Model T now open for subscribers

Miro - you have 6 paragraphs of disclaimer but no real system description. Also, the price is set for $1,000/month which will be a hard sell.

I have six designer models, five of which are outperforming their benchmarks. My large- and mid-cap high-liquidity low-turnover model is boasting an annualized return of 32% since launch; my microcap model has 103% annualized since launch; my small-cap model is returning 29% annualized since launch; and my mid-cap model is returning 38% annualized since launch. All of these were developed after listening to P123 users’ advice. And, of course, I don’t have a single subscriber to any of them. I’m beginning to wonder whether designer models are a waste of everyone’s time.

The last time I checked - about a year ago - the total gross sales of all DMs combined was about $7K/month. The subscription pie isn’t very big and that’s a shame. Before finding P123, I spent some good money on services with much poorer performance. Without more P123 users, I don’t think DMs will provide anything more than beer money to developers.


EDIT: Current total DM revenue is 7595.06/Month, over 290 models.

It’s a confidence issue. There are a bunch of well subscribed DMs but they tend to have a longer trend and results. I subscribe to one of those. I’d probably sign up for more if the price was very low just for giggles. It’s too easy to have a DM that performs in the short-term and then flounders and is taken away. It’d be nice if every DM ever released goes to a separate page that can be looked at.

The # of trades per month on a lot of DM is also a huge turnoff. Most of them have easily over 5 trades per month. I Follow 3-4 of my own models and I make less than 10 trades a month between all of them.

Good point about confidence. That will require longer track records. We’ll see whether subscribers eventually show up. But on the other hand, they do love “hot” models even when when they’re new. Maybe P123 should survey subscribers when they signup and not just when they cancel.

Personally, I’m concerned about the number of trades only w/r/t it’s affect on the average return per trade. I avoid models with a low average return per trade (when those numbers are divulged) and those with low liquidity.

Reviewing graveyard models may be more amusing than enlightening. I know my current models are structured very differently than those I made several years ago.


Two years ago I peaked at $40K+ per year with Designer Models alone. But back then it was a circus with the way model backtest were presented - very open to abuse. Now P123 is doing things properly by downplaying backtests. Yes, revenues are more of a challenge but at least we are not misrepresenting system performance. In the early days it was a 100-yard dash, now it is a marathon.

Hang in there Yuval and don’t get frustrated… try to distinguish your models from the rest of the pack, I’m not just talking about superior performance but also unique strategies that potential subscribers can easily grasp. Keep subscription prices low and maybe one day you will be rewarded with a substantial following. You want to do some marketing yourself. A lot of the designers write Seeking Alpha articles for example.


Yes, I agree, P123 made the right DM changes. However, it’s still common practice for designers, when drumming up business, to post full sim results on the forum. That’s very bad practice and hurts both DM designers and subscribers and should be stopped. Miro announced his model properly. Hopefully, P123 will provide additional designer guidance and we’ll get back to where DMs deliver demonstrable value.


there is a model just below yours with the same 1-year performance. It only costs 2.5% of your model and still doesn’t have a subscriber…

I think the problem on p123 is that most subscribers are designers themselves with all the experience of seeing models falling apart out of sample. Once there is a decent number of models with >2years OOS outperformance, Marco and the team should consider an advertising campaign to attract more investors to the site.

What Needs to happen, if we want to sell Subs? The Need to be able to be booked without beeing a member of p123. There should be a Discount for the members, but I Need to be able to market them to a broader Segment (not only users of p123).
For me right now my modells are a way to build a Reputation, but that takes time (5-10 Years).
Regards Andreas

Just an idea that would address both of these issues, I think. In addition, this has been done before: you can judge for yourself how successful it is regarding marketing success and stock picking success.

Feature idea: Designer Miner

Specifically, the idea is to take the best of StarMine by Thompson Reuters and improve upon it. Image of StarMine output for NFLX below.

People will have better and more creative ideas for improvements and implementation than I have but this is what P123 is really good at and P123 could improve on this greatly. Let me give a few comments to give an idea of what I am thinking would work.

  1. Make no mistake. StarMine uses quantitative models. There are no analysts involved. So P123 and Designers could easily duplicate what Ford Equity Research or Columbine Capital Services or Zacks Investment Research.

  2. So substitute “Seven Sisters Services” and “Miro Institutional Investing” and “Gerstein Professional Management” and “Yuval Financial Factor Research” and “Primus Advanced Mathematical Quantitative Analyitics” and “JudgeTrade Academic Research Engine” for Columbine Capital Services ….etc. More a ranking system that an actual designer model.

  3. StarMine excels by keeping track of the performance of the models and weighing each model according to the historical performance for its output.

  4. P123 CAN REIMBURSE THE DESIGNERS ACCORDING TO THE PERFORMANCE OF THEIR MODEL and the weight placed in the the Designer Miner model…

  5. It would be an option of the designer as to whether a model would be included in the Designer Miner. People would probably want to develop new systems just for this. Again, the Designer Miner model would be more of a ranking system but it could just use the rank of one of the Designer Models.

  6. A Designer’s success with the Designer Miner would be an advertisement for a Designer Model that uses the same ranking system

  7. Designers would be paid a portion of the Designer Miner fees paid by people like me (I would pay for this). Reimbursement would be in proportion to how well a quantitative system performs which would encourage good models. If this takes off it could be a big pot. Money would come from volume.

  8. I would recommend 3 Designer Miner systems. Short, Medium and Long look-back periods for Designer Miner systems. Short look-back periods might adjust more quickly to new market conditions.

People will have better and more creative ideas than I have so I will stop here. But I think this would create a robust system that would outperform. It has already been done: you can evaluate the success of StarMine on your own. I have included an image of StarMine’s performance (below). It is not a high bar for P123 to improve upon.

P123 can do better than StarMine (which works). Some (okay almost all) of the data is old at StarMine. And they only have 5 categories: 3 Bearish for the NFLX example. Ranks would be a big improvement over the 5 categories. Importantly, I think, it competes with Thompson Reuters and not the Standard and Poor’s. The S&P might actually like this or could be convinced to like it. With the appropriate patents, trademarks and advertising the S&P might even end up buying this.

To the extent that it has worked for Thompson Reuters it should be considered for P123 and its designers, I think.


Interesting. Besides StarMine, there are some crowdsourced sites out there – can’t recall the names – but I do see how p123 and the p123 user community could do better. I also can envision how we could address a big weakness of a ranking system; that being the way it’s effectiveness often depends on the universe against which it is applied. IOW, using the NFLX example, we could present an overall rating and additional ratings that relate to particular investment goals (i.e. implemented via the universes against which all the rankings are applied.) Not sure this is making sense; it’s just thinking out loud.

P123 development-product plate is pretty full right now so I can’t promise anything quick. But I’m definitely intrigued by the idea; the wheels are turning, so to speak.

Makes perfect sense. StarMine looks at sectors but similar idea. Anyway the specifics are for P123 and designers. But what you say makes perfect sense.

Investars is another crowdsource site that does the same thing.

And like you, I have no doubt that P123 and designers would do better for multiple reasons. Just details and marketing would determine the popularity.

This is just an idea for P123 and the designers to use if they think there would be any benefit. Perhaps with the consideration of whether it would be an effective marketing tool (or not).

It would probably work as an investing tool. It would probably be robust (like a Decision Jungle in machine learning is). It should be better than a Decision Jungle because of the diverse but rational input of the developers and the adjustment of weights based on out-of-sample results. But it may not be better than what many people are already doing.

It does seem like a cool idea. And when I look at sector performance of the (automated) analysts, which analysts have the best records, the detailed recommendation of some of the participating analysts (e.g., Thompson Reuters/Versus) it keeps me on the web-page for a while. I keep coming back to StarMine when I purchase a stock. But it has not changed any of my purchases (which are based on P123 ports). But then again,it has not been done by P123 yet.



I find the P123 subscription universe a bit confusing. Is there a tier where users can just be DM subscribers w/ access to books?


It’s a momentum model designed for use by advisors.

Currently ranked #1 on the site using a multisort of the criteria pictured below.

Add 2Yr return to the multisort and I offer the next open model! :wink:

No advisor will recommend a model with such short history. There are now several models with over 2yr out of sample performance, and some have really good sharpe ratios. Very difficult to argue for a subscription to a much shorter-range out of sample model. And as others have pointed out, the price is too high - but that also counts for a lot of other models offered on p123.

I wish your model will succeed in the long run, but it will take some patience to see that outperformance.

To avoid any false impressions: Model T has the second-highest Sharpe Ratio of all of the models at 3.08 with the highest being 3.16.

The way the Sharpe Ratio works this is better than 2 years of results with a Sharpe Ratio of 2. [2sqrt(2 yrs) < 3.08sqrt(1 yr)]

I am skeptical of cherry-picking one model out of 259 designer models and drawing any conclusions at all. But as long as we are doing it and talking about Sharpe Ratios…….Well, I think it is fair to say that is a good Sharpe Ratio.


“It’s a momentum model designed for use by advisors.”

So you should really think about modifying the description on the system page because what you have there is butt-covering piled pretty thick with no description. Tell us more than “it’s a momentum model” as there are plenty of those.

If you feel you can attract an advisor’s attention (especially with the price tag) then you will probably have to do a lot of marketing outside P123.

Other than that… congratulations to a great start and good luck into the future!


You and Chipper are doing great work.