do you guys still think you have an edge?


Great points and could not agree more.

But even at P123 there are different tiers of data. Me, I like to backtest more than 5 years—and happily pay for the information that more than five years of data gives me about the stock market.

You have the good sense to say information asymmetry can exist. It exists even among those reading this.

Historical data is data too. We pay for both present and past data (information). It is asymmetrical even at P123.

I am just saying information asymmetry exists. Not how important it is.

BTW, I think you would agree that any information asymmetry that occurs at P123 is not cheating. It is not similar to using mirrors to look at cards.

David, if the Marines need a few satellites (or whatever is needed) to develop some asymmetrical information, I am all for it. I would say Oorah!, but I think I cannot do it right;-)


The Efficient Market Hypothesis seems to assume no data asymmetry (everyone has the same data). I don’t accept EMH and, as pointed out, at least at the individual investor level not everyone has the same data. Even if they did, they would not give it all the same importance.

I’m still trying to shake the brain cyst image…!

@Regallow, the EMH is a model. Not even its progenitors believed it was reality.

To account for this, they cast the EMH in terms of strong-form, semi-strong form, and weak-form. Reality is supposed to lie between the strong and weak forms, but the semi-strong form is too precise to account for variability.

I’d like to add some informational firepower to add to this debate. Here is the rolling alpha (1 year holding period, 1 month intervals) from my flagship strategy. If you were me, and this represented your best efforts, how could you now doubt the information dispersion thesis?

So you still have 8% alpha? Is this theoretical or did it translate to real life alpha?

Personally I have tried some short term systems recently and they have blown up in my face from the extra slipapge/commision/transaction costs/spreads. They may have a ‘small’ edge without these factors.

Information asymmetry should be looked at in terms of what IS and is NOT available to the public.

I think what is more accurate would be to say that data availability and manipulation technique can create an edge. I don’t see that as an information asymmetry.
No where in that definition do you see the word FREE. Let’s try to keep that in mind.

As for an edge lets look at what creates a benchmark. The benchmark is an aggregate of the winners and losers. If nobody beat the benchmark that would mean that nobody lost to the benchmark either.

Right. Wikipedia does not use the word “free” either. In fact, the Wikipedia definition does not seem to care why the asymmetry occurs.

“I do not want to pay more for that data service” does not seem to be excluded as a reason for why someone on one side of a trade might have less information than the other person, however.

In 2001 the Nobel Memorial Prize in Economics was awarded to George Akerlof, Michael Spence, and Joseph E. Stiglitz for their “analyses of markets with asymmetric information”. I assume there was a clear (and probably better or more complete) definition for the term in their papers.


Cool that we get to use our own definitions on the forum—which is the norm now. There is a limit to how much we can learn when everyone has a different definition and are not really talking about the same thing.


Available, yes. Free . . . where did that come from. Your pc, laptop or tablet isn’t free. A cp of coffee isn’t free. Yahoo isn’t free (though you might be lulled into thinking it is because advertisers, not you, pay). Nothing in life is free. Your car isn’t free. Your home isn’t free.

If you want alternative information, go buy it. It’s up to you to assess the cost of the information versus the cost of the excess return. (I hypothesize that the hype is grossly exaggerated relative to the marginal dollars and cents benefits achieved. But if you disagree, find the money, borrow if you must, and go get it.)

It’s between 0% to 8%, depending on which regression you want to use. The given regressions are linear and logarithmic, which are all Excel provides.

My sense is that a hyperbolic regression is more appropriate for modeling information diffusion dynamics. The reason for this is that the rate of decay (diffusion) decreases as a function of itself. Hyperbolic discounting is used in some branches of behavioral finance and in oilfield decline analysis (i.e., Arps’ hyperbolic decline model).

This is theoretical, but jives with what I’ve seen in my accounts.

Yea where did that come from?

Tony first uses the word “free” and says it is not in his definition of information asymmetry. I show a definition from Wikipedia that shows that it is not in that definition either (in response to his post). And of course, you mentioned it.

I have the same question as you Marc: Why are you guys mentioning free?

I said above, I happily pay P123 for the excellent service it provides. And P123 is helping to make the information more available to people. Many people no longer need a SP500 license to get this information.

The reasonable price that P123 charges may give me an edge over someone who tries to get all of their data from Yahoo!, Reuters Google and the web: not necessarily a bad thing IMHO. But those people (using Yahoo!) would be limited to the same information whether P123 provides that information to its members or not.

My only point is that information asymmetry exists for more than one reason and it is not necessarily rare. It isn’t just about satellite images.

There are books one can read about this.

Honestly, I am not understanding your last 2 posts. Not even close to understanding. I am going to assume that you have not read (but skimmed) the above posts. Even so, this “knowledge and information is universal” is straight out of a……Actually, I am not sure that I have seen this even in a movie or a New Age Book.

But to be clear on my end: I am a big fan of P123 and I am not a communist who thinks everything should be free. Marc, you will have to look to your Democratic Party, AOC and Bernie Sanders for that;-)

The people at P123 work hard and deserve to be paid more. In fact, I look forward to helping with that. I look forward to the time when I will have enough capital to justify paying for the coming API. When I get there it will be because of P123.

