Q2: Not much is said in the book about his background. Keying in on his education as an engineer, he paid attention to how components of his thesis might interact. He looked for secondary affects not just the immediate effects a of change. He liked to know how things worked and was curious. Perhaps his education caused him to lean more towards facts, numbers, patterns and some technical analysis. He was also very orderly and structured, especially in how he managed his portfolio reviews.
Q3: Nokia, a messy conglomerate with a hidden gem. A single division, Mobira, had sales growing ~50% a year. Loss from the other divisions were hiding the profitably Mobira. Once he found out the other divisions were going to be shed, he bought in while the company was still undervalued.
Q4: Polly Beck, the conglomerate that got more and more complex and had opaque accounting. Maybe he did, but he did not mention investigation into how the additional businesses (which were widely different from the starting business) were affecting the company.
Q5: I studied 2 that I do not believe Bolton would go for. Reporting here because it was still a good exercise to think through.
DG -- This is not as cheap as it was a few quarters ago; might have been interesting earlier though. Proven CEO brought back in to fix operational problems. Plan showing signs of improvement: reduce theft, simplify inventory, improve labor conditions, renovate stores to improve customer experience. Margins are slowly improving, inventory is coming down, same store sales growing and they are gaining market share (rural markets are their target and are largely underserved).
PTON-- possible super small initial position, but there are big challenges. Also, after looking into this, I ran across a post of someone trying to sell their Peloton that they had only used 2x--I'm not thinking that there is going to be a 'turn around'. Peloton sells premium fitness hardware at near-cost to onboard users into a high-margin recurring subscription ecosystem. They have pricing power and high switching cost for existing client base--if they can keep old users and get new users to buy the hardware. Cost cutting has improved gross margins and slightly improved operating margins. Attempting to branch into commercial business (hotels/gyms) and to integrate AI personalization. Execution of these is yet to show benefits. If hardware sales continue to decline, users won't get sucked into the subscription in the first place. Also, the CFO has recently left.
Q6:
Find recovery or takeover candidate stocks that meet Anthony Boltons criteria:
Strengthening financials
Fixable business problems
Low institutional ownership
Minimal broker enthusiasm
Poor recent share performance
For each stock found:
1) Analyze management performance. Assuming management has a clear turnaround plan, are they following it in measurable ways? Are they doing 'little things' better? Are they beginning to perform in line or better than competitors?
2) How does the company make money? How does it work?
3) What is being assumed in the current price? What key factors are already built in? How plausible and probable are these key factors?
3) Establish the downside at current price: Assume you are looking backwards from a 50% decline in current stock price and create a plausible scenario in which this happens.
Question 2: What about Bolton’s background and circumstances made his approach the right one for him?
Anthony Boulton is an English professional investor with an academic background. He went to Trinity College, Cambridge, as an undergraduate – famous for its music – and he now composes classical music in his retirement.
His whole approach reflects that: structured, organised, thorough, with an analytical process-driven system. Over the top of this sits an intellectual curiosity and questioning, which enables him to take a less conventional approach and take calculated risks to achieve a superior outcome.
He has a checklist-based approach, which I like because it improves the odds of a good outcome and removes judgement and "gut feel" from the places where it is less helpful.
For example
Do I have an investment thesis that I can easily explain? (From Peter Lynch – I like people who listen to others and credit them.)
Does the company have a clear, strong business franchise? Is the business model robust?
Does the company have a weak balance sheet? Does it pass the Altman Z score? If not, it's usually best avoided.
Always follow the cash, and make sure the cash flow backs up the story.
Does he trust the management? "Someone who has let down or disappointed investors once is more likely to do it again."
In all these cases he is not completely black and white, but he specifically says in some cases that he would have had a better outcome if he had been.
Question 3: What are your favourite Bolton investments? Why?
Cairn Energy: here the strengths were good management in an area where trust is vital. Layered on top was a good business model identified by that management of balancing low-risk cash flow-generating assets with big stakes in higher-risk exploration interests, which, if they succeed, would make shareholders a lot of money. The cash flow limits the downside risk and pays for the bets. A 7M investment yielded a £3bn outcome and pushed Cairn into the FTSE 100, making it a 100 bagger for its original investors. A great example of most of his principles in action.
Question 4: What are your least favourite Bolton investments? Why?
