This might have an obvious answer, but here is my question: I have a screen that rebalances weekly; is it possible in a backtest to determine the returns of those stocks that are new to the results (versus those that had appeared in the prior week(s).)?
Doug
Yes. Run it as a simulation with the sell rule NoDays > 1. This simulation will sell all stocks every week and buy them anew. Then download the realized transactions. In Excel, sort first by date, then by ticker. Look at the return for the FIRST buy of each ticker. You’ll see what each ticker did on its first week appearing in your screen.
I’m not sure this is what you’re looking for, though . . .
Thanks, Yuval! Your answer leads to another question. My screens do not work nearly as well when they are turned into ranking systems and run as simulations. Can I run simulations that produce the same (or similar) results as my screens (I thought they had to be converted into ranking systems)? Thanks again for your help.
You don’t have to turn your screening rules into ranking systems in order to run them as simulations. But you have to be a little sophisticated in interpreting the results.
Here’s what you do. Run a backtest of your screen (with no ranking system) and note the MAXIMUM number of stocks that the screen produces.
Now open a new simulation.
If your membership level is ultimate, then go to the rebalance tab and use formula weight rebalancing with the following settings: ideal number of positions: maximum of screen; reconstitution and rebalance frequencies: same as your screen; max portfolio drift: 0; position weight formula: 1; weight constraints: 0, 100; max position drift and min rebalance transactions: 0.
If it’s not ultimate, put the inverse of that max number (in percentage points) in the simulation in the rebalance tab. So, for example, if the screen produces a maximum of 25 stocks, put 4 in the rebalance tab; if it produces a maximum of 50 stocks, put 2.
As your default ranking system, choose one of P123’s default systems. It will be irrelevant.
Put all your screening rules as buy rules.
As sell rules, put the opposite of your screening rules. For example, if a screening rule is EPS%ChgTTM > 15, put EPS%ChgTTM < 15 as a sell rule.
Now run the simulation.
If you have formula weight rebalancing, the result should be pretty close to your screen result.
If not, check the “charts” tab. Is there a great deal of variation in the # positions? Is your % of cash invested quite variable? Probably, depending on your screen rules.
So now you need to transform the data. On the summary tab, download the daily returns by pressing the down arrow under the chart. You’ll get a CSV file with information in columns A through M.
In cell N3, put “leverage” and in cell O3, put “$1 return”
Leave cell N4 blank and in cell N5, put =C5/G5. Copy that all the way down the column.
In cell O4 put 1. In cell O5 put =O4*((G5/G4-1)/N5+1). In cell O6 put =O5*((G6/G5-1)/N5+1). Then copy that formula for O6 all the way down the rest of the column.
That column will give you an approximation of the return of your screen.
Yuval, thank you for taking the time to offer detailed instructions–I appreciate it! I will test it out tonight. Doug