How Accurate is Portfolio123?

General
1

I come across this post from portfolio123 saying pitroski f score give bad performance using other backtest platform:

https://blog.portfolio123.com/why-piotroskis-f-score-no-longer-works/

But if we see some backtest performed using other platform with recent data, all of them have same concenssus saying piotroski is good strategy:

https://alphaarchitect.com/2015/05/value-investing-research-simple-methods-to-improve-the-piotroski-f-score/

https://www.quantconnect.com/research/15728/piotroski-f-score-investing/p1

https://www.oldschoolvalue.com/investing-strategy/piotroski-score-screen-performance/

https://www.equitieslab.com/putting-piotroski-to-the-test/

https://www.quantrocket.com/blog/piotroski-f-score/

https://stockmixology.com/piotroskis-f-score/

Now im curious how accurate is portfolio123 platform, what makes it show different result?

And is there any standarized backtest framework or condition so we have same playing field to make it easy compare between researh. Something like quantopian have in the past when they have list of condition in their strategy contest.

2

If you want to compare the P123 results to these other sources, then you need to verify all these inputs are the same:

  • Similar universe of stocks
  • Liquidity rule
  • Slippage and commission settings and the rebalance frequency
  • Number of holdings (if there is a limit set)

There are probably small differences in the implementation of Piotroski's rules, but you would have to read the details in those links to determine this.

I skimmed the content of the first 4 links to get an idea of the differences:
First link - written in 2015 so results only go to 2015. The results are for 1974 to 2014, so we dont know what the annual return was for 1999-2021 period mentioned in that blog post. I trust that Alpha Architect used reasonable liquidity settings, but I didn't read the details.

Second link - See this comment at the bottom of that page: "the way it is setup right now is essentially it looks nice on the backtest but the liquidity for the stocks is so low you wont be able to realize the same results if we were to actually trade using this strategy it feels like.... and also the strategy completely breaks down if we change the universe to liquid stocks only which is not encouraging"

Third link - testing only goes to 2009 so didn't look at it further.

Forth link - has rule marketcap > 1m. If there is no liquidity rule, the results are meaningless since a 1m marketcap could be a busy hotdog stand - not something you could actually trade. Just for fun I ran our Piotroski screen with the rule mktcap > 1 for the US Incl Foreign Primary universe with 0% slippage and the results for 01/02/1999 - 05/07/2024 was 36.7% annual returns. But as I said, it is totally meaningless without a liquidity rule.

2 Likes