Autonomy. He bought this because of a tip from a colleague. He does not say so, but I don't think he understood what the company did. I say this because I never met anyone who could tell me convincingly how they made money or what they did. Words like "mining", "organising unstructured company proprietary data", and "helping drive insights" were bandied about along with astonishing growth, but details of how they did this? No one ever seemed to know, but the word "proprietary" was used a lot, I remember. He does admit that the only smart thing he did was to sell before the bubble fully burst. Rather depressingly, it was one of his best performers. Sometimes it's more important to be lucky than good, though you might have to put a value investor on a rack to get them to admit that...
Question 5: What are 1 or 2 stocks that Anthony Bolton might find attractive in the current environment?
I have picked Sage Group PLC.
Sage is an accounting and payroll software company based in the UK, but its business is international, with the UK only making up 20% of its market. It offers a series of packages, from a small company using Sage Line 50, with upgrades to multinational, multicurrency support for global companies. Like all software companies, it has been marked down by the indiscriminate sell-off due to the perceived AI threat. It is down from £13 per share to £8.40 in the last year.
I think this share would have appealed to Anthony Bolton. He used a range of valuation methods, and one he specifically mentions is historical PE. Currently it has a PE of 20. The last year you could buy it cheaper than this was 2018, then back to 2014 for the next opportunity, when it traded around a PE of 18. If it grows in line with broker forecasts, it has an FPE of 16. Its metrics are excellent; ROCE, CROCE and EBIT margin are all over 20%, all of which he mentions as hallmarks of a good company. It grows around 9% a year, and it's forecast to do that for the next 3 years. It has paid a dividend for 34 years.
He would have known how deeply entrenched it was in its customers and how many cheap alternatives have tried to displace it over the years, and how skilled it has been in defending itself. I think he would have thought the danger of AI modest because:
Companies don't choose accounting software on cost; they choose it for appropriateness, support, longevity, ecosystem and scalability. No one ever changes their accounting system without very good reason; it's expensive, risky, and rarely produces any benefit. It is usually done to avoid the pain of continuing to live with a system that the company has outgrown or is no longer adequately supported.
Every accountancy practice is familiar with Sage, so support is easily available, and there is no learning curve with your accountant to become familiar with an unknown system.
The packages are scalable, so moving up the ladder as the company evolves is relatively painless (no accounting change is without pain).
There is an ecosystem of support software agents that can build interfaces to other proprietary software and provide tailored software support.
An AI-based system would have to replicate all of this, and even then many companies would only switch when they had to, even if the new system was virtually free, because accounting software is a small cost to most businesses. The pain of transitioning systems is huge, the benefits are zero (accounts are accounts), and accounting disasters are the quickest way to bankruptcy of all. I remember a very successful listed consultancy company in the early 2000s that very nearly went bust when its accounting system was unable to raise invoices for more than a month due to software issues. Its share price did the most spectacular V-shaped drop and recovery, and lots of the company employees bought shares on the dip because they knew it was a sure thing.
Lastly, Sage has proved adept at buying promising competitors and either developing their products or killing them. Never underestimate what a good company with plenty of cash can do to stamp on competitors when they look threatening but are still small. Where it can't do that, it copies them: Xero has been an excellent cloud-based competitor for 20 years, and Sage responded by moving to the cloud, improving its offering and making it easy for small businesses to transition from Xero to Sage as their business grows and needs more features and integration. I think that Anthony Bolton would have appreciated all these kinds of institutional details which make general scares great buying opportunities if the scare does not really apply to the company under consideration.
Question 6: Come up with an AI prompt based on Bolton’s approach.
"As a financial analyst, make a list of companies in the UK FTSE 250 with strong fundamentals that have been heavily sold recently. List them in order of their likely recovery with reasons for whether they might recover or not."
On a scale of 1 (lowest) to 10 (highest), what are the degrees of difficulty in answering the question of how AI will affect Software companies in general and Sage Group in particular?
Like so many questions about the future, this is hard and uncertain, so let's give it an 8. But as investors we are used to this. We know that we have no idea where the stock market will be in a year's time or what interest rates will be in 2 years' time. I'd give both of those an 8 score as well. So can we say that we know the outcome? Definitely not. But is there anything useful we can say about it based on experience? Here, as in so many other areas of the market, I think that the answer is a yes, but always with the humility of bitter experience, I could easily be wrong.
I'm going to break it down into some separate questions to try and give a sensible response.
How good is the current type of AI going to get in the near future? Will it be able to give a step change in current performance rather than a steady evolution of its current capabilities?
