which markets will be the losers, and which will be the winners. For this to be the case three things would have to be true:
- there would have to be some reason which explains why A is winning and B is losing (or it is just likely to be a statistical fluke). For the kind of trading I do, that's usually unlikely. For HFT stuff working on market microstructure with for example different priority rules and rebates it is more plausible.
- there wouldn't be any changes in this relative structure; eg the future would be exactly like the past. Even for HFT this is a stretch, because different players come into the market and change the way it works.
- there would have to be sufficient statistical evidence to show that market A was better than market B
So we're back to statistics again. It's not true that using daily data rather than say monthly or annual data will give you more statistical significance. This is because when you go down to daily data you increase the amount of noise. This exactly compensates for the increase in the number of data points (to be technical, both effects scale at square root of time, ignoring auto-correlation effects which don't make that much difference to the result).
As for 'it worked before but might not work today' I'm trading systems with a relatively slow holding period, at least by the febrile standards of this board. My base assumption is that they always work on all markets for all time. I will never have enough data to prove otherwise. I will never have enough data to prove that I should be trading something else. I will never have enough data to prove that they only work for certain markets.
You are correct that there is no point having 100% correlated markets. However it depends on how much capital you have as to whether you would bother with markets that are 75% correlated, or 99% correlated. All things being equal I'd rather have more markets than less. And with more capital I will be adding markets where the correlation is higher, than for someone with less capital (see http://www.elitetrader.com/et/index.php?threads/a-tale-of-two-positions.292751/#post-4147857 for more explanation on this point).
With enough capital to trade 40 markets, that's what I should do. If I went down to 15 markets by dropping the most correlated pairs I'd reduce my expected Sharpe Ratio by about 15%, or in money terms about £13,000 a year. Why would I voluntarily want to give that up?
(A minor point to make is that what matters is not the correlation between prices that matters, but between the returns of the trading subsystems. The correlation of the latter is lower, so the diversification effect of adding more instruments is higher than you might think)
GAT
There's a new post in my current series here: http://www.elitetrader.com/et/index...atic-trading-system-part-5-accounting.293403/
Please leave any constructive discussion at that thread
GAT
Please leave any constructive discussion at that thread
GAT
It's been about a month since my last update; I've actually been back from holiday a few days but a month is an easier reporting period to deal with.
I discussed in a previous post what action I should take whilst away. In the end I didn't do anything... and for nearly all the time I didn't have internet or even read a newspaper... I can hear you gasping.... why did I miss anything?!
For most of the holiday not very much probably happened, but then with about a week to go apparently things got a little interesting. Net-Net on the moves of the last couple of weeks I'm flat or slightly up, but it has been a wild ride. I hit a new HWM when the equity market fell of a cliff, but gave most of it back on the recovery:
Anyway as of this morning my profit since the last update was around £20K. Current drawdown (off the new, higher HWM don't forget) is 6.9%
Gainers:
KOSPI £5400
MXP £3400
EDOLLAR £3700
PALLAD £3200
AUD £2900
KR3 £2500
Losers:
AEX -£2300
CAC -£2300
JPY -£5400
PLAT -£6900
GOLD -£7600
VIX -£12700
That big VIX loss wasn't really a surprise – I'd sold 3 contracts at a price of 16 or less, then it spiked above 20 very rapidly. Though I cut my position by then the damage was done. There is a similar story with Gold and everything else you can see here. This is why diversification is so important... even within the metals bucket I had radically different p&l. Referencing the discussion on here before, which of the metals should I choose to trade?
Anyway as of this morning my profit since the last update was around £20K. Current drawdown (off the new, higher HWM don't forget) is 6.9%
Gainers:
KOSPI £5400
MXP £3400
EDOLLAR £3700
PALLAD £3200
AUD £2900
KR3 £2500
Losers:
AEX -£2300
CAC -£2300
JPY -£5400
PLAT -£6900
GOLD -£7600
VIX -£12700
That big VIX loss wasn't really a surprise – I'd sold 3 contracts at a price of 16 or less, then it spiked above 20 very rapidly. Though I cut my position by then the damage was done. There is a similar story with Gold and everything else you can see here. This is why diversification is so important... even within the metals bucket I had radically different p&l. Referencing the discussion on here before, which of the metals should I choose to trade? Same question for currencies?
Current positions
Current risk
Although I have fewer, and smaller, positions due to higher market vol the current expected portfolio risk as of last night was virtually unchanged @ 120% of average; on 90% of maximum capital (due to the drawdown) or in money terms £6768 a day or £108K a year.
