R2D2 is like a quant fund "Can't understand a word he's saying. Must be smart. Give him some money." GAT
After trading paper money for more than a year while saving some real play money I finally turned on the live system on Friday Base capital 100k USD, 14 instruments, most of them with thresholding enabled (selected based mostly by the smallest margin requirements but also tried to cover as many asset classes as I could), volume knob is at 25. weights AUD 0.0724101666666 GBM 0.1238042666666 HG 0.0783604666666 ZC 0.0719512666666 CL 0.0621983666666 GE 0.0549407 3KTB 0.0462294666666 LE 0.0737864666666 MXP 0.0912875666666 NQ 0.0774701666666 PL 0.0889830666666 ZS 0.0579291666666 ZT 0.0465405666666 V2TX 0.0541083 Signal weights are bootstrapped with PySystemstrade. positions: ZT 2020-03:-1 GE 2023-06:1 AUD 2020-03:-1 MXP 2020-03:6 V2TX 2020-04:-5 - it wanted to sell more V2TX but couldn't fill on time before EOD. So far I'm up 133$
That's a good start @Kernfusion. In case you aren't aware: there are some more contracts available with small contract size and thus small value volatility per contract: M6E (EUR/USD forex), MES ("baby-ES", also smaller than NQ), EU600 (Europe equity STOXX600). These small contracts could help to increase diversification. Have you considered using NG (natural gas) instead of CL? It currently has a lower value volatility.
Thanks @HobbyTrading, NG instead of CL does seem like a good idea, I'm running them both in paper and can see that at maximum forecast NG had position of 4 contracts but CL only 2, not sure why I choose CL maybe at the time I was looking at them they had more similar margins... Has it always been like that historically (CL was more capital-demanding than NG )? (I just downloaded historical margins from here and it does seem to be the case https://www.cmegroup.com/clearing/risk-management/historical-margins.html) I'll probably switch CL to NG.. Regarding the other mini-contracts, do they have enough liquidity to be able to comfortably roll them and to not loose too much on the spread? Also, I'd need to get historical data for them somewhere, rerun backtests, prepopulate all the roll periods for the new contracts (I have automatic rolls which rely of prepopulated roll-schedule for each contract for the next 80-100 years) - quite a lot of work.. I just want to finally sit, relax and watch the thing do everything by itself I was planning to add more capital next year and these instruments ['EUROSTX', 'WHEAT', 'JPY','VIX', 'LEANHOG' ,'KOSPI','BTP','KR10'], which should give me pretty-wide coverage of different things, and then I'll continue to add more capital and migrate more instruments from paper-system (where I have 31 instrument @350kUSD) to the prod one..
Let's assume that you are not facing any scalability issues (and thus not forced to change any parameters due to liquidity or risk constraints, which for AHL was not always true, I'd recon?). If your primary objective is to preserve the smoothness of your PnL profile, you can actually optimize a scaling schedule based on your Sharpe. I recall deriving something along these lines a few years ago, but my weed-infused brain is blanking on the exact formulation right now.
Darth Vader would be the recipient of my funds. He is a typical activist shareholder and, as far as I can remember, he got the results he wanted at the board meetings.
Although both are considered an "energy instrument" is it my impression that they behave rather differently. Of course is CL more often influenced by global politics, and macro-economics, than NG. See here a chart of both. The blue line (left hand scale) is the forecast value. The red line (right hand scale) is the value volatility per contract in USD. During the last 3 years had CL a value volatility between about 350 and 2k USD per contract. NG is currently relatively low, at about 300 USD, and can fluctuate to about 800~1k USD per contract. So there are times where CL is smaller than NG, but most often (at least during the last 3 years) had NG a smaller value volatility than CL. The forecasts also show a different behavior during the last 3 years. Eyeballing it does not show a high correlation between these two.