Microcap Stock Screener

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Microcap Stock Screener

Post by dothemontecarlo » Wed May 16, 2018 4:43 pm

This is the first time I have posted in many years. I'm returning in hopes of tapping into this forum's terrific theoretical insight.

I am trying to fill a microcap allocation. I am not satisfied with the fund options (e.g., BRSIX, IWC) currently available. BRSIX, in particular, has had a very large tracking error that previous Boglehead commentary suggests was unavoidable.

I am trying to gauge whether it is possible & feasible to create a factor-based basket of microcaps. With guidance from Larry Swedroe's "Your Complete Guide to Factor-Based Investing," I have created an Excel spreadsheet that curates a data dump from a Fidelity stock screener into various factor categories and metrics. I would attach it if I could, but I can provide in response to PM. I have not yet used the spreadsheet in connection with any purchases.

For each factor (e.g., size, value, momentum, performance, quality), I provide weights for each metric (e.g., P/B, FCF/Market Cap, past 5-52 week performance, etc.) as well as for each factor. I can sort the stocks by a single metric, by a single factor (itself a composite of multiple metrics), or by a composite weighted factor score.

As an example, I currently have allocated the following weights to each factor:
Company/Size = 15% (a subcombination of 70% size (rank), 20% security price (rank), and 10% volume (rank))
Value = 20% (itself a subcombination of 40% BtM (rank), 40% FCF/M (rank), & 20% S&P valuation score (scaled by # of companies in data dump))
Momentum = 20% (itself of subcombination of 30% 13-week price performance (exp-stretched rank), 20% 5-52 week price performance (exp-stretched rank), 10% revenue growth (rank), 10% EPS growth (rank), and 30% FCF growth (rank)
Performance & Quality = 20% (subcombination of 10+ metrics involving returns, leverage, SD, turnover, and S&P's quality score)
Analyst Guidance = 5% (e.g., combination of sell/buy recommendations)
Accounting Trustworthiness = 20% (to screen out companies with dicey financials)

To summarize, the composite score is generated mostly from the addition of weighted ranks--the sum over all selected metrics of (metric rank * weight). Companies with missing data in the data dump (e.g., no declared revenues) are deeply penalized. So are penny stocks.

Finally, when generating a portfolio, allocations are 2/3 market-weighted + 1/3 evenly spread

The sheet then scores the portfolio on each of the factors. For instance, an Excel sort of one Fidelity data dump of companies (with less than $650M market cap, an average daily volume of at least 1K, a ROA > 0, an operating margin > 1%, and at least 50 employees) produced a 30-company basket with a weighted average market cap of $321M, a weighted average P/B of 1.86, and having the following percentile factor weights (100% is best):

Company/Size: 52%
Value: 74%
Momentum: 79%
Performance & Quality: 74%
Analyst Guidance: 66%
Accounting Trustworthiness: 73%

This approach could definitely be improved. But how? Relying on rank values, for example, is somewhat arbitrary. What might be better? In the end, the generated basket has fairly weak value, momentum, and other factor characteristics. Some outside commentary claims that it is better to combine a pure value and a pure momentum approach, without muddling the factors.

No doubt, AQR and various multi-factor funds face the same questions. They have the benefit of a greater database of stock data, including historical, and simulators and optimizers --which I don't. But at least a stock basket approach would give an investor the benefit of deep microcap exposure without significantly moving the price with small trades. And this approach would also bypass the significant fees associated with many multi-factor funds.

Looking forward to your insights and critiques.

-Eric

User avatar
Rick Ferri
Posts: 8484
Joined: Mon Feb 26, 2007 11:40 am
Location: Georgetown, TX. Twitter: @Rick_Ferri
Contact:

Re: Microcap Stock Screener

Post by Rick Ferri » Wed May 16, 2018 4:47 pm

Eric,

My response:

Total Stock Market Index Fund

and forgetaboutit.
The Education of an Index Investor: flounders in darkness, finds enlightenment, overcomplicates strategy, embraces simplicity.

Nate79
Posts: 3056
Joined: Thu Aug 11, 2016 6:24 pm
Location: Portland, OR

Re: Microcap Stock Screener

Post by Nate79 » Wed May 16, 2018 6:12 pm

Sounds like a lot of work to me. What do you expect to gain over TSM or the microcap funds that already exist?

