Investing in biotech stocks - My experience so far

Username999

BrockStoker said:

Why? Because It's easier not to have to log in all the time, not to mention less of a security risk.

I have nothing to prove, and I don't have to show actual screenshots of my portfolio, but here you go anyway:

Fair dues, well done.

BrockStoker

thegentleway said:

BrockStoker said:

csgohan4 said:

thegentleway said:

BrockStoker said:

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

Well, if I did OK with 22/25, that suggests I'm getting it right 88% of the time I guess?

That is the most common answer but it is incorrect as it ignores the base rate (1/25). There are only two conditions in which you will pick a stock:

1. You correctly identify a performing stock. Probability is 22/25 x 1/25 = 3.52%

2. You incorrectly identify a non performing stock. Probability is 3/25 x 24/25 = 11.52%

You will therefore pick a stock 15.04% of the time (3.52% + 11.52%). The probability that you picked a performing stock is therefore 23.40% (3.52% / 15.04%).

Even with your magical ability to recognise performing stocks 22 out of 25 times you are going to pick a performing stock less that one quarter of the time.

but his portfolio is undeniably doing well and it wasn't like he chose a stock at random for the odds to kick in. 
His risk appetite will likely be high and understandably has to be in the biotech sector. 

Yes, I'm not exactly sure what thegentleway is trying to say (or prove?) there - is it that "you can't pick 22/25 stocks that are winners"?

Well, theory is fine, but my experience is "real world", and isn't necessarily going to align with theory. Of course, I could get a very different result if I tried to repeat my "experiment", but it doesn't change what I got on the first run.

You asked me what I thought... which is that you are ignoring the base rate. I have not argued with your unrealistic ability to recognise a performing stock 22/25. I’ve explained that this yields a 23.4% probability of picking a winner in real life. It’s basic stats but I appreciate it’s unintuitive hence why I showed the workings and tried to explain it. Ironically it’s very important for medicine; it’s one or the reasons we have control groups. 

Perhaps it would help if you considered a made up scenario where there is only 1 performing stock. Surely you can see that the odds of you picking that one stock is incredibly small even with a 22/25 ability? That’s the effect of the underlying base rate.

I'm sorry thegentleway, but you've completely lost me. What do you mean by "base rate"? Base rate of what?

I don't have a good head for math/figures I should warn you, so showing your calculations doesn't really help.

Either way, your input is appreciated.

BrockStoker

Username999 said:

BrockStoker said:

Why? Because It's easier not to have to log in all the time, not to mention less of a security risk.

I have nothing to prove, and I don't have to show actual screenshots of my portfolio, but here you go anyway:

Fair dues, well done.

Thank you.

To be fair, if when I started this venture I could have gone forward in time, and see where I am today, I might have had trouble believing it too!

thegentleway

BrockStoker said:

thegentleway said:

BrockStoker said:

csgohan4 said:

thegentleway said:

BrockStoker said:

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

Well, if I did OK with 22/25, that suggests I'm getting it right 88% of the time I guess?

That is the most common answer but it is incorrect as it ignores the base rate (1/25). There are only two conditions in which you will pick a stock:

1. You correctly identify a performing stock. Probability is 22/25 x 1/25 = 3.52%

2. You incorrectly identify a non performing stock. Probability is 3/25 x 24/25 = 11.52%

You will therefore pick a stock 15.04% of the time (3.52% + 11.52%). The probability that you picked a performing stock is therefore 23.40% (3.52% / 15.04%).

Even with your magical ability to recognise performing stocks 22 out of 25 times you are going to pick a performing stock less that one quarter of the time.

but his portfolio is undeniably doing well and it wasn't like he chose a stock at random for the odds to kick in. 
His risk appetite will likely be high and understandably has to be in the biotech sector. 

Yes, I'm not exactly sure what thegentleway is trying to say (or prove?) there - is it that "you can't pick 22/25 stocks that are winners"?

Well, theory is fine, but my experience is "real world", and isn't necessarily going to align with theory. Of course, I could get a very different result if I tried to repeat my "experiment", but it doesn't change what I got on the first run.