I'm going to give a cautious no on this. The approach is a continuing increase in computing power and post-processing modules and already shows signs of diminishing returns. This approach of an LLM is always going to weigh towards correlation rather than a true model of the real world, because it is built on correlation, which is by its nature error-prone.
Counterfactual. Already the models are focusing on checking and rechecking their own work for accuracy, and new modules with more bespoke abilities in areas where LLMs are weak, like arithmetic, are being built onto the front of LLMs. Might progress here fully address the weaknesses and allow much more accurate "thinking" to emerge?
Is there evidence that a new and stronger, more general intelligence AI is likely to emerge from the current models which would change the game?
Not that I can see.
I have neither seen nor read nor had any experience of any evidence of a breakthrough in this area, and I have little theoretical reason to think that the current approach is showing any convergence with human intelligence. And if it does, it won't just be software stocks that will have to rethink everything; pretty much the entire planet will need to, including every company, organisation, government and individual.
Counterfactual. Apart from the previous counterfactual, none.
How effective will AI, which is undoubtedly improved but still based on current technology, be in helping programme new platforms? Undoubtedly it will help speed programming and help less skilled programmers do more ambitious things.
Counterfactual. There are questions around whether programming complex systems will actually be speeded up if the code written is of poor quality and the skilled programmer has to spend significant time correcting it, and the unskilled simply misses the problem and writes code which either fails under different circumstances or is very difficult to follow and debug – similar to the so-called spaghetti code written in the early days when software writing programmes were much more freeform. However, I think this is unlikely to be a zero-sum game, and the result will still be quicker and therefore cheaper.
Is AI proprietary and a "winner takes all" technology?
This is, to me, a critical point. ChatGPT, Anthropic and Google are all producing models with very similar abilities, and the lead is passing from one to the other rapidly. There is no frontrunner, and therefore no "winner takes all".
Unless this changes, AI models could easily become commoditised.
Counterfactual. This is linked to whether there is a breakthrough in model performance. Maybe one firm might achieve this and the others not, but it's hard to see this as a likely outcome. There is too much money available to poach the talent and find out what has been done. China has produced very capable models with far fewer computing resources, showing what can be done.
Given that programming becomes easier, does this cost advantage alone mean a displacement of current software companies by cheaper competitors?
I'm going to give a reasonably certain no to this. There have been plenty of technological breakthroughs, and plenty of companies that have grown large and powerful on the back of them, but the shape of this is more driven by the business model than the technology. Meta, Google and similar platforms built businesses by giving free stuff that people wanted to consumers and sold the targeted data to advertisers. This squeezed the whole remaining advertising market and damaged the incumbents there indirectly, rather than by head-to-head competition. The incumbent software companies sell expensive stuff to other corporations with complex needs and embedded systems. Different business model, and probably a different outcome. Most software companies have a very in-depth understanding of their customers' needs and are in a prime position to develop AI solutions for existing customers while still supporting their complex legacy needs. This particularly holds if one AI model is much the same as another in terms of power and reliability.
I see the most likely outcome is that the strong and capable software companies will use AI both to reduce their costs and sell new services based on AI to their customers, leveraging their knowledge of their customers' needs and trust in their solutions, and outsiders will have a long battle to win market share. The worst-case scenario is that their pricing is compromised. The weaker ones, with products that are easy to reproduce and customers not so tied in, are much more vulnerable. Consumer-facing software is the most vulnerable of all, where it is much easier for individuals to switch.
Do I think these arguments apply to Sage? Yes. It will become ever easier to build good accountancy "add-on" packages to existing software, which is something that has happened for many years, so it is not a new problem, but it will grow. It will also be easier for startups to program new accounting AI native software, whose benefits might easily be an ability to do simple accounts without the help of a bookkeeper, which would give a nice sales point to new small traders. More pressure and more competition, but more opportunities as well for acquisition, upselling of services, penetration into smaller markets using lower-cost programming and more rapid development and integration of existing products.
