Yo Bro

I need to hear your point of view on my conclusion. I mentioned in another thread that I gave up on stock picking earlier in my career due to lots of time spent chasing very little reward. A fund I was at (not a small cap) beat it’s benchmark by ~50bps since inception over a 8 year period. However, our asset base was relatively low and the math conducted on my end found fees paid for our expertise ate up any alpha gained. In any case, from a client perspective, I felt that the clients we had would probably be better in a Vangaurd index with all of us moving elsewhere in finance to add real value. This is what I did. I had a really hard time seeing the value we brought to the end client from a fee adjusted return.

On another note, the financial industry is getting saturated with quant shops, hedge funds, algo trading, ai, and what not. How does a traditional analyst conducting a DCF stand a chance against a computer running an arbitrage algo? I know you hate EMH, but with so many participants operating 24/7, how do grand mispricings still exist in say the S&P 500? I realize you operate in a small space with few participants which strikes me as a pseudo PE type of engagement.

I’m starting to spin in circles in thought here, but honestly I’m interested in your point of view of traditional beat the market investing in an age of AI, algo, and heavy quant shops.

I have some of the same thoughts as you CvM. However, I wouldn’t be too concerned with the HFT shops, because a 1/2 cent difference on a buy and a sell doesn’t really affect people with holding periods measured in months and years, except for the flash-crash type risk.

Other algorithms and computer screening probably does mean that there are far fewer “fat pitches” than there were 20 years ago, but that doesn’t necessarily mean that there aren’t some good pitches worth striking at. There are still aspects of human behavior that computers can’t quite sniff out yet. Maybe one day those will disappear too.

It seems to me that what one has to find are mistakes in fundamental valuation that pass under the radar of most analysts. Computers are going to use formulas to calculate fundamental variables and they likely won’t catch the subtlties of the competitive context or changing demand, except indirectly through mass trading behavior. If misconceptions about fundamentals persist, they won’t be observable in price behavior, which means that it will be hard for a computer to spot it unless it is very specifically tuned to look for that. A human is still more adaptable than a computer to a wide variety of contexts, and that may change one day, but that day isn’t here yet.

So my sense is that it is getting more difficult, but opportunities are still out there, in part because of herding behavior, and in part because human beings don’t buy and sell based on the rationality of a computer. Whether it makes sense to play this game vs index and produce a genuinely valuable service in another industry is a separate question.

Curious to hear Bromion’s take on this.

I agree, I think it’s really hard to outperform consistently outside of small caps. Apparently some funds can do this – Einhorn has done it. But he hasn’t beaten the market by that much after fees and that’s with a world class infrastructure. I think the market is as efficient as it has ever been due to the rise of the internet over the last 15 years and then the subsequent rise of quant shops.

I think a traditional analyst conducting a DCF has been completely commoditized. How is your DCF going to be radically different from anyone else’s? I don’t even do DCFs.

I don’t buy S&P 500 stocks, so I can only take a shot in abstract terms. It seems mostly efficient to me unless you are talking about multi-year time arbitrage in which you would sort the 500 and try to buy the stocks that will benefit the most from creative destructionism, and then try to find the cheapest of those. Then you would buy and hold. That strategy would make money.

Keep in mind that the majority of institutional capital cannot buy stocks under $500mm or $1B, so to throw big dollars into the stock market, they are mainly focused on the S&P 500. These are obviously the deepest stocks in the market but they are also the most crowded and efficient. Under $500mm is a completely different world. The upside is most of these stocks do not have thorough analyst coverage, and they have wider and faster earnings swings in both directions. However, there is also a lot of fraud and self dealing that you don’t get in S&P stocks, so it can be a trade off (I personally think the fraud is easy to spot and avoid but every year people get blown out on fraud stocks so maybe not).

What we are doing is a hybrid model. There are lots of smart analysts that underperform. I looked at it said, “I am not going to be the smartest analyst on Wall Street so I need to find a different approach.” I’ve spent basically the entire last year building software. It is more of a software company now than a hedge fund. The software is what would happen if you combined a quant shop with a value shop. So far the results are good but it’s a short time period. We also take companies private and do activist deals, so we can attack things from a few different angles, most of which are not available to most funds.

Not sure if that answered your question.

I think, in general, if you want to distill this down to a sound byte I would say this:

There are 7,700 US listed companies (primary listings only, this excludes secondary listing foreign stocks)

