| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Latest Release » Thread: IMPORTANT: price functions no longer suffer from future splits |
|
Total posts in this thread: 9 |
[Request a Feature] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
marco
|
All price functions like AvgVol(), Close(), SMA etc, now return values unaffected by future splits. In other words when these functions are evaluated in the past (in a simulation for example) all future splits are reversed out. You can now safely use these functions in simulations like: Close(0) > 5 AvgVol(20) > 50000 Previously the above function suffered from forward-looking bias because the result was affected by future splits. This was a very confusing issue and the way around it was to only use price functions relative to other functions. For example: Close(0)> SMA(20) You can now safely do this and Price is now equivalent to Close(0). NOTE: If you were using rules like "AvgVol(20) > 50000" you may now get different results when you re-run simulation. ---------------------------------------- Portfolio123 Staff |
||
|
|
marco
|
FYI, We're going to delete many old posts where the Close(0) vs Price problem was discussed to avoid confusion. ---------------------------------------- Portfolio123 Staff |
||
|
|
rickg777
Advanced Member
|
OK, There goes a tremendous amount of work, over 1 1/2 years, right out the window. All simulations using AvgVol have to be rerun. At least we will have a firmer grasp on reality . Trusting the back testing program will be harder now. It will have to be proven in real time now. Rick |
||
|
|
BJS
Advanced Member
|
Marco, There is still some confusion on my part with AvgVol(10) vs. Vol10DAvg. If I understand it correctly, these two finctions are still not exactly the same. AvgVol(10) would be updated on a daily basis. However, Vol10DAvg is a Reuters parameter, and thus only gets updated once a week. Is this correct? I ran two sims that compared the two. First I ran them with daily rebalance, and got much different results. I would expect this if the above explanation is correct. AvgVol(10) should be more accurate for daily rebalancing, since it is always current. However, I went back and ran them with weekly rebalance and still got much different results. I would expect that the results should be the same, or very close. If you do weekly rebalancing you thoeretically should be using current data for either function. So I still don't understand what is going on. Am I looking at this correctly? Should AvgVol(10) be more accurate than Vol10DAvg? If so, why don't you just get rid of Vol10DAvg to eliminate the confusion. Rick is right - it makes it harder the trust the results if what you thought was fact turns out not to be. Please help. Thanks. Brian |
||
|
|
BJS
Advanced Member
|
....continued... MIght the difference be due to rounding error in the Reuters function? Brian |
||
|
|
marco
|
I don't think it's rounding. For IBM AvgVol(10) returns 5,342,330. Reuters gives us 5.34232999 that multiplied by 1M gives the same result. For a smaller company like CTAC AvgVol(10) gives 55,840, Reuters gives us 0.05589 which is 55,890. So there's a slight difference there but not enough I think. Is the sim you are running based on a few stocks and possibly very small stocks? With sims that hold a few stocks even one stock that is different can cause a huge change. ---------------------------------------- Portfolio123 Staff |
||
|
|
BJS
Advanced Member
|
The sim holds 7 stocks, but they are predominantly small caps. |
||
|
|
rcflyerco
Advanced Member
|
Marco, I thought I understood this topic when you posted the announcement about price functions being corrected for future splits, but I am confused. If there is a discussion of this topic that I have missed, just point me there. My confusion involves the case where a stock splits (say XYZ splits 2:1 in June of this year). How does this affect the values reported prior to the split? Suppose that a simulation is run over the past year and that in April XYZ traded 1000 shares every day the market was open and that every trade occurred at $10 / share. If I understand correctly, Close(0) would return 5 during the month of April. Price would return the same value. Is that correct? Would AvgVol(5) return 1000 for a day in late April or would it return 2000? I think MktCap is not affected by these changes so it would return approximately ShsOutMR * price (the price that traders payed in April, not the split adjusted price). Does AvgDailyTot return the average amount of money that actually changed hands for some numer of days in April? Is that the same value used to calculate PctAvgDailyTot? If I am trying to avoid buying a stock that trades at less than $1 at the time of purchase, whether in April or October, what command should I put in the simulation? Would Close(0)>=1 work? Price>=1? Help, Rick |
||
|
|
DennyHalwes
Advanced Member UNITED STATES Joined: Apr 28, 2004 Posts: 1532 Status: Offline |
Rick, It no longer makes any difference which you use. Close(0) > 1 will yield the same Sim results as Price > 1. 2 weeks ago after Marco started this thread I re-ran 3 of my Sims that had stocks that later split with Close(0) instead of Price, and the Performance was the same. Denny  ---------------------------------------- "The significant problems we face cannot be solved at the same level of thinking that we were at when we created them". Albert Einstein |
||
|
| [Show Thread Printable Version] |
Re: IMPORTANT: price functions no longer suffer from future splits - but I do, lol.