Due to a smaller number of positions I am no longer hitting any risk limits (to put it another way, my exposure to correlation risk has fallen).
Trades
Slippage – expected £648, actual £421
A lot of rolls, but also just a lot of trading because of the markets wild swings. Even profitable positions like MXP had to be traded to adjust my risk as the turmoil affected it.
In other news I've published my final blog post on system construction, any discussion here please http://www.elitetrader.com/et/index...ed-trading-system-checks-and-balances.294024/
And for those who don't follow me on social media, you might not have seen the first review of my new book:
"A remarkable look inside systematic trading never seen before, spanning the range from small to institutional traders. This isn't only for algorithmic traders, it's valuable for anyone needing a structure - which is all of us. Carver explains how to properly test, apply constant risk, size positions and portfolios, and my favorite, his "no rule" trading rule, all explained with scenarios. Reading this will benefit all traders."
Perry Kaufman, author of Trading Systems and Methods.
I'm a massive fan of Perry, and very grateful for this endorsement.
The official release of the book is September 15th, but some of the people who pre-ordered the ebook have received their copies already.
Happy trading
GAT
I discussed in a previous post what action I should take whilst away. In the end I didn't do anything... and for nearly all the time I didn't have internet or even read a newspaper... I can hear you gasping.... why did I miss anything?!
For most of the holiday not very much probably happened, but then with about a week to go apparently things got a little interesting. Net-Net on the moves of the last couple of weeks I'm flat or slightly up, but it has been a wild ride. I hit a new HWM when the equity market fell of a cliff, but gave most of it back on the recovery:
Anyway as of this morning my profit since the last update was around £20K. Current drawdown (off the new, higher HWM don't forget) is 6.9%
Gainers:
KOSPI £5400
MXP £3400
EDOLLAR £3700
PALLAD £3200
AUD £2900
KR3 £2500
Losers:
AEX -£2300
CAC -£2300
JPY -£5400
PLAT -£6900
GOLD -£7600
VIX -£12700
That big VIX loss wasn't really a surprise – I'd sold 3 contracts at a price of 16 or less, then it spiked above 20 very rapidly. Though I cut my position by then the damage was done. There is a similar story with Gold and everything else you can see here. This is why diversification is so important... even within the metals bucket I had radically different p&l. Referencing the discussion on here before, which of the metals should I choose to trade?
Gainers:
KOSPI £5400
MXP £3400
EDOLLAR £3700
PALLAD £3200
AUD £2900
KR3 £2500
Losers:
AEX -£2300
CAC -£2300
JPY -£5400
PLAT -£6900
GOLD -£7600
VIX -£12700
That big VIX loss wasn't really a surprise – I'd sold 3 contracts at a price of 16 or less, then it spiked above 20 very rapidly. Though I cut my position by then the damage was done. There is a similar story with Gold and everything else you can see here. This is why diversification is so important... even within the metals bucket I had radically different p&l. Referencing the discussion on here before, which of the metals should I choose to trade? Same question for currencies?
Current positions
Code:
code contractid positions Lock WrongContract InFwdNotRoll
1 AUD 201509 -3 False False False
5 BOBL 201512 3 False False False
23 BTP 201512 1 False False False
17 COPPER 201512 -2 False False False
18 CORN 201512 -3 False False False
11 CRUDE_W 201512 -1 False False False
21 EDOLLAR 201812 11 False False False
12 EUROSTX 201509 -13 False False False (HEDGE)
3 GAS_US 201511 -4 False False False
7 GBP 201509 -3 False False False
10 KR10 201509 2 False False False
6 KR3 201509 11 False False False
0 LEANHOG 201606 1 False False False
19 LIVECOW 201610 -1 False False False
8 MXP 201509 -6 False False False
16 NZD 201509 -2 False False False
22 PALLAD 201512 -1 False False False
13 PLAT 201510 -2 False False False
14 SOYBEAN 201511 -2 False False False
2 US10 201512 1 False False False
4 US2 201512 3 False False False
20 US5 201512 1 False False False
15 V2X 201510 1 False False False
24 V2X 201511 1 False False False
25 VIX 201510 -1 False False False
9 WHEAT 201512 -3 False False FalseCurrent risk
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
17 VIX -20.1 13098 9846 -1 9846
0 CORN -16.0 10411 3469 -3 10406
29 NZD -19.3 12562 6168 -2 12337
28 MXP -17.7 11499 2104 -6 12624
4 WHEAT -18.0 11695 4237 -3 12710
26 GBP -21.4 13921 4737 -3 14210
34 CRUDE_W -26.0 16928 14271 -1 14271
33 PLAT -16.7 10837 7345 -2 14691
3 SOYBEAN -22.3 14538 8074 -2 16147
24 AUD -22.6 14719 5384 -3 16152
32 PALLAD -21.6 14078 16835 -1 16835
30 COPPER -29.6 19257 9453 -2 18907
35 GAS_US -33.8 21973 5215 -4 20860
36 EDOLLAR 24.2 15720 1294 11 14229Although I have fewer, and smaller, positions due to higher market vol the current expected portfolio risk as of last night was virtually unchanged @ 120% of average; on 90% of maximum capital (due to the drawdown) or in money terms £6768 a day or £108K a year.