User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Re: Microcap Stock Screener

Post by dothemontecarlo » Wed May 16, 2018 6:56 pm

Nate79 wrote:
Wed May 16, 2018 6:12 pm
Sounds like a lot of work to me. What do you expect to gain over TSM or the microcap funds that already exist?
Larry Swedroe's book reported the following annual premiums:
Equity (broad market): 8.3% from 1927-2015
Size (CRSP 6-10 vs CRSP 1-5): 3.3% from 1927-2015
Value (by BtM): 4.1% from 1952-2015
Momentum (ranking stocks by returns over past year but not including past month, and comparing top 30% to bottom 30%): 9.6% from 1927-2015
Profitability, measured as most profitable vs. least profitable: 3.8%
Profitability, based on ROE: 7.4%
Quality, based on various measures: 2.8% to 5.8% for small stocks, based on July 1988-June 2012

Obviously, the factors partially overlap, so one should not expect to reap the sum of the factors through stock selection. But it is conceivable that, if these factors persist, even in reduced amounts, screening for these factors could reap a significant composite premium.

lack_ey
Posts: 6583
Joined: Wed Nov 19, 2014 11:55 pm

Re: Microcap Stock Screener

Post by lack_ey » Wed May 16, 2018 7:11 pm

So is your goal to have multifactor loadings or to have microcaps? You kind of have to prioritize at some point.

What's your starting universe of stocks, anyway?

Rank is probably fine in doing the sorts, but something like z-scores may be a little better. Or something in between. Frankly, z-scores can be problematic when you have outliers in the data set; you could do some kind of variant or cap scores at around 3/-3 sd away from the mean (or some kind of soft cap). Or pre-normalize somehow. I doubt this would make a huge difference.

If you want factor weights more than diversification, you could weight stocks by scores rather than by market cap, at least to some extent.

In general there may be better ways to deal with missing data, but it depends on why data is missing, which figures are not present. Sometimes it's appropriate to impute an average or guess based on other data points available, though if missing data is more like "there's no actual revenues" then you can just interpret it that way. What do you mean by "deeply penalized" anyway?

Tanelorn
Posts: 1502
Joined: Thu May 01, 2014 9:35 pm

Re: Microcap Stock Screener

Post by Tanelorn » Wed May 16, 2018 7:29 pm

Nate79 wrote:
Wed May 16, 2018 6:12 pm
Sounds like a lot of work to me. What do you expect to gain over TSM or the microcap funds that already exist?
Of course. Presumably he hopes to outperform, both TSM and existing microcap funds. Surely the latter can be done with enough effort - not only don't you have the 1-2% expense ratio of a fund, you also have the luxury of barely moving the market with your trades since you're only investing a few $100k or whatever across dozens of small stocks instead of $100M's or $B's like a mutual fund. In addition, you can potentially consider a lot more stocks too - those that would be too small for a fund but not too small for you to invest $1000 in personally.

Is it worth it? Well, that depends on how much OP enjoys the process and values the potential gains in terms of hopefully better returns. As always, no guarantees. You pay your money, and you take your chances.

User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Re: Microcap Stock Screener

Post by dothemontecarlo » Wed May 16, 2018 8:04 pm

lack_ey wrote:
Wed May 16, 2018 7:11 pm
So is your goal to have multifactor loadings or to have microcaps? You kind of have to prioritize at some point.

What's your starting universe of stocks, anyway?

Rank is probably fine in doing the sorts, but something like z-scores may be a little better. Or something in between. Frankly, z-scores can be problematic when you have outliers in the data set; you could do some kind of variant or cap scores at around 3/-3 sd away from the mean (or some kind of soft cap). Or pre-normalize somehow. I doubt this would make a huge difference.

If you want factor weights more than diversification, you could weight stocks by scores rather than by market cap, at least to some extent.

In general there may be better ways to deal with missing data, but it depends on why data is missing, which figures are not present. Sometimes it's appropriate to impute an average or guess based on other data points available, though if missing data is more like "there's no actual revenues" then you can just interpret it that way. What do you mean by "deeply penalized" anyway?
Thanks for your excellent feedback. (And yes, Z-scores would be more elegant!) I'm starting with a universe of microcaps. I hastily figured that if data wasn't available, then perhaps the company's numbers were fishy. I will give this a more careful look. Generally, I "penalize" stocks that don't meet basic screening criteria by giving it a very bad rank score for the affected factor.

I have also screened out companies with fewer than 50 employees. I know that there are very profitable publicly traded companies with far fewer employees, including at least three companies with 0 listed employees. Would you avoid them even if they reported high earnings and margins? I would love some insight, because this screen significantly increases the average weighted market cap of the screened basket.

