Discussion in 'Journals' started by globalarbtrader, Feb 11, 2015.
I don't code up the optimal roll policy in the past. It would be far too much work.
is there a reason for not leaving your funds in a CTA like your old firm vs managing it yourself? Is the management fee that significant, such that it will eat into your returns.
Partly. Also because I can do it myself with minimal ongoing work, and it's fun.
OK thanks. Could this be just be an error in that case? In your gold.csvfile, on 18 Jan 1980 your stitching algo rolls by 10 months which I think it much longer than any other gold roll.
Rolls are only moved in the past data if there is a gap in the available data so that a roll isn't possible on the date it should be.
Thanks for the clarification, GAT
How do you handle market data subscriptions in IB? For tax purposes, I recently moved from a personal trading account to a company trading account (the corporate tax rate is lower than the personal income tax rate in my country); however, I noted that once you're registered as a company with IB, they treat you as 'professional'. This means the market data fees go up between 2x-20x!
E.g., whereas I used to receive CME real time data for $6/month, now it is $90/month, plus now I need to additionally subscribe to numerous separate services that used to be bundled, all of which have gone up in price...
Have you found a way around the very expensive market data costs?
I am strongly considering switching to delayed data, as I am trading quite slowly (once a day max) anyway...
I'd be interested to hear your thoughts.
Also, do you have thoughts on why the mean square error for a trading system decreases by the inverse of the square root of the number of YEARS of data and not the number of DAYS of data?
I'm guessing the tax savings of incorporation are greater than (90-6)*12 or you could just switch back.
I'm really cheap! I get only the most basic feeds and my market data bill is around £10 a month. That means for example I can't trade the UK ICE markets which are really expensive.
The nuclear option is to get end of day data from quandl.com, and then trade market orders blind the next day. But for me at least I saved about 0.4%*400K = £1.6K by not trading market orders last year so this wouldn't make sense.
I'm not sure what you mean by "mean square error" so I'm going to guess.
If returns are uncorrelated and normally distributed then the standard deviation will always scale with the square root of time. So the daily standard deviation will be roughly 1/16 of the annual standard deviation.
The sampling error of the mean (or with certain assumptions the Sharpe Ratio) of a return series (like trading system performance) is equal to the standard deviation divided by the square root of a number of observations. A t-test for positive returns at some frequency is equal to the mean divided by the sampling error of the mean.
For example: Daily returns; average 0.01%. Standard deviation 0.4%
Analogous annual returns: 0.01% * 256 = 2.56% Standard deviation * 16 = 6.4%.
With 16 years of history the annual sampling error on the mean is 6.4 / sqrt(16) = 1.6%
A t-test for positive returns comes out at 2.56% / 1.6% = 1.6 (not significant, significance is normally 2)
With 16*256 = 4096 days of history the daily sampling error on the mean is 0.4% / sqrt(4096) = 0.00625%
A t-test for positive returns comes out at 0.01 / 0.00625 = 1.6 (same value as before)
(Both the numerator and denominator of the t-test have square root of time in them. So changing time has no effect)
The implication of this is that unless the system has a relatively high Sharpe Ratio, and trades relatively quickly, you can't get a narrower confidence interval*. To put it another way only getting a LONGER HISTORY of data will improve your confidence in your backtest, getting MORE FREQUENT returns won't help at all.
(Non Gaussian returns changes this slightly - in particular daily returns give you more chance of seeing the ugly drops that characterise negative skew. Correlated returns also changes this slightly - if they are negatively correlated using annual returns will flatter performance)
* With a high enough Sharpe Ratio just one year or a few months of data is enough to get a significant t-test using daily returns.
Separate names with a comma.