You asked me what I thought... which is that you are ignoring the base rate. I have not argued with your unrealistic ability to recognise a performing stock 22/25. I’ve explained that this yields a 23.4% probability of picking a winner in real life. It’s basic stats but I appreciate it’s unintuitive hence why I showed the workings and tried to explain it. Ironically it’s very important for medicine; it’s one or the reasons we have control groups. 

Perhaps it would help if you considered a made up scenario where there is only 1 performing stock. Surely you can see that the odds of you picking that one stock is incredibly small even with a 22/25 ability? That’s the effect of the underlying base rate.

I'm sorry thegentleway, but you've completely lost me. What do you mean by "base rate"? Base rate of what?

I don't have a good head for math/figures I should warn you, so showing your calculations doesn't really help.

Either way, your input is appreciated. 

The base rate is the prior probability: https://en.m.wikipedia.org/wiki/Base_rate
In this instance the base rate is the probability of any stock performing well. Due to it being so low, this underlying rate has a huge effect on your odds of picking a winner.
To be fair, a lot of people that are good at maths still struggle with stats. Another way to think about it is:
How many companies have you looked at? Have you dismissed 24 for every one you picked?
It would be interesting to track the dismissed stock performance for comparison. Or you could track some random picks as a control group.

Linton

thegentleway said:

BrockStoker said:

thegentleway said:

BrockStoker said:

csgohan4 said:

thegentleway said:

BrockStoker said:

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

Well, if I did OK with 22/25, that suggests I'm getting it right 88% of the time I guess?

That is the most common answer but it is incorrect as it ignores the base rate (1/25). There are only two conditions in which you will pick a stock:

1. You correctly identify a performing stock. Probability is 22/25 x 1/25 = 3.52%

2. You incorrectly identify a non performing stock. Probability is 3/25 x 24/25 = 11.52%

You will therefore pick a stock 15.04% of the time (3.52% + 11.52%). The probability that you picked a performing stock is therefore 23.40% (3.52% / 15.04%).

Even with your magical ability to recognise performing stocks 22 out of 25 times you are going to pick a performing stock less that one quarter of the time.

but his portfolio is undeniably doing well and it wasn't like he chose a stock at random for the odds to kick in. 
His risk appetite will likely be high and understandably has to be in the biotech sector. 

Yes, I'm not exactly sure what thegentleway is trying to say (or prove?) there - is it that "you can't pick 22/25 stocks that are winners"?

Well, theory is fine, but my experience is "real world", and isn't necessarily going to align with theory. Of course, I could get a very different result if I tried to repeat my "experiment", but it doesn't change what I got on the first run.

You asked me what I thought... which is that you are ignoring the base rate. I have not argued with your unrealistic ability to recognise a performing stock 22/25. I’ve explained that this yields a 23.4% probability of picking a winner in real life. It’s basic stats but I appreciate it’s unintuitive hence why I showed the workings and tried to explain it. Ironically it’s very important for medicine; it’s one or the reasons we have control groups. 

Perhaps it would help if you considered a made up scenario where there is only 1 performing stock. Surely you can see that the odds of you picking that one stock is incredibly small even with a 22/25 ability? That’s the effect of the underlying base rate.

I'm sorry thegentleway, but you've completely lost me. What do you mean by "base rate"? Base rate of what?

I don't have a good head for math/figures I should warn you, so showing your calculations doesn't really help.

Either way, your input is appreciated. 

The base rate is the prior probability: https://en.m.wikipedia.org/wiki/Base_rate
In this instance the base rate is the probability of any stock performing well. Due to it being so low, this underlying rate has a huge effect on your odds of picking a winner.
To be fair, a lot of people that are good at maths still struggle with stats. Another way to think about it is:
How many companies have you looked at? Have you dismissed 24 for every one you picked?
It would be interesting to track the dismissed stock performance for comparison. Or you could track some random picks as a control group.

Your maths seems right to me though I havent checked the details. However I have issues with your application in this case.....

Firstly for those who have not met the issue:  
Suppose there is a fatal illness which affects 1% of the population.  There is a screening test which amazingly is 90% correct when the disease is present    However it also has a 10% chance of falsely finding the disease is present when it isnt.  You take the test and get a positive - should you be rushing out to write your will?