Thanks,
Eric

wbrianwhite
Posts: 49
Joined: Fri Dec 19, 2014 9:44 am

Re: Microcap Stock Screener

Post by wbrianwhite » Wed May 16, 2018 8:19 pm

I'd check out http://www.aaii.com, they have a model portfolio of microcap stocks and they review it regularly to add or remove based on screening guidelines. They run a real portfolio and have a long track record for it

I always thought it was interesting, but too much work for me.

wbrianwhite
Posts: 49
Joined: Fri Dec 19, 2014 9:44 am

Re: Microcap Stock Screener

Post by wbrianwhite » Wed May 16, 2018 8:21 pm

Microcap is really the only area where broad outperformance is possible, because institutional money just really can't play there. The downside is spotty analysis and limited data.

DenisD
Posts: 18
Joined: Sun Apr 19, 2009 9:16 pm

Re: Microcap Stock Screener

Post by DenisD » Wed May 16, 2018 8:25 pm

I would sign up for a trial account at www.portfolio123.com and backtest some of your ideas. Unfortunately, they just raised the prices and reduced the maximum backtest length for new clients. AFAIK, you can still pay by the month. If you wanted to refresh your microcap's once per year, you'd only have to pay for 1 month. But I would confirm that.

I've been an intermittent client for more than 10 years running large and small-cap screens.

User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Re: Microcap Stock Screener

Post by dothemontecarlo » Wed May 16, 2018 8:31 pm

Brian & Denis,

Thank you for those great links!

Eric

User avatar
whodidntante
Posts: 3742
Joined: Thu Jan 21, 2016 11:11 pm

Re: Microcap Stock Screener

Post by whodidntante » Wed May 16, 2018 8:31 pm

Rick Ferri wrote:
Wed May 16, 2018 4:47 pm
Eric,

My response:

Total Stock Market Index Fund

and forgetaboutit.
Or you can sell S&P 500 futures and load up on microcaps for a nice loading on size. :wink:

User avatar
abuss368
Posts: 12594
Joined: Mon Aug 03, 2009 2:33 pm
Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!

Re: Microcap Stock Screener

Post by abuss368 » Thu May 17, 2018 9:57 pm

Rick Ferri wrote:
Wed May 16, 2018 4:47 pm
Eric,

My response:

Total Stock Market Index Fund

and forgetaboutit.
:sharebeer
John C. Bogle: "You simply do not need to put your money into 8 different mutual funds!" | | Disclosure: Three Fund Portfolio + U.S. & International REITs

User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Re: Microcap Stock Screener

Post by dothemontecarlo » Mon May 21, 2018 5:02 pm

Following up...

In my Microcap stock selection worksheet, I finished incorporating lack_ey's much-appreciated suggestions (e.g., using Z-scores and (partially) weighting the final stock selection by the composite scores, giving those with higher scores more weight). I have also an optional sector weighting scheme, to force the planned portfolio to be weighted by sector targets you provide.

The spreadsheet is designed for Fidelity stock-screen data dumps. It could be (relatively easily) modified to other stock screen dumps. If anyone is interested in a copy of the Excel file, please send me a PM.

I still haven't implemented anything, just in case I am persuaded to abandon this effort. In the meantime, feel free to "front run" me!

-Eric

User avatar
aegis965
Posts: 128
Joined: Sat Sep 03, 2016 6:06 am
Location: Taipei, Taiwan

Re: Microcap Stock Screener

Post by aegis965 » Mon May 21, 2018 8:58 pm

dothemontecarlo wrote:
Wed May 16, 2018 4:43 pm
For instance, an Excel sort of one Fidelity data dump of companies (with less than $650M market cap, an average daily volume of at least 1K, a ROA > 0, an operating margin > 1%, and at least 50 employees) produced a 30-company basket with a weighted average market cap of $321M, a weighted average P/B of 1.86
Why $650M?
I may be biased.

User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Re: Microcap Stock Screener

Post by dothemontecarlo » Mon May 21, 2018 10:02 pm

aegis965 wrote:
Mon May 21, 2018 8:58 pm
dothemontecarlo wrote:
Wed May 16, 2018 4:43 pm
For instance, an Excel sort of one Fidelity data dump of companies (with less than $650M market cap, an average daily volume of at least 1K, a ROA > 0, an operating margin > 1%, and at least 50 employees) produced a 30-company basket with a weighted average market cap of $321M, a weighted average P/B of 1.86
Why $650M?
It's the way Fidelity's stock screen defines "microcap," if I remember correctly. One of the limitations of Fidelity's stock screener is that it only allows you to download the first 200 stocks and related stock data in the list. By reversing the alphabetical order, you can increase that to about 400 stocks. I'm not aware of any way to download more than that number. For that reason, I chose limits within the Fidelity stock screener to produce a list of no more than 400 stocks.

