I have similar implementation. Now I have noticed that quarterly micro DAX contract has new name (i.e. FDXS 20230915 M instead of FDXS SEP 23). At first it was presented by EUREX that monthly contract names stay the same, and only weekly contracts would get W1, W2, W3, W4 at the end.
Maybe it's the recent market turbulence but I'm finding my DO system opening and closing the same positions this week. For example: 3/13 DAX [-1] close MXP [-1] close NIKKEI [1] RUSSELL [1] close TOPIX [-1] close 3/14 DAX [1] DOW [-1] GBP_micro [-1] RUSSELL [-1] US5 [1] 3/15 (on instrument order stack) RUSSELL [1] NIKKEI [1] DAX [-1] DOW [1] Shadow cost and tracking error buffer at default values. Capital ~100k. I realize this is probably expected behavior for DO given the large price changes in the past few days but I'm seeking reassurance that there isn't some weirdness in my config somewhere. Thanks!
I would imagine these are all pretty cheap instruments, so it wouldnt' be a surprise if it was tinkering at the margin with them. I saw similar behaviour. Down about 6% for the whole of 'volmageddon' since Wednesday; AHL down about 11%. Rob
Great TTU episode again, GAT. The whole volmageddon thing (down about 8% from highs this year) has had me scrambling back to my risk overlays to figure out if I doing them right. Going momentarily deep into the weeds, the correlation matrix I use to calculate the daily portfolio risk is essentially the instrument correlation matrix adjusted down by the subsystem factor of .7 (or more accurately sqrt(1/2)) as in ST. But as is often said in TTU, trend following may appear simple, but the devil is in the details. Arguing with devil (not aloud) I am proposing the correct correlation matrix might be the unadjusted one as we are looking at static risk on a single day not the dynamic subsystem. So who is right?
For current portfolio risk you should use the correlation of instrument returns and your current positions. The correlation of subsystem returns (actual, or using the 0.7 approximation) is what you use for instrument weight and IDM calculation Rob
Thanks for the response Rob. I justified the use of subsystem correlations because the alternative to multiplying the intrument risk vector by the corrleation matrix was simply looking at the volatility of portfolio returns (over the same lookback period you would use to generate the correlation matrix) and seeing if this was too high and then making appropriate adjustments. Since portfolio returns came from individual subsystems not instruments as described in your excellent book, I concluded the adjusted subsystem correlations were the right one to use when measuring portfolio risk. Still, I guess at the end of the day all of this is becomes moot if we simply making the appropriate adustment on the cap level depending on which measure is used. But wanted to play devil's advocate to ensure I haven't missed something.....
I see the flaw in my thinking now. A risk snapshot is based on CURRENT posns. Historic risk is not. Please ignore this last comment, nothing to see here.
New to your thread. Do you trade fully automatically? What type of software do you use? Do you also trade stocks?