Due to a smaller number of positions I am no longer hitting any risk limits (to put it another way, my exposure to correlation risk has fallen).
Trades
Code:
code contractid filled_datetime filledtrade filledprice
4741 AEX 201509 2015-08-06 13:17:37 1 500.200000
4777 AEX 201509 2015-08-13 07:56:45 -1 483.700000
4990 AUD 201509 2015-08-26 01:28:57 -1 0.710400
5086 AUD 201509 2015-09-03 03:31:48 -1 0.702000
4843 BOBL 201509 2015-08-17 07:28:42 1 130.510000
4927 BOBL 201509 2015-08-25 10:02:17 -3 130.470000
4930 BOBL 201512 2015-08-25 10:02:17 3 128.620000
4846 BTP 201509 2015-08-17 08:29:10 1 136.880000
4867 BTP 201509 2015-08-18 14:54:13 -1 136.480000
4912 BTP 201509 2015-08-25 09:36:09 -1 135.330000
5008 BTP 201509 2015-08-27 09:41:02 -1 134.880000
5011 BTP 201512 2015-08-27 09:41:02 1 133.200000
5014 BTP 201512 2015-08-27 09:45:49 -1 133.280000
5035 BTP 201512 2015-08-28 10:49:52 1 133.750000
5077 BTP 201512 2015-09-02 15:25:36 -1 132.710000
5113 BTP 201512 2015-09-03 15:24:02 1 133.640000
4822 BUND 201509 2015-08-14 07:30:53 1 155.040000
4957 BUND 201509 2015-08-25 13:02:24 -1 154.020000
4960 BUND 201512 2015-08-25 13:02:24 1 154.210000
4996 BUND 201512 2015-08-26 07:26:07 -1 153.980000
4783 CAC 201509 2015-08-13 07:59:42 -1 5005.500000
4999 CAC 201509 2015-08-26 07:59:04 -1 4494.000000
5017 CAC 201509 2015-08-27 10:00:17 1 4640.500000
4756 COPPER 201512 2015-08-10 13:07:10 -1 2.348000
4762 COPPER 201512 2015-08-10 16:15:42 1 2.398500
4810 COPPER 201512 2015-08-13 12:10:50 -1 2.347000
5029 COPPER 201512 2015-08-27 16:55:39 1 2.334500
4852 CORN 201512 2015-08-17 15:15:26 -1 375.250000
5074 CORN 201512 2015-09-02 14:30:00 -1 368.750000
4798 CRUDE_W 201512 2015-08-13 11:57:17 -1 46.150000
5032 CRUDE_W 201512 2015-08-27 17:00:54 1 43.400000
5071 CRUDE_W 201512 2015-09-02 14:12:30 1 47.020000
4807 EDOLLAR 201812 2015-08-13 12:04:23 1 97.705000
4831 EDOLLAR 201812 2015-08-14 12:03:57 1 97.710000
4873 EDOLLAR 201812 2015-08-19 12:00:30 1 97.700000
4945 EDOLLAR 201809 2015-08-25 11:58:50 -2 97.985000
4948 EDOLLAR 201812 2015-08-25 11:58:50 2 97.885000
5005 EDOLLAR 201812 2015-08-26 17:55:45 1 97.875000
4744 EUR 201509 2015-08-06 13:59:49 -1 1.088900
4753 EUR 201509 2015-08-10 01:28:17 1 1.096000
4801 GAS_US 201510 2015-08-13 11:59:15 1 2.934000
4849 GAS_US 201511 2015-08-17 12:14:52 -1 2.903000
4864 GAS_US 201511 2015-08-18 12:11:41 -1 2.847000
4882 GAS_US 201511 2015-08-19 17:52:57 -1 2.824000
4951 GAS_US 201510 2015-08-25 12:02:37 1 2.690000
4954 GAS_US 201511 2015-08-25 12:02:37 -1 2.768000
4750 GBP 201509 2015-08-07 01:46:12 -1 1.551000
4774 GBP 201509 2015-08-13 00:59:57 1 1.560800
4918 GBP 201509 2015-08-25 09:39:34 1 1.578900
4987 GBP 201509 2015-08-25 19:26:17 -1 1.568500
5023 GBP 201509 2015-08-27 15:52:25 -1 1.538500
5041 GBP 201509 2015-08-31 01:35:03 -1 1.542000
5083 GBP 201509 2015-09-03 01:34:12 -1 1.530000
4768 GOLD 201512 2015-08-11 12:07:07 1 1111.500000
4834 GOLD 201512 2015-08-14 12:14:54 1 1120.000000
4891 GOLD 201512 2015-08-20 12:14:06 1 1138.200000
4885 JPY 201509 2015-08-20 01:40:06 1 0.008076