About ~4,500 are above $20mm of cap (really hard to buy below that level and many of these are just shells or companies about to go bk). Only about 1,700 of these are above $1B of cap – this is “institutional Wall Street” That leaves 2,800 below $1B of cap, of which about 2,200 are below $500mm This last category is what I call the Abyss – most institutions can’t invest below this level, but there are 2,200 stocks available here. In my entire career, I’ve only looked at maybe 500-750 total stocks (in around 7.5 years). Most of these stocks are invisible. They don’t have analyst coverage (or have sparse coverage) and they don’t regularly attend conferences. The only way they get traded up or down is based on results that show up on screens. The game is front running the traditional screens. If you could, for example, find a way to reliably identify earnings beats and misses or guidance revisions, then you could place the trade before these factors become “known” – however, quant funds are also trying to do this. We took a hybrid approach, identifying both qualitative and quantitative factors. The qualitative stuff is really interesting and hard to screen for, that’s the challenge and the opportunity. There are only a handful of things a stock can do: The results can get better, they can get worse, the company can get bought, or it can go bankrupt. That’s about it. After a while you begin to quickly dismiss possibilities or at least likelihoods. You can look quickly and say, “XYZ is this type of stock, and these are the relevant questions, can I get an edge there if I try? No? Next.” If you look at a few hundred a year from a carefully filtered list, you will find mispricings below $500mm, that is a guarantee. Of course it takes several years of good experience before you can even really step up to the plate, but it can be done.

So to give one quick example, I bought stock in AGYS at around $8 about a year ago. It’s around $14 today.

The analysis was really simple – money losing company with a good software business underneath two bad divisions (one of which was already spun out). The software is recurring revenue in a growing niche. New management was stripping away the dead wood. The balance sheet is a rock.

If you look at the competitors, 2/3 of their largest competitors are in the same small town in Georgia. That’s really bullish, AGYS should be a buyout. The new CEO / CFO have experience fixing, growing and selling software companies. When I talked to them, it was clear that around 65% of the OPEX would be redundant to a strategic… just down the street. That’s fantastic, the stock was even cheaper than it appeared relative to peers on an EV/Sales multiple (peers trade at 3x, but AGYS was selling for <1x at the time since the market hadn’t rerated it up to its new peer group).

It’s going to take a while to play out – they rolled some of the cash into R&D to improve their software. It’s a SaaS model and they are the first / best mover with this product. It is starting to roll through the P&L, but it will accelerate once the R&D expense (which is depressing earnings) rolls out of the base in 2H14.

In a buyout, which I think is highly likely, the stock is in the $20s. The largest shareholder is a smart fund that can’t get out without a sale.

Meanwhile, insiders have been buying hand over fist and the company previously refused a buyout at $11 several years ago. The cash burn was minimal and improving. It’s “priced for nothing” since it was a dog for years due to the legacy businesses, the company had no coverage at the time, and they never attended conferences or spoke with Wall Street.

You can figure that stuff out in about 1 day. The stock is up 70-75% since then and I think it could be up another 75-100% from here over time.

I don’t care what the DCF says, the change is the opportunity because it will screen better later.

There are plenty of other examples like this in the market at any point in the cycle.

I believe it was O’Shaughnessy that found that the small cap effect is real, but is most effective for small-caps that are so small that institutions are basically unable to invest in them because of restrictions on ownership sizes. This does seem to be the opportunity for the rest of us non-institutional types.

Thanks for your insight, bromion. Always good to hear a pro’s thinking.

Any suggestions on a good programming language to learn?

I hire out all the programming, but I assume C# is probably good. I’m not the best person to ask though.

My opinion is you want to focus on what is going to give you the biggest bang for the buck. In my case, I could learn to program, but the thousands of hours it would take is better spent elsewhere (for me personally). The most money in this business is made by business people who solve problems and get results, not tech people. You can make six figures in a tech job and that’s a good living in a good job, but understanding business and being able to execute offers significantly more upside IMO.

If you want to do HFT, C++ is kinda the standard.

If you want to do algorithmic prototyping, and getting the answer a few seconds (or minutes) faster or slower doesn’t matter for what you are doing, Python is good. Matlab and/or R are also options. I am most comfortable in R, but am putting more work into doing Python these days.

If you want general purpose programming but don’t want to go into the full intricacies of C++, then Java isn’t bad. Android uses a Java framework too. C# is basically a Microsoft dialect of Java that integrates a bit better with MS systems.

If you are a fundamental analyst and just want to automate some of your spreadsheet work, then VBA is your baby.

If you are primarily running stuff in browsers, Javascript and JQuery is good.

If you are querying databases, you’ll need to know SQL, but often you don’t need to know all that much unless you are building the databases as well.

Yes VBA is what most of my stuff is built off of, but you can hire that out really easily. I don’t even really consider VBA programming – I guess it is, but it is programming-lite compared to some of the stuff bchad noted.

VBA is a great skill to learn relatively quickly that adds value in the industry. I wish I knew it.

Good stuff, thank you.

This is really good reading, lots to think about. Thanks for taking the time to write out your thoughts everyone!

Bro, mad respect!

I’m aware of a HF manager in Chicago that plays in the space bromion spoke of earlier. Basically he finds 10-15 companies that are under covered and where the value play or catalyst isn’t yet recognized. Over the past 10 years I think he’s averaged 33% which is remarkable.

The downside of this strategy is that its incredibly high conviction and hard to scale up. He isn’t able to manage more than around 400M.

