When I run a backtest with the screen using the sp1500 as the universe why do I get substantially different results than when I use all fundamentals for the universe and buy rule Universe(sp1500) ? Shouldn’t they both yield the same results? Are they not both point in time.
Any input much appreciated.
Mike,
if you have FRank or Rank rules in your screens, they will select different stocks, according to your 2 universes (Stocks, which are not selected by a rank rule with “all fundamentals” universe can be selected by the same rank rule with SP1500 universe). Then the results of the screens should differ also.
Matthias
Mike,
It is counter intuitive but even if you do not use Frank changing the universe will usually affect the relative ranks unless you have only one factor/function in your ranking system.
Consider the example where you have Market Cap as a node in your ranking system. In the SP500 universe a stock xxx may rank 40 in this node. However, it may rank say 95 in the all fundamentals universe for this node. On the other hand, a stock like Apple would go from 100 to 100 (no change) when the universes are switched. If that node has adequate weight, the 40 rank in that node for stock xxx could really drag down the overall rank for that stock in the SP500 universe but would drag it down less in the all fundamentals universe. This could change the relative rank compared to another stock in the SP500 where everything else except the market cap are similar. For example, if stock xxx was a little better than Apple with regard to all of the factors in your ranking system except for market cap then stock xxx could rank better than Apple in the all fundamental universe but rank far less less than apple in the SP500 universe.
Confirmed. The sp1500 from the dropdown is slightly different than using Universe(sp1500) with All Fund. It’s the same if you use All Stocks.
Try this
Set universe to ALL STOCKS
Enter these rules
universe(sp1500) and universe(allfund)=false
In 3/1/2005 there are two stocks MKTSQ.^08 and WAMUQ^12 that apparently are in the Sp1500 but have no fundamentals.
I’ll have to investigate. Thanks
Thank you everyone for your replies.
Marco did you find anything with your investigation. Yesterday I was getting backtests with 5% performance difference Now it seems to be around 2% I am not sure if I changed something from my end. Thank you
Still waiting. Not sure what the 5% vs 2% is , we didn’t make changes. Sometimes there are data fixes from them for old data.