4903 JPY 201509 2015-08-24 00:57:14 1 0.008216
4909 JPY 201509 2015-08-25 00:57:57 1 0.008411
4870 KOSPI 201509 2015-08-19 02:58:19 -1 233.850000
5038 KOSPI 201509 2015-08-31 01:01:27 1 230.350000
5131 KOSPI 201509 2015-09-04 02:36:39 1 229.200000
4894 KR10 201509 2015-08-21 01:40:29 1 124.320000
4840 KR3 201509 2015-08-17 01:47:50 1 109.340000
4897 KR3 201509 2015-08-21 02:00:22 1 109.450000
5128 KR3 201509 2015-09-04 01:46:33 1 109.610000
5116 LIVECOW 201510 2015-09-03 16:55:29 1 142.450000
5119 LIVECOW 201610 2015-09-03 16:55:29 -1 135.100000
4765 MXP 201509 2015-08-10 18:11:37 1 0.061870
4792 MXP 201509 2015-08-13 09:59:07 -1 0.061240
4855 MXP 201509 2015-08-18 02:24:11 -1 0.060690
4921 MXP 201509 2015-08-25 09:40:38 1 0.058530
4936 MXP 201509 2015-08-25 11:26:57 1 0.059000
4993 MXP 201509 2015-08-26 01:39:01 -1 0.057770
5026 MXP 201509 2015-08-27 16:17:41 1 0.059320
4858 OAT 201509 2015-08-18 07:32:41 1 149.860000
4876 OAT 201509 2015-08-19 15:21:25 -1 149.340000
4888 OAT 201509 2015-08-20 07:31:50 1 150.020000
4915 OAT 201509 2015-08-25 09:37:06 -1 149.010000
4963 OAT 201509 2015-08-25 13:08:30 -1 148.380000
4939 PALLAD 201509 2015-08-25 11:57:25 1 559.850000
5065 PALLAD 201509 2015-09-02 13:25:09 1 574.300000
5068 PALLAD 201512 2015-09-02 13:25:09 -1 574.900000
4879 PLAT 201510 2015-08-19 15:48:20 1 1008.400000
4942 PLAT 201510 2015-08-25 11:57:55 1 991.400000
4780 SMI 201509 2015-08-13 07:57:51 -1 9297.000000
4759 SOYBEAN 201511 2015-08-10 15:46:16 1 981.500000
4819 SOYBEAN 201511 2015-08-13 16:22:26 -1 913.500000
5002 SOYBEAN 201511 2015-08-26 12:21:49 -1 874.500000
4747 SP500 201509 2015-08-07 08:56:52 -1 2073.500000
4813 US10 201509 2015-08-13 15:33:12 1 127.671875
4966 US10 201509 2015-08-25 13:55:42 -1 128.656250
4969 US10 201512 2015-08-25 13:55:42 1 128.101562
4816 US2 201509 2015-08-13 15:38:40 1 109.476562
4972 US2 201509 2015-08-25 13:56:17 -3 109.687500
4975 US2 201512 2015-08-25 13:56:17 3 109.468750
4978 US5 201509 2015-08-25 13:57:13 -1 120.570312
4981 US5 201512 2015-08-25 13:57:13 1 120.179688
4984 US5 201512 2015-08-25 14:04:00 1 120.171875
5020 US5 201512 2015-08-27 14:07:59 -1 119.609375
4708 V2X 201509 2015-08-05 11:02:38 -1 19.500000
4711 V2X 201510 2015-08-05 11:02:38 1 19.900000
4738 V2X 201510 2015-08-06 08:11:06 -1 20.000000
4789 V2X 201510 2015-08-13 08:14:17 -1 20.700000
4933 V2X 201510 2015-08-25 10:38:30 1 23.850000
5047 V2X 201510 2015-08-31 09:08:46 1 26.000000
5137 V2X 201511 2015-09-04 08:22:25 1 27.300000
4714 VIX 201510 2015-08-05 11:04:33 -2 15.800000
4720 VIX 201509 2015-08-05 14:02:19 2 15.000000
4795 VIX 201510 2015-08-13 10:05:07 -1 16.000000
4924 VIX 201510 2015-08-25 10:02:31 1 20.550000
5062 VIX 201510 2015-08-31 11:00:07 1 22.800000
4861 WHEAT 201512 2015-08-18 12:02:26 -1 503.750000
5080 WHEAT 201512 2015-09-02 16:24:15 -1 477.000000Slippage – expected £648, actual £421
A lot of rolls, but also just a lot of trading because of the markets wild swings. Even profitable positions like MXP had to be traded to adjust my risk as the turmoil affected it.
In other news I've published my final blog post on system construction, any discussion here please http://www.elitetrader.com/et/index...ed-trading-system-checks-and-balances.294024/
And for those who don't follow me on social media, you might not have seen the first review of my new book:
"A remarkable look inside systematic trading never seen before, spanning the range from small to institutional traders. This isn't only for algorithmic traders, it's valuable for anyone needing a structure - which is all of us. Carver explains how to properly test, apply constant risk, size positions and portfolios, and my favorite, his "no rule" trading rule, all explained with scenarios. Reading this will benefit all traders."
Perry Kaufman, author of Trading Systems and Methods.
I'm a massive fan of Perry, and very grateful for this endorsement.
The official release of the book is September 15th, but some of the people who pre-ordered the ebook have received their copies already.
Happy trading
GAT
A quiet couple of weeks after all the excitement in August.
As of this morning my loss since the last update was around £7K. Current draw down is 8.7%.
Gainers:
Crude oil £6K
Platinum £3.2K
Palladium £2.2K
KOSPI £1.4K
Yen £1.2K
Losers:
KR3 £1.2K
Live cow £1K
S&P 500 1.8K
Current positions:
Current risk:
My risk has fallen with the weakening of many trends; the current expected portfolio risk as of last night was virtually unchanged @ 90% of average; on 91% of maximum capital or in money terms £5162 a day or £83K a year.
Trades:
A lot of rolls.
Slippage £349 vs expected £473
GAT
As of this morning my loss since the last update was around £7K. Current draw down is 8.7%.
Gainers:
Crude oil £6K
Platinum £3.2K
Palladium £2.2K
KOSPI £1.4K
Yen £1.2K
Losers:
KR3 £1.2K
Live cow £1K
S&P 500 1.8K
Current positions:
Code:
code contractid positions Lock WrongContract InFwdNotRoll
4 BOBL 201512 3 False False False
20 BTP 201512 2 False False False
14 COPPER 201512 -1 False False False
13 CORN 201612 -3 False False False
7 CRUDE_W 201512 -1 False False False
12 EDOLLAR 201903 3 False False False
17 EDOLLAR 201812 10 False False False
19 EUROSTX 201512 -13 False False False (hedge)
1 GAS_US 201512 -1 False False False
2 GAS_US 201511 -4 False False False
11 KR10 201512 2 False False False
9 KR3 201512 9 False False False
0 LEANHOG 201606 2 False False False
15 LIVECOW 201610 -2 False False False
6 MXP 201512 -5 False False False
10 NZD 201512 -2 False False False
18 PALLAD 201512 -1 False False False
8 PLAT 201510 -2 False False False
3 US2 201512 2 False False False
16 US5 201512 1 False False False
21 V2X 201511 3 False False False
5 WHEAT 201612 -2 False False FalseCurrent risk:
Code:
code multisignal expected_annual_risk expected_annual_risk_per_contract position expected_annual_risk_rounded_pos
4 WHEAT -12.8 8245 4182 -2 8363
28 MXP -14.3 9268 1854 -5 9268
0 CORN -11.8 7618 3105 -3 9315
2 LIVECOW -14.4 9323 5256 -2 10511
30 COPPER -23.1 14902 10651 -1 10651
29 NZD -16.2 10470 6157 -2 12314
32 PALLAD -12.9 8317 13331 -1 13331
33 PLAT -23.5 15199 6668 -2 13335
34 CRUDE_W -19.8 12799 14363 -1 14363
35 GAS_US -35.9 23175 4853 -5 24267
7 BOBL 9.9 6421 1859 3 5576
1 LEANHOG 10.6 6823 3006 2 6011
5 KR10 7.9 5119 3042 2 6084
6 KR3 9.2 5953 705 9 6349
8 BTP 20.9 13506 7393 2 14786
36 EDOLLAR 27.0 17431 1446 13 18799My risk has fallen with the weakening of many trends; the current expected portfolio risk as of last night was virtually unchanged @ 90% of average; on 91% of maximum capital or in money terms £5162 a day or £83K a year.
