We have a new leader in the designer model competition

It should be noted that @judgetrade has taken the winning position in the designer model competition using default parameters.
https://www.portfolio123.com/app/r2g
It is not easy game to compete with Yuval.
Congrats! :clap:

3 Likes

Amazing work!

What default parameters? When I click Default in the Multi Sort I see @SZ models

1 Like

I think sorting by 5 year return maybe is what he means.

This is my default screen when I click on DMs.
But you maybe right that I personalised this earlier and it may be different for others.

2 Likes

@pitmaster,

They both do great (Yuval and Andreas).

One thing I like about Yuval and his models is that he has no survivorship bias. I think he may have removed a single model a very long time ago.

Just a positive for Yuval and not a comment on any other models. I generally do not keep track of survivorship bias and just happen to have followed Yuval close enough to be aware of this for his models alone.

I would love to have a t-test and a p-value on the models and do a Bonferroni correction using the number of models the designer has submitted to do the correction with. I think that is a feature requests that P123 would not be interested in which is fine. But I know what the Bonferroni correction would be for Yuval (not so large of a correction which is a good thing).

Comparisons are difficult for me without knowing about any survivorship bias there might be.

Best,

Jim

1 Like

I see. Your Multi Sort is ↑DDown1Y and ↑# Holdings but you are specifically sorting by the 5Y return column. So the Multi Sort is not used for the ordering.

We should have a 10y return field as well now that enough time has passed to have the field be useful. Capturing performance across wider time frames can help better capture how models behave across different economic environments.

1 Like

Every day in the morning (Europe time) I like to check DMs leader board.

While the leader is the same (based on 5Y return), I decided to calculate general statistics related to all models performance against benchmark, as show in the table below:

image

The results show that only 31% of DMs outperformed benchmark during last 2 years (5y/10y excess returns would be useful to check).

P123 has been doing a great job to help us to beat benchmark.

However, only ~30% users is able to do it using current methods. I hope the introduction of time-varying (ML) models will change this situation.

I would support @SZ request to add 10y return field. In addition to that, I would add 5y and 10y excess return.

I would bet every single one of those models beat their benchmark in the training sample. Without exception.

When R2G first started there was no point in having the 10y there, but now we have had more years pass and it finally makes sense to add it.

I will say, I notice a very concerning and huge focus on "the benchmark" on this site, by investors, and in the industry in general. It is almost obsession-like and people over the decades have lost sight of what the S&P 500 actually is. The S&P 500 is an arbitrary "algo" like ours with its own criteria and subject to its own issues and randomness. There is nothing special about it and from a scientific standpoint it does not reflect the median or average stock at all. For me my focus when looking at a strategy is on performing very well relative to the median stock in my universe on any given year and on good long term results vs other alternatives including but not limited to any one index.

Equal-weighting is more reflective of a more diversified index so I normally prefer S&P 500 equal weight or similar when doing studies. If you fine tune to the cap-weighted S&P 500 too much you may hurt your own performance as you are fine tuning toward outliers (given most S&P 500 gains come from a small % of stocks). The S&P 500 is currently heavily weighted towards businesses with great long term economics so it will probably keep doing well over the long term and focus on it will keep increasing, but its composition can change greatly over time. 2022 was a glimpse at what can happen when the whole market is focused on and concentrated on just 10 names. This is coming from someone whose personal account is beating the index. While possible, it is really not necessary to beat a specific index every year. If you look at most good track records they did not do that at all. I say this because I would not want to see people discouraged just because on year 1 you can't top some arbitrary algo namely the S&P 500. Nvidia, a single security, accounted for about a third of S&P 500 YTD gains. You are not immediately a bad investor just because you didn't hold it.

If the index spiked aggressively due to a huge amount of say quantitative easing and AI hype due to say 3 stocks in the future and those stocks go for 500x sales will it be a good benchmark? Will it be safe? Ask people in Japan who thought their popular index was safe in the 90's. I am very concerned not now but in the long term about what an indexing obsession can do and the risks it carries for our society when investors blindly assume it is safe, fairly valued, or representative of the average stock regardless of valuations, macroeconomic environment, or weightings.

7 Likes

New year, no changes in a leader based on 5yr CAGR but I checked 10yr CAGR performance. Top 3 DMs are as follows:

  • Super 5 Small Caps, DebtCycle Peak, 5 holdings, 21.6%
  • Sharper, Michael Thompson, 6 holdings, 21.6%
  • Keating's 20 5th Gen Microcap Trader, 20 holdings, Oliver Keating, 18.9%

I would like to also check 10yr: Sharpe, Sortino, MaxDD, StdDev, Downside Deviations.

I could make mistakes since I did it manually , ... @marco please think about it :slight_smile:

Oh, my! 77 trades last month and an average return of <1% per trade.

I wonder what happened to the designer? His last login was 5 years ago.

Good question.

He was from London with a master's degree in physics interested in becoming a finance professional. Also had skills (including math skills) consistent with this CV. So,I guess roughly 15% chance this is he (depending on how common this name is in England as a base case). Still 85% this is not him as a conservative estimate but best candidate I could find. Would need some sort of confirmation from someone at P123 who knew him better to know for sure: Oliver Keating

Edit: Hmmm… as per Grok 3 with DeepSearch turned on: "The full name "Oliver Keating" is rare in England, with an estimated 5 to 6 males sharing this combination." So maybe.

I communicated with him 10 years ago. His email address then was:
okeating@gmail.com

Also check the forum for: olikea on the forum.

Yes, that is definitely the gentleman who designed 5th Gen Microcap. I recognize his photo. I was around here in those days.