Apparently you try the following: - you buy 100% of your cash in T-Bills/short term US Tsy - the securities segment will go negative (as it funds the futures segment margin) - that negative balance, because it is collateralized by T-Bills, won't generate an interest charge. It is not clear to me what happens when you have a negative day, the variation margin is paid and you end up with a securities segment cash balance that is more negative than the initial margin of the futures segment.
Can't be that bad ....I guess you just pay some interest in the negative cash balance at base+1.5% using the leverage that the T bill position will give you. One just needs then to sell the t-bill to cover that and avoid paying any interest to Tommy.
A question for CSI users: They also have cash prices for most of their futures. Currently I do not use their cash prices because I have another provider which does not have cash prices and I wanted to be consistent. I guess using cash prices opens another can of worms due to possible misalignment of prices due to different closing times. It seems to me that some of the cash prices are also not tradable because it is only an index and so on. Has anybody here used cash prices for carry calculations? Is this an endeavour worth undertaking or a waste of time?
I have a data query for the hive mind. For my next book I'm going to be running some simple backtests of stocks. Something like: - every year for the last N years (ideally N>20), get the S&P 500 constiutents - get ideally a history of prices adjusted for splits, dividends and other corporate actions. If it doesn't have dividends this isn't a deal break. There would also be some work with ETFs where I would supply the tickers but again get price history. From what I've heard, CSI is a good go to place for this. Can anyone confirm that? Are there alternatives? And does anyone have some python that would save me time to get this data? Many thanks Rob
Not sure how legit he is, but Marsten Parker is a well-known systematic equities trader (he was in the newest Market Wizards book) and he uses Norgate for his systems.
CSI have the following data options available to handle stock adjustments … Adjust for stock splits - Adjusts the data prior to the split using an appropriate multiplier to force a smooth transition from pre-split data to post-split data. For example, in a 2 for 1 split, which would normally result in the stock price falling by 50%, the pre-split data would be multiplied by .5. Adjust for stock dividends and capital gains - Much like the split adjustment, an appropriate multiplier is used to compensate for the value removed from the stock price by the dividend or distribution. Due to the effects of inflation, this method can cause adjusted stock prices to fall below zero. Negative values are not accepted by all trading systems. Adjust Dividend/Capital Gain proportionally (Not right P/L, but no negatives) - This adjustment method also smoothes the transition from pre-distribution prices to post-distribution prices, but the adjustment is based on the dividend or capital gain as a percentage of price, not on the dollar value actually paid to shareholders. This technique avoids negative values in your adjusted stock file. Another excellent alternative for quality historical stock data is Norgate Data who do have a Python interface … https://pypi.org/project/norgatedata/
Yes I'm hearing good things about Norgate. I think Andreas (Clenow) uses them as well. The fact the data is held in a 'walled garden' on your computer which is closed after subscription was a bit of a concern but apparently you can export in ASCII and actually thinking about it if there is a python api then once the data is sucked out into a dataframe it's effectively exported. The other annoying thing is that the local database is windows only so I would have to get a small python instance working under windows to get the data out. I also found this: https://github.com/fja05680/sp500 Rob
The CSI cash data for Stock Indices and FX futures is excellent and can certainly be used for carry calculations. For US rates, also excellent, however outside of the US, CSI have no available cash data for any interest rate contracts. Other than financial futures, the CSI cash data for a lot of the physical commodity markets can be somewhat dodgy with many many days seeing no change in price, so for commodities I’d suggest you avoid using cash for carry calcs and just stick to futures prices.
I can second that for stocks. Absolutely top notch, especially quality of splits/capital adjustments, dividends and historical index constituents. That the software is windows-only kind of sucks but at least you can export everything via python (or any other language if you reverse engineer their interface) on the same machine. For stocks I prefer Norgate's pricing and better API.