Trades:
Code:
code contractid filled_datetime filledtrade filledprice
5164 AUD 201509 2015-09-08 16:23:20 1 0.701700
5218 AUD 201509 2015-09-10 06:56:38 1 0.701800
5221 AUD 201512 2015-09-10 06:56:38 -1 0.698450
5248 AUD 201509 2015-09-10 07:06:08 1 0.701800
5251 AUD 201512 2015-09-10 07:06:08 -1 0.698450
5323 AUD 201512 2015-09-14 01:29:59 1 0.707700
5437 AUD 201512 2015-09-16 07:41:01 1 0.712900
5113 BTP 201512 2015-09-03 15:24:02 1 133.640000
5152 BTP 201512 2015-09-07 07:34:06 1 134.390000
5371 BTP 201512 2015-09-15 11:40:02 1 134.780000
5428 BTP 201512 2015-09-15 15:58:34 -1 134.140000
5470 COPPER 201512 2015-09-18 16:07:57 1 2.383500
5404 CORN 201512 2015-09-15 14:35:56 3 393.500000
5407 CORN 201612 2015-09-15 14:35:56 -3 414.000000
5146 EDOLLAR 201812 2015-09-04 12:18:57 1 97.880000
5278 EDOLLAR 201903 2015-09-10 12:00:18 1 97.760000
5377 EDOLLAR 201903 2015-09-15 12:04:39 1 97.820000
5386 EDOLLAR 201903 2015-09-15 12:20:43 1 97.810000
5434 EDOLLAR 201812 2015-09-15 18:35:47 -1 97.810000
5461 EDOLLAR 201812 2015-09-18 12:13:41 -1 97.970000
5365 EUROSTX 201509 2015-09-15 08:02:02 13 3188.000000
5368 EUROSTX 201512 2015-09-15 08:02:02 -13 3175.000000
5473 GAS_US 201512 2015-09-18 18:18:02 -1 2.842000
5161 GBP 201509 2015-09-08 13:18:02 1 1.540400
5224 GBP 201509 2015-09-10 07:01:47 1 1.536500
5227 GBP 201512 2015-09-10 07:01:47 -1 1.535700
5266 GBP 201509 2015-09-10 07:58:28 1 1.536500
5269 GBP 201512 2015-09-10 07:58:28 -1 1.535700
5329 GBP 201512 2015-09-14 02:38:06 1 1.544800
5452 GBP 201512 2015-09-18 01:52:52 1 1.556400
5149 GOLD 201512 2015-09-04 14:20:48 -1 1117.100000
5281 GOLD 201512 2015-09-10 12:05:39 -1 1106.300000
5458 GOLD 201512 2015-09-18 12:05:03 1 1136.500000
5476 GOLD 201512 2015-09-21 12:04:49 1 1137.100000
5131 KOSPI 201509 2015-09-04 02:36:39 1 229.200000
5290 KR10 201512 2015-09-14 01:11:27 2 124.670000
5305 KR10 201512 2015-09-14 01:12:08 2 124.670000
5311 KR10 201512 2015-09-14 01:14:19 2 124.680000
5314 KR10 201512 2015-09-14 01:16:48 2 124.650000
5317 KR10 201512 2015-09-14 01:18:12 2 124.640000
5320 KR10 201509 2015-09-14 01:19:54 -2 124.790000
5335 KR10 201512 2015-09-14 06:30:13 -8 124.420000
5128 KR3 201509 2015-09-04 01:46:33 1 109.610000
5158 KR3 201509 2015-09-08 04:31:46 -1 109.580000
5215 KR3 201509 2015-09-10 02:30:37 1 109.630000
5296 KR3 201512 2015-09-14 01:03:19 11 109.570000
5299 KR3 201512 2015-09-14 01:07:45 11 109.570000
5302 KR3 201512 2015-09-14 01:10:46 11 109.570000
5308 KR3 201509 2015-09-14 01:12:30 -11 109.620000
5338 KR3 201512 2015-09-14 06:30:52 -21 109.520000
5449 KR3 201512 2015-09-18 01:45:36 -2 109.620000
5455 KR3 201512 2015-09-18 04:49:37 -1 109.650000
5176 LEANHOG 201606 2015-09-09 15:00:11 1 80.150000
5116 LIVECOW 201510 2015-09-03 16:55:29 1 142.450000
5119 LIVECOW 201610 2015-09-03 16:55:29 -1 135.100000
5479 LIVECOW 201610 2015-09-21 16:14:37 -1 131.375000
5167 MXP 201509 2015-09-08 16:54:46 1 0.059580
5236 MXP 201509 2015-09-10 07:04:38 2 0.059370
5239 MXP 201512 2015-09-10 07:04:38 -2 0.058970
5254 MXP 201509 2015-09-10 07:10:37 2 0.059370
5257 MXP 201512 2015-09-10 07:10:37 -2 0.058970
5260 MXP 201509 2015-09-10 07:15:41 1 0.059370
5263 MXP 201512 2015-09-10 07:15:41 -1 0.058970
5230 NZD 201509 2015-09-10 07:02:48 1 0.639200
5233 NZD 201512 2015-09-10 07:02:48 -1 0.635050