30%+ compound returns on a portfolio of $400mm where you return a portion of the capital in some years, replacing it with your own capital, is a great business model. Most people would be thrilled with that. Personally, I am not really excited about hosing my LPs by raising the maximum AUM I can and diluting the returns. Most of the street uses that model, but I’d rather be small and nimble with excellent returns.

I think I could scale to a billion (in theory – practice is a lot harder) because we have other lines of business than just equities. The equity portion would be hard pressed to go above $500mm unless we started moving up the cap range, which is possible but we’re not set up for that today. The software program I built focuses primarily under $2B of cap. There is a limit to how much you can deploy. But even with a $300mm portfolio, a 33% return is $100mm and the manager takes 20% of that. Not terrible.

Unlike this other HF manager though, I go with low conviction. I do not like to “fall in love” with individual stocks, and diversification provides really significant benefits. We will probably target 150 positions across both short and long as we scale it up. What I have found, which is counter intuitive, is that the longer I spend researching a stock, the worse I do. I generally know with all of my best picks within 10 minutes if it is a good idea or not, and then I might spend another 1-2 days vetting it. If it’s not a very strong candidate within 1 hour, I will move on. It’s like speed dating.

That’s an interesting observation about “more research; less reliable.” And it seems to line up with other research that shows that more information does not necessarily lead to better decisions.

It may come down to diminishing returns of additional information.

Maybe it comes down to additional research making you fall in love with the stock.

Maybe it comes down to the fact that more time researching leads to more opportunity for the market to correct any mispricings and therefore missing out on the fat pitch.

It probably is also somewhat ideosyncratic in terms of what kinds of mispricings bromion’s approach is best able to find. That is to say, that the kinds of things he can identify in an hour may not be the same kind of things that a someone else can identify in an hour, but that’s ok, it may be best to identify, put it into the hopper, and move on to the next identifiable opportunity. In theory, one would want to find every possible market opportunity imaginable, but in practice, it’s generally good enough just to spot the ones that you yourself are suited to find and keep going.

Bromion, how do you plan your position exits? Do you have a target return or a target valuation or a target holding time? Do you have stop-loss techniques? Depending on your strategy, each of these can work, but I’m curious what works best with the way you think (to the extent that you can talk about it).

My premise has always been that there are only a handful of factors that matter for a particular stock. The 3rd or 5th most relevant factor is exponentially more important than the 20th most relevant factor. If you can nail the key factors within a short time frame, additional effort is not necessary. Said differently, if I can’t figure it out in 1-2 days then what good will 10-20 days be?

There is also an element of, “This is obviously a good idea based on X, Y and Z with no major offsets to those” as opposed to “I talked myself into this after spending a huge amount of time researching.” The only times I have lost money in the stock market are where I convince myself some convoluted situation is good or true. There are thousands of stocks, why ever go for something convoluted? I always hear people say stuff like, “This might be a fraud but it’s so cheap, I have to buy some…” Wait, whaaat? No, don’t buy that.

There are also additive learning benefits from looking at many stocks – you learn more and you learn in a non-linear way. The learning compounds faster, and you can amortize that effort: “I passed at X but now it’s X-30%, this could be really interesting now.” Then you can dust it off quickly.

By covering a lot of ground quickly and taking small, fast positions in high probability winners, you can also structure a low risk (diversified) and low volatility book while maintaining the upside characteristics of most concentrated portfolios. The best performing portfolios will always be very concentrated, but so will the worst performing portfolios. In order to win, you must first not lose. Compounding takes care of the rest.

In terms of position exits, I will sell if the reason I own it changes adversely. I will sell if it hits some estimate of fair value, and I’ll sell if it looks like the upward move is running out of steam based on the fundamentals. Really strong performers usually overshoot fair value, so no need to sell immediately if it’s ripping. Because the positions are small, I am fine holding them for long periods of time, several years if necessary. Obviously anyone would prefer to be immediately right, but that’s not required as long as I am ultimately right within roughly 2-3 years. If I’m running a concentrated book, I need to be right immediately. If the position goes against me, I have to sell or cover to avoid blowing up my year. The diversification is the stop loss – even if one goes to zero, it’s 1% of capital or less.

The strategy is more of a value harvesting approach. It’s not about being smarter than other people, it’s about finding mispriced stocks on a multi-year time period, harvesting the inefficiencies, and then waiting. It requires excellent sourcing, good judgment and time efficiency.

On AGYS, it took me about an hour to figure it out. I bought some stock. I spent the next 1-2 days researching it in depth, talked to the company, bought more. My good friend also owns the stock. He runs a book of 12 stocks and spent two months researching it. He will make more money on it, but will he make more money on a time efficient basis? What if we had both been wrong? Then he would have lost a lot more money or had a much higher opportunity cost.

This only makes sense for a professional money manager, not for a PA, but it makes a lot of sense in the right context.

One of my buddies in equity research bought some AGYS as well. I think he is a bro fan, even though he has never met him ha ha

Lots of good info on here. Bromion- appreciate all this.

AGYS gapping up on Monday morning, I think so.

Good stuff Bromion.