some advice about launching designer models

I was planning to launch some designer models in a few months and I had a few questions.

Right now I am planning to launch eight models. I’m listing them in order of potential returns, with the first model probably having the highest return.

  1. Minimum daily dollar volume $250,000, 5X turnover, weekly rebalance
  2. Minimum daily dollar volume $250,000, 5X turnover, weekly rebalance, low beta
  3. Minimum daily dollar volume $250,000, 2X turnover, 3-month rebalance
  4. Minimum daily dollar volume $250,000, 2X turnover, 3-month rebalance, low beta
  5. Minimum daily dollar volume $1 million, 5X turnover, weekly rebalance
  6. Minimum daily dollar volume $1 million, 5X turnover, weekly rebalance, low beta
  7. Minimum daily dollar volume $1 million, 2X turnover, 3-month rebalance
  8. Minimum daily dollar volume $1 million, 2X turnover, 3-month rebalance, low beta

All of these models are loosely based on the ranking systems I currently use for my own investments. See below for my live portfolio returns based on this system, with variable slippage.

Are these all appealing or are some less appealing? Do you have any suggestions about this? How many stocks should I put in each model? I’m thinking of charging $100 per subscriber—how many subscribers should I allow per model, and does this price make sense? Should some of these models be more expensive than others? I like to update my ranking systems from time to time as I perform more backtests, as I come up with new ideas, and as the markets change—is this OK or is it frowned upon?


These are all micro-caps. If $2,500 is invested in each stock, then you have a portfolio value of 17 x $2500= $42,500 and approx slippage will be 1% of trade value. So if you have 5 subs they should limit their total investment to $10,000 each for acceptable slippage.

Slippage is related to trade size, average volume traded, trade duration, daily volatility and shares outstanding. The slippage formula can be found in this article:
http://www.cims.nyu.edu/~almgren/papers/costestim.pdf

I’m sure everyone has different criteria when searching for models, but I generally look for:

  • about $50 max
  • 5 to 10 holdings. I may want more than one model in a book, so too many holdings spreads my already meager cash too thin.
  • average winners return > 10%. Some models have good returns, but they make many trades at about 1-2% each. Slippage or bad timing can make these unprofitable.
  • some kind of market timing. It might just be scaling back the number of holdings when conditions aren’t optimal.
  • some kind of liquidity filter. I’d like to avoid illiquid stocks if possible.
  • tweaking your model over time isn’t a terrible idea. I like to know that the designer is keeping his finger to the pulse. However, frequent revisions might mean the model is too curve fitted.
  • the number of subscribers doesn’t really matter to me unless it’s a micro cap model.

That’s about it. Your Crazy Returns 1 model looks great! How did it backtest?

Georg -

My crazy returns model is a microcap model but the ones I’m proposing, with the higher minimum average daily volume, wouldn’t be so microcap-oriented. And I only use variable slippage, so the high slippage is already factored in there.

Manster -

Thanks so much for the advice. In terms of liquidity, is $250,000 minimum average daily dollar volume high enough? Is $1 million?

My crazy returns model backtested very well indeed. The liquidity is low, though–$50,000 minimum average daily dollar volume. I’ve been steadily improving it since I launched it. Right now a 7-year backtest gives an annualized return of 57% and a Sharpe ratio of 2.18 with 17 holdings.

  • Yuval

Yuval,
$50,000 seems low to me. It’s probably fine for 1 person to trade, but i usually look for at least $250,000 when choosing designer models.
The 57% return sounds amazing though…

Hi Yuval,

I offer one small/micro-cap designer model with 150,000$ ADT. I think this low liquidity is OK given that turnover is only 3x. So effectively your first 4 model options are very similar.
All my other designer models have 1MM $ ADT, and then with lower slippage the turnover is no more a concern.
If I had to subscribe to one of your options, I would take 1, 2, 5 or 6. I don’t like very low turnover models because I think the backtest results might statistically not be very significant. You might want to check that using a rolling backtest with 1 year periods and 1-3 month offset.

Looking forward to see your designer models live!
:slight_smile:

Rolling 1 year returns with 1 week offset is better.

yuvaltaylor,

Pleaese, post your model’s simulation return graph for 2014 and 2015.

2016 most of the models posted positive returns along with all the positive returns bench mark.

In 2014 and 2015, very few designer model performed good in micro and small cap universe.

Thanks
Kumar :slight_smile:

Sevensisters and Georg -

Yes, I develop my low turnover models using rolling 1-year returns and a 1-week offset period. Sevensisters, thanks for pointing out that with high liquidity, turnover slippage is less of a concern. I was thinking that some investors might not want to bother rebalancing so often and would like a simple 3-month rebalance. If that’s not the case, I could get rid of the low turnover models. Thanks for the feedback!

Kumar -

Here is my method simulated for 2014 and 2015, as requested. As I said above, this is for a $50,000 minimum dollar volume; my $250,000 minimum would certainly look somewhat different.

  • Yuval

These days I only offer large cap models because I myself got burnt, and I want my subscribers to succeed.

In the early days of R2G, I subscribed (briefly) to a very promising microcap model–together with maybe a dozen others. Slippage ate up my entire return.

It wasn’t so much the slippage from the other subscribers directly, (although this was also an issue), but the fact that among the many subscribers some inevitably did some sloppy trading which telegraphed the trades to the market, and the market pounced on it.