Out of a set of 100 screens on the general population the test will find 0.9 people with the illness but give a false positive to 10, a total of 11 (forgetting the 0.1 person) positives.  So the chance that you actually have the disease is only approx 9% even though the test is extremely accurate.  The results of the screen would have to be confirmed with very accurate tests which will show in 91% of cases that you dont have the disease - the screening is not specific enough for individual cases.

Now my problem with the interpretation here: 
We are seeing the confirmed results - the OP states the 23 out of 25 of his choices actually are good performers, not false positives in the selection algorithm.  Using the medical example I gave above, the detailed post-screening checks are actually finding that the disease is usually present.  Therefore the statistical assumptions must be wrong.

thegentleway

Linton said:

thegentleway said:

BrockStoker said:

thegentleway said:

BrockStoker said:

csgohan4 said:

thegentleway said:

BrockStoker said:

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

Well, if I did OK with 22/25, that suggests I'm getting it right 88% of the time I guess?

That is the most common answer but it is incorrect as it ignores the base rate (1/25). There are only two conditions in which you will pick a stock:

1. You correctly identify a performing stock. Probability is 22/25 x 1/25 = 3.52%

2. You incorrectly identify a non performing stock. Probability is 3/25 x 24/25 = 11.52%

You will therefore pick a stock 15.04% of the time (3.52% + 11.52%). The probability that you picked a performing stock is therefore 23.40% (3.52% / 15.04%).

Even with your magical ability to recognise performing stocks 22 out of 25 times you are going to pick a performing stock less that one quarter of the time.

but his portfolio is undeniably doing well and it wasn't like he chose a stock at random for the odds to kick in. 
His risk appetite will likely be high and understandably has to be in the biotech sector. 

Yes, I'm not exactly sure what thegentleway is trying to say (or prove?) there - is it that "you can't pick 22/25 stocks that are winners"?

Well, theory is fine, but my experience is "real world", and isn't necessarily going to align with theory. Of course, I could get a very different result if I tried to repeat my "experiment", but it doesn't change what I got on the first run.

You asked me what I thought... which is that you are ignoring the base rate. I have not argued with your unrealistic ability to recognise a performing stock 22/25. I’ve explained that this yields a 23.4% probability of picking a winner in real life. It’s basic stats but I appreciate it’s unintuitive hence why I showed the workings and tried to explain it. Ironically it’s very important for medicine; it’s one or the reasons we have control groups. 

Perhaps it would help if you considered a made up scenario where there is only 1 performing stock. Surely you can see that the odds of you picking that one stock is incredibly small even with a 22/25 ability? That’s the effect of the underlying base rate.

I'm sorry thegentleway, but you've completely lost me. What do you mean by "base rate"? Base rate of what?

I don't have a good head for math/figures I should warn you, so showing your calculations doesn't really help.

Either way, your input is appreciated. 

The base rate is the prior probability: https://en.m.wikipedia.org/wiki/Base_rate
In this instance the base rate is the probability of any stock performing well. Due to it being so low, this underlying rate has a huge effect on your odds of picking a winner.
To be fair, a lot of people that are good at maths still struggle with stats. Another way to think about it is:
How many companies have you looked at? Have you dismissed 24 for every one you picked?
It would be interesting to track the dismissed stock performance for comparison. Or you could track some random picks as a control group.

Your maths seems right to me though I havent checked the details. However I have issues with your application in this case.....

Firstly for those who have not met the issue:  
Suppose there is a fatal illness which affects 1% of the population.  There is a screening test which amazingly is 90% correct when the disease is present    However it also has a 10% chance of falsely finding the disease is present when it isnt.  You take the test and get a positive - should you be rushing out to write your will?

Out of a set of 100 screens on the general population the test will find 0.9 people with the illness but give a false positive to 10, a total of 11 (forgetting the 0.1 person) positives.  So the chance that you actually have the disease is only approx 9% even though the test is extremely accurate.  The results of the screen would have to be confirmed with very accurate tests which will show in 91% of cases that you dont have the disease - the screening is not specific enough for individual cases.

