Going live tonight: a better Open price estimation algorithm

Announcements
1

Dear All,

As you may know the Compustat database did not record Open prices before approx. mid 2004. When we discovered this we added a simple Open estimation as (hi+lo)/2

We did some statistics with actual open prices and we found that a better Open price estimation is as follows:

If the privous close is between hi & lo, use previous close
Otherwise use (hi+lo)/2

This the best we can do for Open prices with a very simple algorithm. We do have the real historical open prices from ourInteractive Data databse. However this project is much more difficult since it involves merging two database. We do hope to do this in the near future. In any event a better open price algorithm is always useful because there’s always the possibility of it missing.

If your sims/backtests use the Open price some slight differences are to be expected.

Let us know of any issues.

Thanks

2

Marco,

Do you have a plan to merge the 2 databases so we can have the actual open prices prior to mid 2004? If so, do you have an expected date? Most everything about the Compustat data is better than the Reuters data except this. Our Sims and Screener backtest would be more accurate with the actual data.

Denny :sunglasses: