Base Rate View-- Mauboussin PDF

[netnet]

Mauboussin has written a (small) book on base rates.  It's very good and free.

 

It describes the what, why and how of using base rates to inform your decision making in the first 26 pages or so.

 

Then he looks at various base rates of various metrics such as growth, return on investments, and price increases etc.

 

As Kahneman says, remember the outside view, using base rates really help.

 

http://csinvesting.org/wp-content/uploads/2016/10/The-Base-Rate-View-by-Mauboussin.pdf

[netnet]

For those who don't know, base rate can be thought of as the underling percentages, growth, prospects, or success rate of something that closely matches the situation you are looking at. For example, how often the team with the best slugging percentage and ERA won the World Series.  (The reference class is not that the neither the Cubs nor the Indians have won the series in over 65 years!)

[Jurgis]

Since you seem to be lonely here, I'll +1 that I liked Base Rate View somewhat.

 

I and friends had a long discussion (to be continued perhaps) about "Thinking Fast and Slow".

 

My one big conclusion from the discussion (in)applicable to investing:

 

- You can possibly get outside view probabilities - not always but perhaps

- You can have inside view

- It's very hard to impossible to assign probabilities for inside view for various reasons (humans cannot think probabilistically, there's no prior data set for inside view, etc.)

- So that's why almost nobody combines outside view probabilities to inside view probabilities in their process - or produce probabilistic models for investing.

 

There are exceptions - somewhat. That's where people like Thorp make/made their money. Even he used mostly outside view + heuristics. I'm not sure he explicitly assigned inside view probabilities and did integration with outside view. Edit: when you can construct probabilistic models, they tend to work really well, like Thorp's friends/proteges who shoveled gains in Asian horse racing, etc.

[rawraw]

Since you seem to be lonely here, I'll +1 that I liked Base Rate View somewhat.

 

 

My one big conclusion from the discussion (in)applicable to investing:

 

- You can possibly get outside view probabilities - not always but perhaps

- You can have inside view

- It's very hard to impossible to assign probabilities for inside view for various reasons (humans cannot think probabilistically, there's no prior data set for inside view, etc.)

- So that's why almost nobody combines outside view probabilities to inside view probabilities in their process - or produce probabilistic models for investing.

 

Quant funds heavily rely on base rates.  I think to assume they aren't used in investing is very myopic

 

Almost all the research I've seen shows combining base rates with inside view always produces better forecasts.  Knowing what values typically are seems fundamental to the mean reversion many value investments rely upon.  Surprised someone would conclude that having a forecast for a company based in the reality of the industry/sector occurs by "nobody" -- why then are we focused on industry comps?   

[Jurgis]

I think you are misinterpreting my last statement. I did not say that outside view is not used. I said:

 

- So that's why almost nobody combines outside view probabilities to inside view probabilities in their process - or produce probabilistic models for investing.

 

So let's ask this: do you combine outside view probabilities with inside view probabilities in your process? If so, how do you calculate inside view probabilities?

[rawraw]

Yes, of course I do.  I've been trying to be very rigorous in my approach to thinking about the future, which is why I'm a fan of mauboussin. He has a similar approach to thinking.

 

Depends on the situation. In the banks I evaluate, I use the base rate of various financial performance metrics before adjusting it based on my inside view. And I Iook back in history to see how those traits sustained or changed in history when forecasting ratios increasing or decreasing.

 

This weekend I intend to look at the recently announced mergers in the market given interesting commentary by Levine this week. I'll use a base rate of deals closing before then adjusting the probability based on the inside view of the specific transaction.

 

I feel the above is very basic, so apologize if I answered the wrong question. But the base is the starting point from which I adjust a forecast based on inside knowledge.

[Jurgis]

If you can answer, how do you assign probabilities to the inside view? Maybe an example would help. :)

 

Thanks and good luck.

