Fully automated futures trading
- Thread starter globalarbtrader
- Start date
I know it hurts, but a very sharp reversal is often the best thing that can happen to a trading system like this. The vol scaling will cut your positions quickly, reducing further losses. The alternative, a slow reversal, means you have to wait for forecasts to reduce your position and by then you will have lost almost all your gains.
Of course that assumes we are the end of the trend... if not then at least you are in a position to benefit from . Important thing is risk is reduced, and rightly, since we are less certain about what is happening next.
Rob
Thanks,
Im just a casual follower of this thread, i typically read the first few pages of this thread every now and then and then the last few pages. Tend to ignore everything in between
I noticed in the first few posts on this thread, eg. post number 14, it seemed like you said you started out with 300K (either actual or notional) back in 21014. And you made approx 350K profit in the first year, between April 2014 and March 2014.
I might be wrong, but your latest daily % figures, which include that period, don't seem to show 100%+ returns, more like 50% for that period.
So are you using a different notional starting figure these days.
This is actually (sort of) explained in the second post of the thread: https://www.elitetrader.com/et/threads/fully-automated-futures-trading.289589/#post-4085237
What happened was when I started trading I had a 50% risk target, and then I dropped to 25% in December 2014 because it was way too scary.
So for proper comparision when my code calculates the notional capital it automatically resizes historic figures so that my % returns are consistent; effectively it doubles my capital base pre December 2014 so it's initially 600K rather than the 300K notional I actually had.
(Incidentally my actual account starting value was about 550K, but that isn't relevant; although it's how fundseeder and IB calculate my performance FWIW so again there are some differences in those early years)
To work out my percentage returns I divide the actual £££ p&l by the notional capital on any given day. So I end up 50% for that earlier period rather than 100%, because the capital is doubled. I did actually make 100% on the notional capital I was employing, and when looking at cash profits that is reflected, but I'd rather factor out the change in risk target.
(A lesson here is don't change your risk target, it makes life very hard!)
The pair of graphs I posted in the very first post shows this effect happening, which is why I felt the need to explain it in the second post. So this was a retrospective change, but once I've applied since before this thread started.
Just to make things even more complicated, prior to early December 2014, I was also compounding up my profits. So my max capital was actually as high as £420K, or notionally doubled £840K,
From January 2015 it's simple because I'm using 25% risk on a notional capital of £400k, which is what I use today, so no further adjustments are required. In the very first post I say my risk is "25% annualised on the notional capital of £400,000)." So I'd already settled on this methodology by the time the thread started.
I've said several times before that the £££ profit I have is basically higher than it would be if I had the same risk target. If you just add up all my % returns (adding up makes sense as I don't cumulate returns) you get about 144% to yesterdays close, or £576K on £400k. My actual accumlated profit to yesterdays close was £757K, about £180K less. That extra £180K was made by a combination of accumulating profits and having double the risk target in my first 9 months or so of trading.
On the other hand, if I'd compounded my returns I would have made 333%, which even on my original notional £300k would have been a nice round million quid. On the other, other hand Fundseeder and IB that assume I cumulate and use my current account value as the capital divisor reckon I've made over 400%.
Hope that helps.
Rob
Thanks for the taking the time to explain.
I copied your daily results into a nice monthly table, like hedge funds show monthly performance. So i can personally follow your results more easily. You might have posted something similar somewhere on your various blogs yourself, i didn't take a hard look. I have only seen you post that daily results spreadsheet. (I hope i didn't make any errors when creating the table from the daily % figures)
Your trading is very inspirational and motivational, in that you are able to trade such a hard system so successfully through long drawdowns!
Last edited:
That's awesome. Fundseeder have something similar, but without the colours. If I can trick something similar up in python, I'll see if I can get an auto updating table like that.
Rob
I know you like to roll your own, but this is pretty nice https://github.com/quantopian/pyfol...4b7d83502e5c6f2e81aa/pyfolio/plotting.py#L143
Code:
sns.heatmap(
monthly_ret_table.fillna(0) *
100.0,
annot=True,
annot_kws={"size": 9},
alpha=1.0,
center=0.0,
cbar=False,
cmap=matplotlib.cm.RdYlGn,
ax=ax, **kwargs)
ax.set_ylabel('Year')
ax.set_xlabel('Month')
ax.set_title("Monthly returns (%)")
return axSmall up today. As you'd expect after a big day, absolute s***load of trades, mostly closing.
