@globalarbtrader Have spent the past two weeks trying to replicate all of the signals from AFTS. Am struggling with a few of the signals and was hoping for your insight. Carry: I am computing carry based on the front two contracts; and my numberes check out ( at least directionally with the prices on the contract I am seeing live). I am thinking that maybe you are using different contracts (besides the front two) for some of your instruments. Can you clarify if this is the case and/or what else may be the issue? Skew: My numbers are way off from yours. I am using the back adjusted front contract for this. The raw skew calculation I am getting seems to be accurate; but again the # I am seeing is way off from what I see posted on your AFTS website. Again I wonder if my issue is using the front contract and not a different contract; or if there is some sort of adjustment you have made to the prices before computing skew (or something else). Anyway any feedback you can provide would be appreciate. My trend signals (except asset trend) all match up very very closley with yours so that is reassuring. I am bit hestiate to put any rules lives that I am not extremely confident with the coding on.
https://github.com/robcarver17/pysystemtrade/blob/develop/data/futures/csvconfig/rollconfig.csv shows how I roll, and thus compute carry. RollOffsetDays shows how many days before expiry I roll. So for example, for VIX I aim to roll 60 days before expiry. So I am usually holding the second contract. CarryOffset shows which contract I calculate carry on. -1 for VIX means its the one before the priced contract. So I'm calculating carry on 1st vs 2nd. Off the top of my head the only contracts where I'm likely to be doing something different are those i hold as seasonal. That's where there is only one letter in the HoldRollCycle column. Eg Corn, I always hold Z, and I calculate carry -1 which is U. So it's U-Z always. When you say "from what I see posted on your AFTS website.", what do you mean exactly? which numbers, which page? There are two explanations, different data or different code. Which code are you using? Skew: is very sensitive to outliers, so this will show the biggest difference. Rob
Mexican Peso historical data and charts still not working. IB really messed this one up. I'm surprised enough people haven't complained yet that they've fixed it.
I just submitted a web ticket. Should have done it a week ago. Perhaps they'll fix it if enough people complain.
@globalarbtrader Hi Rob, in AFTS Strat 9, to calculate FDM, you are using the correlation matrix derived using capped forecasts, right?
Hi, That was a helpful post thanks. When you say "rates", you mean the STIRs and the 10Y/2YY (10Y/2Y yield) not the bonds/notes? If you've looked at bonds as well, what's your opinion on the CSI cash prices for these? For 2Y Note June contract, priced at 103.22, the cash price is 99.70. We can't use these in the same way as others or it implies way higher carry than there should be. I really don't know bonds outside of just following time series trends. Does anyone know what these prices represent? It can't be the rate based (100 - interest rate) that the usual rate contracts use, because the CSI cash price does fluctuate over 100 a lot for these as well. It actually behaves like the bond prices do, but much lower. It could be including duration and expectations at settlement time etc, I just don't know the source of these quotes. Some examples for June contract: 2Y Futures: 103.22 / cash: 99.70 10Y Futures: 109.94 / cash: 97.97 TBond Futures: 112.66 / cash: 96.63 Probably going to use term structure for these, but maybe if I can learn something from these. This is the 2Y Cash (only close prices): This is the 10Y cash (only close prices):
Hi, guys. I have a question about carry forecasts. I downloaded the prices for the Corn future that trades at the MATIF (EMA in IB). According to the rollconfig.csv in pysystemtrade, for carry calculations I should use the nearer contract (CarryOffset = -1). Right now I'd be holding the Aug25 contract and I should use the Jun25 contract price for carry. But the Jun25 contract expired last week (on 4th June), so now the carry forecast cannot be calculated anymore and the overall forecast "jumps". I see some ways to fix this: - Switch to the later contract (Nov25) on these gaps where there is no nearer contract (so using a CarryOffset = +1 for some periods in the year) - Use the value of the last correct carry forecast (on 4th June) from that day on until I roll to the Nov25 contract and have a nearer contract available again - Assume the carry forecast is 0 during the gap - Remove the carry rule from the overall forecast (so its weight is zero and during the gap I only care about the rest of the rules – trend in my case) How should I treat this situation? Thanks a lot.