5242 NZD 201509 2015-09-10 07:05:07 1 0.639200
5245 NZD 201512 2015-09-10 07:05:07 -1 0.635050
5347 OAT 201512 2015-09-14 09:39:00 1 150.670000
5431 OAT 201512 2015-09-15 16:01:47 -1 149.600000
5389 SOYBEAN 201511 2015-09-15 12:33:19 1 886.000000
5392 SOYBEAN 201611 2015-09-15 12:33:19 -1 879.000000
5395 SOYBEAN 201511 2015-09-15 12:38:34 1 885.750000
5398 SOYBEAN 201611 2015-09-15 12:38:34 -1 878.750000
5401 SOYBEAN 201611 2015-09-15 12:48:31 1 878.500000
5464 SOYBEAN 201611 2015-09-18 12:30:07 1 882.000000
5446 US10 201512 2015-09-17 14:11:42 -1 126.640625
5467 US2 201512 2015-09-18 15:54:06 -1 109.382812
5137 V2X 201511 2015-09-04 08:22:25 1 27.300000
5275 V2X 201511 2015-09-10 08:12:02 1 27.500000
5341 V2X 201510 2015-09-14 08:35:17 -1 27.800000
5344 V2X 201511 2015-09-14 08:35:17 1 26.750000
5362 VIX 201510 2015-09-15 06:47:25 1 22.850000
5410 WHEAT 201512 2015-09-15 15:28:37 1 494.750000
5413 WHEAT 201612 2015-09-15 15:28:37 -1 533.000000
5416 WHEAT 201512 2015-09-15 15:43:22 1 495.500000
5419 WHEAT 201612 2015-09-15 15:43:22 -1 533.500000
5422 WHEAT 201512 2015-09-15 15:46:48 1 496.000000
5425 WHEAT 201612 2015-09-15 15:46:48 -1 534.000000
5440 WHEAT 201612 2015-09-16 15:59:10 1 526.250000
5443 WHEAT 201612 2015-09-16 16:04:28 -1 526.000000A lot of rolls.
Slippage £349 vs expected £473
GAT
GAT,
Great post and great blog.
I have a few questions as I'm digesting all the info:
1. the 25% capital at risk is at 95% confidence (I believe I read that in one of your posts), and therefore 2 times the standard deviation of annual returns, correct? If I'm following correctly, then this means that the standard deviation of annual returns is 12.5% -- is this based on your backtesting results?
2. if my understanding in #1 above is correct, why do you not include the expected annual return in the equation of capital at risk? What I mean is, why is not capital at risk equal to expected maximum draw-down at 95% confidence -- in other words -- maximum_draw_down = mean_annual_return - 2 * annual_std_dev ?
3. I'm amazed by your graph (in prior post) showing the expected and actual risk. You mentioned that your "risk" calculations are based on standard deviation of the daily returns. I assume that the calculation of expected and actual risk is a measure of standard deviation of returns based on some look-back period, is that correct? If so, what is the duration of the look-back that you use?
BTW, I'm new to futures trading, actually never traded a single futures contract yet
I found your blog when I was looking for some info on the IB Gateway API -- it seemed amazing to me that there was no way to log in without the GUI -- I guess there isn't. I'm working on a futures trading system now, and I've been trading a tactical asset allocation system (using ETFs) for about 6 months. But that system is automated only up to the point of execution, it sends me emails of what to buy/sell and I do that part manually. The average holding period is about a month, so the burden is not big.
Anyhow, I hope my questions make sense -- meaning that I've not misunderstood or misread your blog and this thread. Please keep in mind that I'm not questioning your methods in #2, I simply would like to understand why this is your approach and perhaps why the approach I mention in the question is not appropriate.