[vinod1]

After reading all of the 158 pages here is what I ended up learning:

 

Buying cheap and quality leads to good returns. He talks about inside view, outside view, HOLT.... and on and on. When it comes finally to examples it is really simply buying cheap. I think someone has written about it all the way back n 1934.

 

Netnet - Please do continue to share interesting research. Nothing against you. I am disappointed with Mauboussin. That he had to write so much to convey so little.

 

Vinod

[KCLarkin]

When it comes finally to examples it is really simply buying cheap. I think someone has written about it all the way back n 1934.

 

I don't think it was well-written or well-organized BUT there is a tremendous amount of insight hidden in the jargon in this report. I certainly don't think the takeaway should be simply buying cheap. His definition of cheap would not be a low-multiple stock.

 

[vinod1]

When it comes finally to examples it is really simply buying cheap. I think someone has written about it all the way back n 1934.

 

I don't think it was well-written or well-organized BUT there is a tremendous amount of insight hidden in the jargon in this report. I certainly don't think the takeaway should be simply buying cheap. His definition of cheap would not be a low-multiple stock.

 

My definition for cheap is also not low multiple stock, it is just as Buffett defines it - cheap as compared to IV, which incorporates growth, quality, etc.

 

The concept of outside view - just having base rates in mind is something I imagine every value investor incorporates when making an investment decision, it certainly is something that is very central to my own approach. Mean reversion which is what data indicates is the very essence of Graham's 1934 book.

 

What I did not see in this report is anything new that adds to what Graham and Buffett had already covered. I have a written investment approach - nothing fancy, pretty much what Buffett, Graham, Munger, Greenwald, Fisher, etc. have repeated over and over again and  a checklist - my own mistakes and few pointers from the same gang. So when I read an investment book or article, I am primarily looking for something new to add or modify in either the investment approach or checklist. After reading this report, I did not find one thing that I can add to either of these. What I find mildly useful are the two pages of base rate data on earnings growth. Nothing new, pretty much in line with my understanding but I did not have this data before.

 

See after all that is written, he gives examples, where quality is high and stock is cheap and returns are good. But the main thing is stock is cheap. Is that really a surprise? This is before even going into such things as expected returns over 90 day period.

 

Vinod

 

 

[KCLarkin]

When it comes finally to examples it is really simply buying cheap. I think someone has written about it all the way back n 1934.

 

I don't think it was well-written or well-organized BUT there is a tremendous amount of insight hidden in the jargon in this report. I certainly don't think the takeaway should be simply buying cheap. His definition of cheap would not be a low-multiple stock.

 

My definition for cheap is also not low multiple stock, it is just as Buffett defines it - cheap as compared to IV, which incorporates growth, quality, etc.

 

The concept of outside view - just having base rates in mind is something I imagine every value investor incorporates when making an investment decision, it certainly is something that is very central to my own approach. Mean reversion which is what data indicates is the very essence of Graham's 1934 book.

 

What I did not see in this report is anything new that adds to what Graham and Buffett had already covered. I have a written investment approach - nothing fancy, pretty much what Buffett, Graham, Munger, Greenwald, Fisher, etc. have repeated over and over again and  a checklist - my own mistakes and few pointers from the same gang. So when I read an investment book or article, I am primarily looking for something new to add or modify in either the investment approach or checklist. After reading this report, I did not find one thing that I can add to either of these. What I find mildly useful are the two pages of base rate data on earnings growth. Nothing new, pretty much in line with my understanding but I did not have this data before.

 

See after all that is written, he gives examples, where quality is high and stock is cheap and returns are good. But the main thing is stock is cheap. Is that really a surprise? This is before even going into such things as expected returns over 90 day period.

 

Vinod

 

Perhaps not useful to you but useful to many others. Especially to investors who build very detailed models with very unrealistic expectations.

[Sionnach]

I agree with other comments - I excitedly printed out this whole PDF only to find the majority of it was useless.

 

The important insight I gained from it was the importance of the outside view and having greater humility when it comes to the inside view.

 

This was helpful to me when looking at FOGO which has supposedly attractive economics (inside view) within a brutally competitive and low barrier-to-entry industry (outside view).

[rawraw]

If you can answer, how do you assign probabilities to the inside view? Maybe an example would help. :)

 

Thanks and good luck.

Just saw this.  I'd assume most assign probabilities to inside view qualitatively.  So when combined with base rates, you'd say there is a 45% chance of this happening.  But variable X, Y, Z change the percentage in my view by 10% to 55%.  My preference would be to get more detailed on the base rate.  For example, how does the base rate change for cases in the sample with variable X.  But sometimes the sample isn't large enough for that sort of refinement. 

 

I think the Base Rate book isn't supposed to be a new investing methodology.  I think it is more supposed to be a helpful way to keep multi-year forecasts in check, by understanding the distribution of differing variables over time.  I think the purpose is very mechanical in nature; more focused on improving forecasting than improving the thinking behind investing.

[rawraw]

If you can answer, how do you assign probabilities to the inside view? Maybe an example would help. :)

 

Thanks and good luck.

Just saw this.  I'd assume most assign probabilities to inside view qualitatively.  So when combined with base rates, you'd say there is a 45% chance of this happening.  But variable X, Y, Z change the percentage in my view by 10% to 55%.  My preference would be to get more detailed on the base rate.  For example, how does the base rate change for cases in the sample with variable X.  But sometimes the sample isn't large enough for that sort of refinement.  And I always like to check how sensitive my forecast is to changing probabilities.

 

I think the Base Rate book isn't supposed to be a new investing methodology.  I think it is more supposed to be a helpful way to keep multi-year forecasts in check, by understanding the distribution of differing variables over time.  I think the purpose is very mechanical in nature; more focused on improving forecasting than improving the thinking behind investing.

[SharperDingaan]

You might want to look at the concept a little more critically; then how it's applied.

 

This is a probability weighted FORECAST. The actuals will be different - to find out how different, simply go back & compare.

A 5% forecast error in Yr 1 (pretty good), compounds to 15%+ (unusable) after 3 years. It's a forecast at < 1yr, & a 'feel' at >1 yr.

 

There's nothing wrong with 'feel' as long as it is supportable, & can withstand outside scrutiny. But it is not a guarantee that XYZ will occur.

Hence you invest 'with' todays forecast for 1 yr or less, & 'against' it > 1 Yr (Talebs antifragility); it's application of the Gladwell 'fast & slow'.

Alternatively, reforecast at 1 yr.

 

Most folks are far better at 'feel', primarily because its closer to real life. I 'feel' like something wants to eat me, so I stay in the cave.

But you cant sell 'feel', you can sell 'forecast' - talking heads do so daily.

In NA we don't trust 'feel', in a Russia or China it keeps you alive.

 

SD

 

 

[rawraw]

You might want to look at the concept a little more critically; then how it's applied.

 

This is a probability weighted FORECAST. The actuals will be different - to find out how different, simply go back & compare.

A 5% forecast error in Yr 1 (pretty good), compounds to 15%+ (unusable) after 3 years. It's a forecast at < 1yr, & a 'feel' at >1 yr.

 

There's nothing wrong with 'feel' as long as it is supportable, & can withstand outside scrutiny. But it is not a guarantee that XYZ will occur.

Hence you invest 'with' todays forecast for 1 yr or less, & 'against' it > 1 Yr (Talebs antifragility); it's application of the Gladwell 'fast & slow'.

Alternatively, reforecast at 1 yr.

 

Most folks are far better at 'feel', primarily because its closer to real life. I 'feel' like something wants to eat me, so I stay in the cave.

But you cant sell 'feel', you can sell 'forecast' - talking heads do so daily.

In NA we don't trust 'feel', in a Russia or China it keeps you alive.

 

SD

Thanks for the reply! Can you please specify "this"  in your opening sentence. I really have no idea what you are specifically addressing