You certainly should. You sound like you want to apply some rigour to what you want to try to do. Using free data would undermine that from the start.
I noticed this as a gap (for me) the other day. But given that kind of synthetic series you're alluding to can be coded up in custom functions I'm interested in why you see this as a gap. I noticed this gap because I've been coding up some functions in Tradersstudio and have been plotting them for verification. An indicator based upon a synthetic series can't be plotted as Tradersstudio stands now. What I will say is that building on the "data alias" feature of the data universe to allow construction of syntheic series would be a good design move for Tradersstudio. But like I said above, you can simulate any seriies you wish in code - but maybe that's not your point. Thx D
Hmmm....well, maybe I just don't know how to do it in Tradersstudio. So, let's say I want to calculate the % of stocks in an index that are above a 50 day simple moving average. I want to plot this, and I want to buy, when, say, the index is less than 20 and sell when it is higher than 60. Any sample you can provide would be greatly appreciated.
Well, I don't do stocks (which I think you know) so I don't have an example of this type of thing. How I might try to do this (in pseudo-code) if I'm understanding you correctly is: "Function StockStrengthIndex(StockIndex as Array, Lookback) For StockSeries in StockIndex BenchMA = Average(StockSeries[UBound(StockSeries) -1], Lookback) If StockSeries > BenchMA Then StockStrengthIndex = StockStrengthIndex + 1 Else StockStrengthIndex = StockStrengthIndex - 1 Next End Function" ...this is where my thinking would start from anyway. I don't think you'd have a problem with this. You've done quite a lot of coding haven't you? Open Question: My assumption is that a 2D Array of stock ticker series would be loaded into an Array(StockIndex) as input to the for loop. Tradersstudio may have a more elegant way of management stock index components than that. As I said, I don't do anything with stocks right now...maybe i should for some coding laughs Interestingly, I know that a bunch of array handling functions have been added to tradersstudio in the most recent beta (I think it's that one). I suspect those improvement may, in part, be motivated by the need to manhandle groups of stocks in custom functions when bararrays can't be used. Thx D
I do not see how this is a problem. At any given time when a purchase is going to be made in a backtest, the current value of the stock at that point needs to be used (the $34 instead of the 11 cents) for position sizing to be accurate, and then since we are only going forward from that point, it should be relatively easy for someone to program how the computer handles the price in the event of a future stock split. The price is going to fall by exactly half, or by 3 or something to that nature, one could program the software to recognize this I would think. Could this not be done with Amibroker?
Thanks for your response. Is there no way to get more than 20 years of data? I would really like at least 50 years... Is this amount of data available for other programs?
Fifty years of what? If you want stock data going back that long then you're talking about paying many thousands of dollars. I'm not a data guru but older data has more accuracy issues associated with it - particularly stock data.
I've considered doing something similar but that method, without storing the result in another data array, is pretty darn slow. Every time you run a system, you'll run that across all stocks for every bar. I suppose that I could write out the result to a file but I'm not familiar with that area in Tradersstudio. As a side note, I've asked Murray about how to program a rotational trading basket about 10 times and he says it can be done but so far I haven't seen an example.
So what you'd like to do is store the calculated values for later use in order to speed up testing. Okay. Yes, i suppose you'd need to dump the output to a data file. I do think the suggestion of creating synthetic data series as part of the data handling system is a good one though. Keep pushing him. From my experience of it's capabilities I'd expect tradersstudio to be able to do that. It's generally well put together for exploring reasonably complex systemic appoaches. Thx D
Agreed on the synthetic data series - I think it'd be a great addition. The issue for me is that I often create synthetic data series that I reference outside of a trading system - so I might be trading all stocks, but want my synthetic series created out of a small subset. My other issue with TS is workflow - I find the process of using Trading Sessions and Systems, overall, to be very slow to work with. The number of times I've had to unload and reload a trading session because I've made changes in the underlying system - it drives me a bit batty.