Now that I think about it, I am probably missing something. You mentioned that your Sharpe (or information ration since it does not include the risk-free rate) is roughly 0.9. If std. dev. of annual returns is 12.5%, then average annual return is roughly 11%. But I remember reading that you expect your system to produce about 16% annually. Sorry, I probably need to go back through some of your posts.
Thanks for your time and for making all this information available.
--Maciej
Great post and great blog.
I have a few questions as I'm digesting all the info:
1. the 25% capital at risk is at 95% confidence (I believe I read that in one of your posts), and therefore 2 times the standard deviation of annual returns, correct? If I'm following correctly, then this means that the standard deviation of annual returns is 12.5% -- is this based on your backtesting results?
2. if my understanding in #1 above is correct, why do you not include the expected annual return in the equation of capital at risk? What I mean is, why is not capital at risk equal to expected maximum draw-down at 95% confidence -- in other words -- maximum_draw_down = mean_annual_return - 2 * annual_std_dev ?
3. I'm amazed by your graph (in prior post) showing the expected and actual risk. You mentioned that your "risk" calculations are based on standard deviation of the daily returns. I assume that the calculation of expected and actual risk is a measure of standard deviation of returns based on some look-back period, is that correct? If so, what is the duration of the look-back that you use?
BTW, I'm new to futures trading, actually never traded a single futures contract yet
I found your blog when I was looking for some info on the IB Gateway API -- it seemed amazing to me that there was no way to log in without the GUI -- I guess there isn't. I'm working on a futures trading system now, and I've been trading a tactical asset allocation system (using ETFs) for about 6 months. But that system is automated only up to the point of execution, it sends me emails of what to buy/sell and I do that part manually. The average holding period is about a month, so the burden is not big. Anyhow, I hope my questions make sense -- meaning that I've not misunderstood or misread your blog and this thread. Please keep in mind that I'm not questioning your methods in #2, I simply would like to understand why this is your approach and perhaps why the approach I mention in the question is not appropriate.
Now that I think about it, I am probably missing something. You mentioned that your Sharpe (or information ration since it does not include the risk-free rate) is roughly 0.9. If std. dev. of annual returns is 12.5%, then average annual return is roughly 11%. But I remember reading that you expect your system to produce about 16% annually. Sorry, I probably need to go back through some of your posts.
Thanks for your time and for making all this information available.
--Maciej
It's absolutely fine to ask these questions, and hopefully I can clear up your confusion.
The long run target expected average standard deviation of my annual returns is 25% - that is what I mean by 'risk'. The amount of capital I have at risk is £400,000; so the long run expected blah blah is £100K a year.
I should get slightly less than this in my backtest, as I have an exogenous risk overlay that reduces my risk when it is particularly vunerable to tail events. However over short periods of my backtest I will obviously have expected risk different from the target; eithier because my forecasts are weak or strong, or because correlations are different from their long run average.
I'm trying to remember where I used a 95% confidence interval; perhaps relating to a t-test around what the distribution of my returns should be, but I don't think I've ever used it in the context of risk. If so apologies. More likely I've mentioned VAR in passing - but to be clear I don't use VAR and in any case VAR would normally be quoted as a daily measure.
The maximum drawdown at 95% confidence is path dependent and so can't just be calculated from the expected annual standard deviation. I think your equation would give the 95% confidence interval for expected annual loss if targeted risk was constant (i.e. it wouldn't make sense for trend following or for someone using the kelly criteria to scale capital at risk).
Still for fun, if we assume my SR is 0.9 as in the backtest, then we get 0.9 x 25% - 2 * 25% or perhaps if we factor out the expected standard deviation 25% x (0.9 - 2) = 28%. That is probably a little conservative but worth bearing in mind.
FWIW the DAILY equivalent of that formula makes a lot more sense; daily 1.5625% x (0.05625 - 2); or about 3%. Note this is the same as a 95% VAR if you assume gaussian returns. If you put a gun to my head then I'd say my 95% VAR was probably about 4% on average.
Due to jump effects and so on I'd say that 3% isn't sufficiently conservative and the 95% daily loss is probably a little worse than that in reality.
I use a lookback of about one month for measuring standard deviations, and a longer lookback of about 6 month for correlations (and using weekly returns).
Yes to reiterate my std. dev of annual returns is 25% (or a little lower in backtest), as I'm a little conservative I don't expect to see a SR of 0.9, so 16% is probably a reasonable figure for expectations.
GAT