I would consider offering a microcap model only if it is limited to very few subscribers; maybe 2 or 3.

Chaim -

How likely is this to happen with stocks with a $250,000 minimum average daily dollar volume? I have no idea how much money is traded by subscribers. I can see slippage being a problem if millions of dollars are being spent purchasing my recommended stocks, in which case perhaps I should only offer the $1 million minimum models?

  • Yuval

Chipper, I totally agree with you. If you realistically can get a +20% annualized return with reasonable drawdown from a large cap or ETF model then you have a desirable model. Slippage is a huge problem for microcaps.

[quote]
Chaim -

How likely is this to happen with stocks with a $250,000 minimum average daily dollar volume? I have no idea how much money is traded by subscribers. I can see slippage being a problem if millions of dollars are being spent purchasing my recommended stocks, in which case perhaps I should only offer the $1 million minimum models?

  • Yuval
    [/quote]Yuval, I can’t tell you for sure.

It depends on the number of subscribers, the amount that each subscriber invests, and the time/patience of their trading.

There are at least one or two subscribers to designer models who manage millions. Presumably, they use VWAP or other algos to spread trades over a few days. Still one of those subs alone can trade 10% of volume. [EDIT: Even without a fund subscribing, I am under the impression that the average subscriber allocates about $5k to $10k per position. Let’s assume that the average is about $10k to be a bit conservative (some subs don’t trade at all, they just reserve a spot). If you get 10 subs, that means that they are buying about $100k altogether. That’s 10% of the volume for a $1m ADT stock.]

Another example: Let’s say you have 10 subs. 3 of them don’t have the time to monitor the market on Monday morning from 9:30am - 11:00am (or so). Those three impatient subs may get very aggressive with their orders and move the market.

Alternatively, if 5 subs put in limit orders at the same time, the market makers will look at the market depth and notice the excessive demand. They will smell blood and raise their asking price. In this case, even a low % of volume can indirectly move the market.

You have no control over what your subs do.

I’m thinking this wouldn’t be as much of a problem with the buy-and-hold-for-three-months models. Slippage is much less of an issue there.

But my other quick thought is that if people are investing millions and paying only $50 a month for our hard work, why are we selling our ideas for such prices in the first place? It seems like we’re enabling our own downfall.

Sorry, there is something wrong with my calcs.
How does one delete an attached picture file?


Yuvaltor,

design model with more liquidity and tradable mumber of positions 10 or 8 etc., and reasonable return.

You will get better investment ideas in the process.

Thanks
Kumar

Yuval,

You need to re-assess the price you plan to charge for your systems.
I assume that you want to launch Designer models in order to make some additional cash.
Below are all the current Designer systems that have a price between $50 and $100 and also have subs.
There are 51 system in this price range, and only 10 have any subs, with a total of 30 subs.
All but 9 of those willing to pay over $50 are requiring 100% gain since launched.
And more importantly, no one is invested in a system in that price range that isn’t at least 2 years old.
So it is obvious that subs are requiring high return and long out of sample record before they will pay that price.

:+:Quantonomics 20 Small Cap Gems:+:; Gain 91% since 05/16/2013, 4 subs, Price $100

Best9(Russell1000); Gain 30% since 12/03/2013, 3 subs, Price $60

TWY 5 stocks HG EMA 11 SYS; Gain 334% since 03/17/2013, 10 subs, Price $88

1st! 5 Liquid Earnings Kickers; Gain 110% since 12/29/13, 4 subs, Price $100

Sherman’s way 2 go with a low Max Drawdown; Gain 105% since 03/24/2013, 1 sub, Price $100

1st! 5 Stock Different; Gain 106% since 03/02/14, 2 subs, Price $100

TWY 5 stocks Liquid MicroCap $350m; Gain 44% since 09/07/2013, 1 sub, Price $88

1st! 5 New Liquid Estimators; Gain 103% since 02/21/2014, 2 subs, Price $100

SMS SP500 Market Neutral; Gain < 1% since 09/22/2014, 1 sub, Price $65

Keating’s Extreme 5 (mktcap>$100m, vol>$400k); Gain 105% since 10/21/13, 2 subs, Price $100

Denny -

Thanks so much for the advice.

I’m very grateful for the responses I’ve received to this post. They have been, however, very discouraging, and I’ll probably not offer a designer model—at least not like the ones I’ve proposed. I just don’t see a big upside to it. Why would I want to put in all the hard work to enable someone to make huge profits for a recompense of $40 a month? Especially if it has the potential to hurt my own investing success through increased slippage.

I mean, take this, for example. Here’s a simulation of a 20-stock $1 mil minimum average daily volume model, and exactly the same model with 50 stocks just to show it’s not curve-fit, both using variable slippage. With 62% winners, a Sharpe ratio of 1.64, a 4X turnover, and an annualized return of 36%, is that worth only $40 a month? That seems nuts to me.

  • Yuval


“And more importantly, no one is invested in a system in that price range that isn’t at least 2 years old.”
Denny, one way to deal with that is you introduce your model at a lower price and then openly tell subscribers that as the performance is demonstrated, that the price will go up. Subs can leave at any time if they choose.
I have not tracked the existing models but I don’t think many developers do that now.
I think that is only fair (and rational from an economic standpoint).
If I had a ‘high’ certainty that I would make much more than $100/month in profit from using a model (this of course is related to my portfolio size), and it has been demonstrated, then it would make sense to pay it.