 |
| 2015.02.24 XOI.X Cycle Model Chart |
The XOI.X index is a bit ahead of itself but on track for a rise into April. There are no changes to the long term model. A previous iteration of the model is shown here on the blog and below.
 |
| 2014.10.17 XOI.X Cycle Model Chart |
5 comments:
Thanks Paolo - this is really interesting, as I've noticed something that could be useful. I don't trade XOI or oil as a proxy of sorts, but have just considered the correlation of XOI to SPX. The correlation waxes and wanes in a cyclic fashion, with overperformance into mid-2014 and underperformance thereafter. The correlation nonetheless exists.
Your XOI projection suggests a mid-April peak which matches well with some other traditional analyses (EW, trendlines, sentiment indicators) along with macro expectations (Grexit v2, Fed rates) and the "general expectation" of a correction following some of blow-off top.
Do your neural models hold any insight?
Mr. Anon
Well, here is the output of a neural model for SP which is a run on an out-of-sample set. The optimization set was two years prior to 7/8/2014. Overall performance was about 68%. Here are the details.
Name All Trades Long Trades Short Trades
Start Date/Time 7/8/2014
End Date/Time 2/20/2015
Total Number of Bars 158
Beginning Price 1963.710353
Ending Price 2110.299163
Change in Price 146.5888
Percent Change in Price ("Buy And Hold") 7.46%
Annual Percent Change in Price 11.95%
Return on Trades 18.26% 12.83% 5.43%
Annual Return on Trades 29.23% 20.53% 8.70%
Total Number of Trades 16 8 8
Number of Winning Trades 11 7 4
Number of Losing Trades 5 1 4
Percent Profitable Trades 68.75% 87.50% 50.00%
Net Profit/Loss 35573.25 24660.12 10913.13
Largest Winning Trade Profit 15528.72 15528.72 7749.8
Largest Losing Trade Loss 7664.228 7664.228 1680.189
Profit/Loss per Trade 2223.328 3082.514 1364.141
Profit/Loss per Bar 225.1471 156.0767 69.07045
Commissions Paid 3.2 1.6 1.6
Sorry about the formatting but here is a breakdown of the trades. Obviously risk management by stops where if a stop is reached, the action is to hold until the next signal helps increase performance. BTW, in neural models, this is about as good as it gets!
Signal Signal Date Fill Date Fill Price Profit/Loss %Return Cum Profit/Loss Cum %Return
Sell Short 7/8/2014 7/9/2014 1972.829331
Reverse to Long 7/10/2014 7/11/2014 1967.57087 525.6461 0.27% 525.6461 0.27%
Reverse to Short 9/2/2014 9/3/2014 2000.71921 3,314.634 1.68% 3,840.28 1.95%
Reverse to Long 9/16/2014 9/17/2014 2001.569697 (85.2487) -0.04% 3,755.031 1.91%
Reverse to Short 9/18/2014 9/19/2014 2010.400028 882.8331 0.44% 4,637.864 2.35%
Reverse to Long 9/22/2014 9/23/2014 1982.769731 2,762.8297 1.37% 7,400.694 3.72%
Reverse to Short 10/9/2014 10/10/2014 1906.129452 (7,664.2279) -3.87% (263.5337) -0.14%
Reverse to Long 10/15/2014 10/16/2014 1862.760861 4,336.6591 2.28% 4,073.125 2.13%
Reverse to Short 10/30/2014 10/31/2014 2018.050045 15,528.7184 8.34% 19,601.84 10.47%
Reverse to Long 11/5/2014 11/6/2014 2031.210094 (1,316.2049) -0.65% 18,285.64 9.82%
Reverse to Short 11/6/2014 11/7/2014 2031.91911 70.7016 0.03% 18,356.34 9.85%
Reverse to Long 11/18/2014 11/19/2014 2048.719003 (1,680.1893) -0.83% 16,676.15 9.03%
Reverse to Short 11/19/2014 11/20/2014 2052.750746 402.9743 0.20% 17,079.13 9.22%
Reverse to Long 12/1/2014 12/2/2014 2066.550353 (1,380.1607) -0.67% 15,698.96 8.55%
Reverse to Short 12/24/2014 12/26/2014 2088.770721 2,221.8368 1.08% 17,920.80 9.63%
Reverse to Long 1/13/2015 1/14/2015 2011.270719 7,749.8002 3.71% 25,670.60 13.34%
Close Long Position 2/20/2015 2/20/2015 2110.299163 9,902.6444 4.92% 35,573.25 18.26%
Also, the run closes the position but the SP model is still long. In contradistinction, the VIX model, which has an overall accuracy of 61%, more typical, is still on "Sell" at this moment. However, the model output has been rising suggesting a possible change in status in the next few days.
Wow - thanks for the detailed breakdown! An annual RoI of 30% with 2/3 profitable trades and 2:1 ratio on average profit/loss per trade is outstanding.
Awesome stuff indeed,
Mr. Anon
Post a Comment