Indeed. There is a 91.67% chance that you do not have the disease if your screen test is positive. This is precisely what I was trying to explain.

Linton said:

Now my problem with the interpretation here: 
We are seeing the confirmed results - the OP states the 23 out of 25 of his choices actually are good performers, not false positives in the selection algorithm.  Using the medical example I gave above, the detailed post-screening checks are actually finding that the disease is usually present.  Therefore the statistical assumptions must be wrong.

I answered the question I asked, which was

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

The probability is 23.40%

You raise a valid point though and perhaps the question I should have asked was:

2. Given a track record of 22/25 performing stocks, how good are you at picking stocks assuming that only 1/25 stock perform well?

The answer is 99.44% - this means that if you give @BrockStocker 177 stocks, he will correctly identify 176 of them and only make 1 mistake

Linton

thegentleway said:

Linton said:

thegentleway said:

BrockStoker said:

thegentleway said:

BrockStoker said:

csgohan4 said:

thegentleway said:

BrockStoker said:

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

Well, if I did OK with 22/25, that suggests I'm getting it right 88% of the time I guess?

That is the most common answer but it is incorrect as it ignores the base rate (1/25). There are only two conditions in which you will pick a stock:

1. You correctly identify a performing stock. Probability is 22/25 x 1/25 = 3.52%

2. You incorrectly identify a non performing stock. Probability is 3/25 x 24/25 = 11.52%

You will therefore pick a stock 15.04% of the time (3.52% + 11.52%). The probability that you picked a performing stock is therefore 23.40% (3.52% / 15.04%).

Even with your magical ability to recognise performing stocks 22 out of 25 times you are going to pick a performing stock less that one quarter of the time.

but his portfolio is undeniably doing well and it wasn't like he chose a stock at random for the odds to kick in. 
His risk appetite will likely be high and understandably has to be in the biotech sector. 

Yes, I'm not exactly sure what thegentleway is trying to say (or prove?) there - is it that "you can't pick 22/25 stocks that are winners"?

Well, theory is fine, but my experience is "real world", and isn't necessarily going to align with theory. Of course, I could get a very different result if I tried to repeat my "experiment", but it doesn't change what I got on the first run.

You asked me what I thought... which is that you are ignoring the base rate. I have not argued with your unrealistic ability to recognise a performing stock 22/25. I’ve explained that this yields a 23.4% probability of picking a winner in real life. It’s basic stats but I appreciate it’s unintuitive hence why I showed the workings and tried to explain it. Ironically it’s very important for medicine; it’s one or the reasons we have control groups. 

Perhaps it would help if you considered a made up scenario where there is only 1 performing stock. Surely you can see that the odds of you picking that one stock is incredibly small even with a 22/25 ability? That’s the effect of the underlying base rate.

I'm sorry thegentleway, but you've completely lost me. What do you mean by "base rate"? Base rate of what?

I don't have a good head for math/figures I should warn you, so showing your calculations doesn't really help.

Either way, your input is appreciated. 

The base rate is the prior probability: https://en.m.wikipedia.org/wiki/Base_rate
In this instance the base rate is the probability of any stock performing well. Due to it being so low, this underlying rate has a huge effect on your odds of picking a winner.
To be fair, a lot of people that are good at maths still struggle with stats. Another way to think about it is:
How many companies have you looked at? Have you dismissed 24 for every one you picked?
It would be interesting to track the dismissed stock performance for comparison. Or you could track some random picks as a control group.

Your maths seems right to me though I havent checked the details. However I have issues with your application in this case.....

Firstly for those who have not met the issue:  
Suppose there is a fatal illness which affects 1% of the population.  There is a screening test which amazingly is 90% correct when the disease is present    However it also has a 10% chance of falsely finding the disease is present when it isnt.  You take the test and get a positive - should you be rushing out to write your will?