Interestingly I made £2k from not doing these trades at the close yesterday, due to the mini rebounds (most of that was in Brent). Bid/ask spread was £260, but I made £111 back on the execution
Here's my current risk - v.low:
Of course this means if the upward march in energies resumes I'm now less likely to benefit; but that's life
Interesting that Crude is more correlated with Corn than with European oil equities.
Rob
Code:
================================================================================================
Broker orders
================================================================================================
instrument_code strategy_name contract_date fill_datetime fill filled_price
order_id
32033 BRE dynamic_TF_carry [20220500] 2022-03-10 15:02:43 [1] 0.19525
32030 BRENT-LAST dynamic_TF_carry [20220600] 2022-03-10 14:00:35 [-1] 112.97000
32028 EU-OIL dynamic_TF_carry [20220600] 2022-03-10 08:00:46 [1] 306.90000
32027 FTSECHINAA dynamic_TF_carry [20220300] 2022-03-10 03:00:34 [-1] 13877.00000
32019 IRON dynamic_TF_carry [20220500] 2022-03-10 01:25:37 [1] 159.85000
32025 MSCISING dynamic_TF_carry [20220300] 2022-03-10 02:35:19 [-1] 309.65000
32032 MXP dynamic_TF_carry [20220600] 2022-03-10 15:01:45 [-1] 0.04691
32031 RUSSELL dynamic_TF_carry [20220300] 2022-03-10 15:00:37 [1] 1999.10000
32034 SOYOIL dynamic_TF_carry [20220700] 2022-03-10 15:30:36 [-1] 70.90000
32020 TOPIX dynamic_TF_carry [20220600] 2022-03-10 01:45:32 [1] 1801.25000
32029 V2X dynamic_TF_carry [20220600] 2022-03-10 08:01:41 [1] 30.80000Interestingly I made £2k from not doing these trades at the close yesterday, due to the mini rebounds (most of that was in Brent). Bid/ask spread was £260, but I made £111 back on the execution
Here's my current risk - v.low:
Code:
********************************************************************************
Risk report report produced on 2022-03-10 16:01:19.104883
********************************************************************************
Total risk across all strategies, annualised percentage 11.0
========================================
Risk per strategy, annualised percentage
========================================
risk
dynamic_TF_carry 11.1
===================================================================================================================================================================================================================================================================
Instrument risk
===================================================================================================================================================================================================================================================================
daily_price_stdev annual_price_stdev price daily_perc_stdev annual_perc_stdev point_size_base contract_exposure daily_risk_per_contract annual_risk_per_contract position capital exposure_held_perc_capital annual_risk_perc_capital
JPY 0.00 0.00 0.01 0.34 5.37 9520364.06 82346.39 276.15 4418.38 -1.0 378567.3 -21.75 -1.17
MSCISING 4.83 77.21 307.65 1.57 25.10 56.07 17249.99 270.56 4328.93 -1.0 378567.3 -4.56 -1.14
KR10 0.36 5.84 119.74 0.30 4.87 619.20 74143.54 225.82 3613.16 -1.0 378567.3 -19.59 -0.95
FTSECHINAA 191.31 3060.91 13708.00 1.40 22.33 0.76 10440.41 145.70 2331.28 -1.0 378567.3 -2.76 -0.62
NIKKEI400 211.98 3391.61 16275.00 1.30 20.84 0.66 10689.15 139.22 2227.55 -1.0 378567.3 -2.82 -0.59
BRE 0.00 0.03 0.20 1.04 16.62 76162.91 15034.56 156.18 2498.86 1.0 378567.3 3.97 0.66
EU-OIL 6.50 104.07 303.80 2.14 34.26 41.96 12747.52 272.92 4366.68 1.0 378567.3 3.37 1.15
CORN 9.50 152.03 634.75 1.50 23.95 38.08 24172.20 361.85 5789.60 1.0 378567.3 6.39 1.53
SILVER 0.50 7.92 25.96 1.91 30.52 761.63 19771.89 377.15 6034.46 1.0 378567.3 5.22 1.59
GOLD_micro 25.86 413.73 1985.40 1.30 20.84 7.62 15121.38 196.94 3151.07 3.0 378567.3 11.98 2.50
SOYBEAN 15.94 254.99 1474.25 1.08 17.30 38.08 56141.59 606.89 9710.30 1.0 378567.3 14.83 2.57
IRON 4.63 74.11 159.85 2.90 46.36 76.16 12174.64 352.79 5644.63 2.0 378567.3 6.43 2.98
CRUDE_W_mini 3.97 63.58 100.92 3.94 63.00 380.81 38433.71 1513.30 24212.74 1.0 378567.3 10.15 6.40
Sum of abs(notional exposure % capital) 113.8
Sum of abs(annualised risk % capital) 23.8
Net sum of annualised risk % capital 14.9Of course this means if the upward march in energies resumes I'm now less likely to benefit; but that's life
Code:
==============================================================================================================================
Correlations
==============================================================================================================================
BRE CORN CRUDE_W_mini EU-OIL FTSECHINAA GOLD_micro IRON JPY KR10 MSCISING NIKKEI400 SILVER SOYBEAN
BRE 1.00 0.16 0.18 0.09 0.08 0.18 0.10 0.08 0.00 0.05 0.14 0.16 0.13
CORN 0.16 1.00 0.20 0.05 0.03 0.15 0.16 0.12 -0.08 0.02 0.03 0.13 0.41
CRUDE_W_mini 0.18 0.20 1.00 0.17 0.07 0.13 0.24 0.02 -0.01 -0.07 0.05 0.15 0.17
EU-OIL 0.09 0.05 0.17 1.00 0.13 -0.06 0.10 -0.07 -0.07 0.31 0.26 -0.01 0.08
FTSECHINAA 0.08 0.03 0.07 0.13 1.00 0.10 0.08 0.05 -0.02 0.18 0.21 0.11 0.12
GOLD_micro 0.18 0.15 0.13 -0.06 0.10 1.00 0.09 0.29 0.08 -0.05 0.04 0.47 0.13
IRON 0.10 0.16 0.24 0.10 0.08 0.09 1.00 0.01 -0.03 -0.09 -0.08 0.13 0.19
JPY 0.08 0.12 0.02 -0.07 0.05 0.29 0.01 1.00 0.14 -0.02 0.02 0.24 0.06
KR10 0.00 -0.08 -0.01 -0.07 -0.02 0.08 -0.03 0.14 1.00 -0.01 0.03 0.06 -0.16
MSCISING 0.05 0.02 -0.07 0.31 0.18 -0.05 -0.09 -0.02 -0.01 1.00 0.34 -0.01 0.08
NIKKEI400 0.14 0.03 0.05 0.26 0.21 0.04 -0.08 0.02 0.03 0.34 1.00 0.05 0.07
SILVER 0.16 0.13 0.15 -0.01 0.11 0.47 0.13 0.24 0.06 -0.01 0.05 1.00 0.10
SOYBEAN 0.13 0.41 0.17 0.08 0.12 0.13 0.19 0.06 -0.16 0.08 0.07 0.10 1.00Interesting that Crude is more correlated with Corn than with European oil equities.
Rob
Last edited:
The timing on this closing trade turned out to be spot on, close to the top of the day
(I'm assuming times are UTC).How did it execute at that time? I know at some point your execution algo was checking that the bid-ask spread is no bigger than X multiples of the minimum tick size. Does your execution try every X minutes and then finally it succeeds when there's enough liquidity that the spread is small enough, or something else?
From the backtest
https://1.bp.blogspot.com/-8PPER8WM...0wqqDowT3d9Mu8xgCLcBGAsYHQ/s1920/Figure_1.png
Average (once I have all the markets in my dataset) is probably about 25, max 45
13 is really low; but then I have 40% of my average risk on so that makes sense
Rob