updated P123 ranking systems

yuvaltaylor · July 18, 2019, 9:37pm

We are replacing the “Basic” ranking systems with new “Core” ones. The old ones had not been revised in a very long time. All the old default ranking systems are still available in a folder labeled “Old P123 Ranking Systems.” The new ones are in a folder called “Core Investing Styles.”

Most of the main components of the “Basic” ranking systems remain in place, but we’ve added a few new twists. Please check them out and let us know if you have any questions.

We are planning to replace the default ranking system in new simulations with “Core Combination,” but that might take a little time. I’d recommend you do so yourself in the meantime.

jmh · July 18, 2019, 9:44pm

Hi Yuval,

New twists and ideas are always good!

When you have a minute, could you please elaborate on the major changes and why.

No need for a ton of statistical proof but a few words on:

Which node / component has changed?
Reason? Benefit?

Any one of them appears to you like a wonder standing tall above all others?

Thanks,

Jerome

RTNL · July 18, 2019, 10:50pm

I agree, please provide some color/rationale

yuvaltaylor · July 19, 2019, 1:08pm

Sure. Here’s a quick rundown:

We’ve added a brand new core ranking system: Low Volatility.
We’ve added EV and FCF components to the Value ranking system.
We’ve added short interest to the sentiment ranking system.
We’ve added industry momentum to the momentum ranking system, as well as incorporating a feature of the old “balanced” ranking system based on three-month intervals.
We’ve added asset turnover and excess gross margin to the quality ranking system, as well as made most of its components industry-comparable.
We’ve added operating income growth to the growth ranking system.
The combination ranking system incorporates all six of the others, with the most emphasis given to quality and value.

Marc, Riccardo, and I felt that the old core ranking systems were somewhat lacking in that they ignored volatility and EV-based value factors. I think I speak for all three of us when I say that these are the two most important differences between the old systems and the new ones. I thought that the other things I added/changed were important to improve performance. After all, the most basic membership level we offer does not allow users to create their own ranking systems. They’re confined to the ones we offer. So we wanted to make available high-quality, updated systems whose out-of-the-box performance will, we hope, be better than that of the older, more limited systems.

That said, if you think that our approach is flawed in some way, you could likely convince me to make some changes.

Jrinne · July 19, 2019, 1:20pm

Yuval,

Great stuff. And of course, we can change the weights or delete a factor or function that we do not like on our own.

Nothing but good. We should not make the perfect (not possible) the enemy of the good (or excellent in this case).

-Jim

Barn · July 19, 2019, 1:28pm

I think it’s a welcome change. I didn’t use the P123 systems much but I do like the thoughtfulness that has been added to these new ones and it gives me something to add to my arsenal

jmh · July 19, 2019, 4:31pm

Yuval - Interesting - thank you.

On a quick read of your email, this makes sense. Curious to have a play and see how well they fare vs old.

Jerome

piard2 · July 19, 2019, 5:34pm

Yuval, the words “we are replacing” are a bit scary. You say old systems are still available, I understand you ensure backward compatibility, which means you have created something new but not changed existing ranking systems and all screens and models using old basic ranking systems will keep their old behaviors. Is it correct?
Example: will the rule below still work, and select the same stocks in the past:
Rating(“Basic: Quality”)>60

yuvaltaylor · July 20, 2019, 4:46am

All the old rankings remain in place and we’re not planning to get rid of any of them. Your old rules will still work. However, at some point in the near future, new simulations will default to Core Combination rather than to Balanced.

piard2 · July 20, 2019, 1:57pm

Yuval, Thanks for your answer and for the new features.

Monday33 · July 21, 2019, 7:55pm

Great news!!! Thanks a lot!

superelastic · December 14, 2019, 7:08pm

That “Core Combination” RS is hot fire in a large-cap simulation I am playing with. Thanks for your good work Yuval.

SnortingElk · December 16, 2019, 7:36am

Very nice!

To add to the future suggestion list, from a portfolio standpoint, I feel like the current content is primarily focused on equities. Therefore, if I had a wish list around additional content it would include the following: (1) sim/ranking for real estate; (2) sim/ranking for bonds (bear markets); and perhaps (3) market timing. With a nice sample book that pulls everything together. I realize that international is coming next year, but that would also be included so the major asset classes are covered. Basically, P123’s take on a Swensen Portfolio.

Just a thought; what you’ve added is great.

judgetrade · December 16, 2019, 8:20am

Yes Core is very interesting!

Thank you!!!

See attachment: 25 Stocks, SP500 Universe…

judgetrade · December 16, 2019, 8:29am

See PDF

Sorry, do not know how to delete an attachment, First version is with nasdaq 100 second with sp500…

Core Combination.pdf (1.66 MB)

geov · December 16, 2019, 3:42pm

One can do better with “Core Combination” ranking system and 25 positions.
25 Seasonal Sector Stocks of the Portfolio123 Large Cap

judgetrade · December 16, 2019, 4:15pm

cool

geov · December 16, 2019, 4:27pm

Andreas can you run this from 1999. My research membership only gives me 15 years.

Here are the buy rules (all previously discussed in the forum):
Price>10 & avgDailyTot(20)>1000000 & LoopMin(“Vol(CTR)*Close(CTR)”,20)>1000000
Eval(between(Month,1,4),InSet(Sector,Discretionary,Energy,Materials,Industrial,TECH),True)
Eval(between(Month,5,10),InSet(Sector,Staple,Healthcare,tech),True)
Eval(between(Month,11,12),InSet(Sector,Staple,Healthcare,Discretionary,Energy,Materials,Industrial,TECH,),True)
(close(0,#spepscy)>ema(75,0,#spepscy)) or (close(0,$sp500)>ema(75,0,$sp500))

Sell rules:
nodays>25 | rank<70
PctFromHi <-10
(close(0,#spepscy)<ema(75,0,#spepscy)) and (close(0,#bench)<ema(200,0,#bench))

Note that between May and October the model only selects form sectors Staple, Healthcare & Tech.
Also Financials are not in this.
Also Andreas’ market timer is used.

I forgot to post the Formula Weight Rebalancing rules:
Positions = 25
Reconstitution = 1 week
Rebalancing = 4 weeks
Port drift = 0
Weight Formula = 1
Max position drift = 20
Min rebalancing transaction = 0

judgetrade · December 17, 2019, 6:15pm

Hi Georg,
yep looks good! this one is on the sp500