Out of a set of 100 screens on the general population the test will find 0.9 people with the illness but give a false positive to 10, a total of 11 (forgetting the 0.1 person) positives.  So the chance that you actually have the disease is only approx 9% even though the test is extremely accurate.  The results of the screen would have to be confirmed with very accurate tests which will show in 91% of cases that you dont have the disease - the screening is not specific enough for individual cases.

Indeed. There is a 91.67% chance that you do not have the disease if your screen test is positive. This is precisely what I was trying to explain.

Linton said:

Now my problem with the interpretation here: 
We are seeing the confirmed results - the OP states the 23 out of 25 of his choices actually are good performers, not false positives in the selection algorithm.  Using the medical example I gave above, the detailed post-screening checks are actually finding that the disease is usually present.  Therefore the statistical assumptions must be wrong.

I answered the question I asked, which was

1. Assuming that 1/25 stock perform well and you have the ability to recognise a winning stock 22/25 of the time, what do you estimate the probability of you picking a performing stock is?

The probability is 23.40%

You raise a valid point though and perhaps the question I should have asked was:

2. Given a track record of 22/25 performing stocks, how good are you at picking stocks assuming that only 1/25 stock perform well?

The answer is 99.44% - this means that if you give @BrockStocker 177 stocks, he will correctly identify 176 of them and only make 1 mistake

From where do you get you figure of 1/25 stock perform well?  And if you can does not performing well does not mean performing badly, many may hang around for years not doing anything in particular.  Your calculations must assume a probablility of false positives.  Can you justify that?  In any case the nature of this market makes it likely that one stock performing very well will pay for many stocks performing very badly - you cant lose more than the initial cost of  purchase but you can make a lot more.

Surely it is better to make your theories and set up your models on the basis of real life experiments rather han decry real life experiments because you believe you already know the answer from your theories.

Username999

Stats, what a great way to derail a thread!

BrockStoker

Thank you thegentleway, and Linton for explaining/contributing.

I think I more or less understand now.

I'm inclined to agree with Linton here. I think statistics are more useful when applied to large samples/lots of data, which is not really the case here. Also, companies are "organic" entities, which grow/evolve uniquely on a case by case basis, making them less predictable as far as statistics go, especially so if the sample is small.

One other factor which I think has played a large part, and that I neglected to mention in my most recent posts here (although I have mentioned it previously), is that many of the companies I chose, were previously highlighted by SA analysts here:

https://seekingalpha.com/stock-ideas/healthcare

So effectively, the companies have been screened once to pick the best, and then I've come along, and picked what I believe to be the "best of the best". I have said in the past that I believe many SA analysts do a pretty good job when it comes to biotech (no idea about the other sectors they cover).

BrockStoker

Username999 said:

Stats, what a great way to derail a thread!

No biggie. It's all good, and I'm sure I can get this thread back on track without too much effort. I'll give it a try now anyway:

There was some good news regarding Evofem today - at least I find it encouraging that I'm not the only one who thinks the future is bright.

Evofem Biosciences Wins Business Intelligence Group's BIG Innovation Award 2021

https://finance.yahoo.com/news/evofem-biosciences-wins-business-intelligence-140500486.html

Here is my reasoning why I think their product Phexxi will be successful:

- Many women don't like/won't use hormonal birth control, and find other current options to be less than satisfactory (I know at least one personally).

- When it was trialed, a high proportion liked it, and said it increased pleasure/satisfaction during intercourse, which is something no other form of birth control can claim to do.

- It's been FDA approved, meaning it's effective enough to be considered as comparably effective alternative option compared to existing options.

- Initial sales have been better than expected. So if Evofem was expecting to be successful, a start that beats expectations is a very bullish sign IMHO.

Obviously it will take time for word to spread with a product like this, but the seeds have been sown, and sooner or later I would expect it to start to snowball. It only needs to take a small chunk of the overall market to be successful, but I think it may go on to take a significant chunk of the market. With no alternative alternatives out there I think Evofem is in a very sweet spot. At least one medical insurance company already offers it, and I expect the others to follow.

One other thing to note is that Evofem's CEO is very passionate about what she is doing, and this bodes well once again. Anyone who is not convinced needs to see her in action:

https://www.youtube.com/watch?v=2DUzLMQvbGI