User avatar
aegis965
Posts: 128
Joined: Sat Sep 03, 2016 6:06 am
Location: Taipei, Taiwan

Re: Microcap Stock Screener

Post by aegis965 » Wed May 23, 2018 9:20 pm

dothemontecarlo wrote:
Mon May 21, 2018 10:02 pm
aegis965 wrote:
Mon May 21, 2018 8:58 pm
dothemontecarlo wrote:
Wed May 16, 2018 4:43 pm
For instance, an Excel sort of one Fidelity data dump of companies (with less than $650M market cap, an average daily volume of at least 1K, a ROA > 0, an operating margin > 1%, and at least 50 employees) produced a 30-company basket with a weighted average market cap of $321M, a weighted average P/B of 1.86
Why $650M?
It's the way Fidelity's stock screen defines "microcap," if I remember correctly.
This makes the average market cap of your portfolio larger than many mass market microcap funds. If you end up in the same playfield IMO defeat the purpose of doing it yourself.
I may be biased.

madpunster
Posts: 109
Joined: Sun Nov 02, 2008 12:58 am

Re: Microcap Stock Screener

Post by madpunster » Wed May 23, 2018 10:17 pm

The AAII Stock Investor Pro software might suit you. You can screen the US universe by your criteria and rank quantitatively offline. The Shadow Stock Portfolio (SSP) is the screen discussed above; mkt cap 30-400 million.

The farther away from large caps, the wackier the numbers, the more prayer involved, bigger spreads etc. So screening on fundamentals and executing on technicals is what I've been kicking around. The program has simple charts with monthly bars.

The other odd twist is that there is a noticeable bump in the selected stocks after portfolio additions are announced quarterly. So some are using this program to make good guesses on which stocks might be chosen prior to the announcement and sell on the bump up. I think Jason Zweig wrote about this in the WSJ.

I have no connection to AAII other than being a member and using this program when I occasionally fall off the Boglehead wagon. :oops:

User avatar
aegis965
Posts: 128
Joined: Sat Sep 03, 2016 6:06 am
Location: Taipei, Taiwan

Re: Microcap Stock Screener

Post by aegis965 » Wed May 23, 2018 10:37 pm

I suggest 25-100m market cap. This way you have reasonable liquidity and considerably less big players. But as madpunster said, data are noisier, so rudimentary DD is a must even if you want to go full quant.
I may be biased.

KyleAAA
Posts: 6626
Joined: Wed Jul 01, 2009 5:35 pm
Contact:

Re: Microcap Stock Screener

Post by KyleAAA » Wed May 23, 2018 11:28 pm

Are those factor loadings? How did you get such high exposure to both value and momentum at the same time? They usually dilute each other. Doesn’t smell quite right.

User avatar
aegis965
Posts: 128
Joined: Sat Sep 03, 2016 6:06 am
Location: Taipei, Taiwan

Re: Microcap Stock Screener

Post by aegis965 » Thu May 24, 2018 12:50 am

KyleAAA wrote:
Wed May 23, 2018 11:28 pm
Are those factor loadings? How did you get such high exposure to both value and momentum at the same time? They usually dilute each other. Doesn’t smell quite right.
I think those are characteristic rankings not factor loadings.
I may be biased.

User avatar
dothemontecarlo
Posts: 539
Joined: Tue Nov 06, 2007 10:19 am
Location: Colorado
Contact:

Re: Microcap Stock Screener

Post by dothemontecarlo » Thu May 24, 2018 10:44 am

aegis965 wrote:
Wed May 23, 2018 9:20 pm
This makes the average market cap of your portfolio larger than many mass market microcap funds. If you end up in the same playfield IMO defeat the purpose of doing it yourself.
The average weighted market value of my portfolio is $161M. (This is weighted by the purchase amount of each stock). This is accomplished by giving adequate weight to the size score.
KyleAAA wrote:
Wed May 23, 2018 11:28 pm
Are those factor loadings? How did you get such high exposure to both value and momentum at the same time? They usually dilute each other. Doesn’t smell quite right.
I would need a definition & mathematical formula for determining one's "factor loading" to answer your question. I just calculated percentiles. They do dilute each other. But there are some stocks with both above-average value & momentum characteristics.

Post Reply