Yea, I will get that Mac Pro when I get access to P123’s API. I do not think it will be free;-)


Hyperbolic discounting is an interesting topic!!! I cannot confirm it has use for this but it belongs in any detailed look at the Dividend DISCOUNT Model.

David, thank you for raising this interesting topic.


On the topic of edge. Here’s a recent article that is not very in depth on the topic but does make some valid points as to why this is hard. However I disagree that a model can’t beat the market.

“Computer Models Won’t Beat the Stock Market Any Time Soon”

One of those overview articles that tries to make everyone happy even if they contradict their own headline: "Using machines to beat the markets is a really difficult challenge,……But I don’t think it’s impossible.

But I do like this:

“One tool that Renaissance uses is linear regression, which a high school student could understand”

Yea, some of the machine learning techniques used by Renaissance Technologies are simpler than optimizing a P123 ranking system. A linear regression is probably not as good, however. Not as good because of nonlinearity of the data and the problems of collinearity (for multiple regression) that P123 can handle with good optimizations.


I think it’s quite hard if the other person has better information (knows if a stock is likely to beat or miss earnings).

For example BBBY (the computers were buying this one because it was revised up and cheap on a value metrics. But the smart money were selling it before the earnings. They knew BBBY is likely to miss earnings (due to the massive miss in HOME). Of course now BBBY guides down and AQR and co are selling now. If you know how these rule based quants work, then you can often front run them…

This comment made me curious to try my own system, the one that I use to trade every week, to see if the results were the same. I’m glad they weren’t. The results are below. Except for the GFC period, the alpha has been pretty consistent between 2004 and 2018, with a big drop in the last six to twelve months (which I hope is temporary). (Please note that the simulation uses some leverage so you should subtract a portion of the alpha pictured.)

I don’t think this proves anything about information dispersion one way or the other. If I tried extremely hard, I might be able to design a strategy whose alpha has steadily increased since 2004.

Yuval said: “the alpha has been pretty consistent between 2004 and 2018, with a big drop in the last six to twelve months (which I hope is temporary)”

This is the quandary I am in. All of my sims look good BUT the OOS performance over the last two years, compared to SPY, is poor. This is leading me to believe that none of my ports are really robust for the current environment - the regime that they were simulated in is no longer reality. And I am not sure what the new reality is.

I will let Yuval make any clarifications but I think I am in agreement.

I recently listened to an O’Shaughnessy podcast. He made me remember what his father, Greenblatt (and everyone) has said. There will be periods where any system does not do well. That CAN extend a few years although I am looking for relief about now (unless there is a recession). In any case, you have to be able to hold through those periods—whatever it takes. That may even mean holding more cash after learning from this drawdown (in the future). It also means having a good system that you have confidence in.

I think Yuval said now may be like 1999. But whatever this period is like, it will probably revert back to what it was. I do not think there has been a sudden change in information asymmetry or anything discussed above. Even if the Quant Funds became much better suddenly (because of quantum computers or whatever), they still have about the same amount of money (as before) to make the market more efficient through arbitrage. That limits how quickly things could change.

BTW, about the only thing I ever disagree with Yuval about is the importance of cross-validation and out-of-sample stuff. And here he provides his out-of-sample results to support his view.

Anyway, I think I am saying the same thing as Yuval and I leave any clarifications (or poor understandings on my part) to him.


From OSAM’s last quarterly report…
Same thing with AQRs products. Wes Gray’s (alpha architect) ETFs have done awful the last 12-18 months. Basically it’s been a bad period for everyone. This wasn’t just a slow decay, it was everything in the toolbox stopped working all at once. The old timers like like Jim O’Shaughnessy and Cliff Asness say they’ve seen periods like this before in their careers, so I guess we’ll see if this time it’s different.

I want to thank you for this. It’s good to hear the general case stated so baldly; it’s good to know that we’re not alone in suffering through this period in which all our beliefs seem to be reversed.

Josh Brown has pegged the underlying reason behind this shift: the incredibly low cost of capital. When Everything That Counts Can't Be Counted - The Reformed Broker It’s a great read and it’s scary as hell. And it’s why I’ve been wondering whether 2019 is the new 1999.

Short answer, yes.

But it doesn’t look like it at this very moment. The unexplainable happens from time to time for the same reason that you have an edge: mr. market is irrational. There are going to be periods when things don’t work and there are going to be periods when you will crush the market. Very difficult to explain and understand why, but, historically when things look bad and don’t make sense it is usually a good time to be patient, hone your skills, remind yourself why things are the way they are, and continue buying those tasty hamburgers you know are a good deal.

Can you even think of a better alternative? Cash, commodities, crypto, bonds, real estate? Compared to stocks, those are pretty laughable investments in the long run (excluding leveraged real estate). Sure you might be able to make money dancing in and out of different assets if you are lucky. But thats not really investing, its speculation.