Q1, a contrarian - against the crown investing, owned limited (max 50) issues - most funds own many more securities
based the investment in a company on a thesis - which if violated, would give a reason to re-evaluate or sell
always looked at portfolio as if he was building for the first time (from-scratch) as opposed to holding on just because you own it
used the importance of valuing management as opposed to just relying on numbers to evaluated an investment, he needed to separate the spin as opposed to facts from management
would reduce outsized positions in bear markets
he "dug deeper" , searched widely and kept an open mind regarding the investments he made,
he needed the company to have a competitive advantage or moat
a company also needed to be a cash generator
he looked at under owned sectors and companies
he also looked for underfollowed companies
he used technical analysis to evaluate the results he got from his fundamental analysis
he incorporated market sentiment into his evaluations
Q2. He graduated with a degree in engineering. This gave his a different perspective than those who had graduated from business schools in finance. His first jobs were in investment management as opposed to the more popular investment banking field. He was able to see the operations of the investment profession from a first hand basis a watch it evolve over his career.
Q3. Cellnet and Security Services as they were his largest holdings over a number of years and were undervalued.
Q4. Parkfield and Sportingbet - He was referred Parkfield by a broker prior to meeting the CEO. Financial difficulties were later disclosed, probably a result of financing acquisitions. Sports betting - a bad investment.
Q5. Landstar - a slightly undervalued mid-sized transportation company.
Q6. Evaluate both a European and South Asian indices for mid-size companies using Bolton's criteria for evaluating companies, management, shares, sentiment, risks, financials, valuations, takeovers and special situations.
Q1: https://docs.google.com/document/d/e/2PACX-1vQb4ofaY3GZmlP9loRjEskcJmVBUrpL-7EBqDsnHHFSiMQEfXRpGK091KVK_t2LKQp-ZtFRYxBDAwYM/pub
Q2: Not much is said in the book about his background. Keying in on his education as an engineer, he paid attention to how components of his thesis might interact. He looked for secondary affects not just the immediate effects a of change. He liked to know how things worked and was curious. Perhaps his education caused him to lean more towards facts, numbers, patterns and some technical analysis. He was also very orderly and structured, especially in how he managed his portfolio reviews.
Q3: Nokia, a messy conglomerate with a hidden gem. A single division, Mobira, had sales growing ~50% a year. Loss from the other divisions were hiding the profitably Mobira. Once he found out the other divisions were going to be shed, he bought in while the company was still undervalued.
Q4: Polly Beck, the conglomerate that got more and more complex and had opaque accounting. Maybe he did, but he did not mention investigation into how the additional businesses (which were widely different from the starting business) were affecting the company.
Q5: I studied 2 that I do not believe Bolton would go for. Reporting here because it was still a good exercise to think through.
DG -- This is not as cheap as it was a few quarters ago; might have been interesting earlier though. Proven CEO brought back in to fix operational problems. Plan showing signs of improvement: reduce theft, simplify inventory, improve labor conditions, renovate stores to improve customer experience. Margins are slowly improving, inventory is coming down, same store sales growing and they are gaining market share (rural markets are their target and are largely underserved).
PTON-- possible super small initial position, but there are big challenges. Also, after looking into this, I ran across a post of someone trying to sell their Peloton that they had only used 2x--I'm not thinking that there is going to be a 'turn around'. Peloton sells premium fitness hardware at near-cost to onboard users into a high-margin recurring subscription ecosystem. They have pricing power and high switching cost for existing client base--if they can keep old users and get new users to buy the hardware. Cost cutting has improved gross margins and slightly improved operating margins. Attempting to branch into commercial business (hotels/gyms) and to integrate AI personalization. Execution of these is yet to show benefits. If hardware sales continue to decline, users won't get sucked into the subscription in the first place. Also, the CFO has recently left.
Q6:
Find recovery or takeover candidate stocks that meet Anthony Boltons criteria:
Strengthening financials
Fixable business problems
Low institutional ownership
Minimal broker enthusiasm
Poor recent share performance
For each stock found:
1) Analyze management performance. Assuming management has a clear turnaround plan, are they following it in measurable ways? Are they doing 'little things' better? Are they beginning to perform in line or better than competitors?
2) How does the company make money? How does it work?
3) What is being assumed in the current price? What key factors are already built in? How plausible and probable are these key factors?
3) Establish the downside at current price: Assume you are looking backwards from a 50% decline in current stock price and create a plausible scenario in which this happens.
The market gave DG credit in the stock price very quickly, well before strong evidence of a turnaround.
Question 1: Please “map” Bolton as an investor on as many dimensions of an investment style as possible.
https://datawrapper.dwcdn.net/sia30/1/
Question 2: What about Bolton’s background and circumstances made his approach the right one for him?
Anthony Boulton is an English professional investor with an academic background. He went to Trinity College, Cambridge, as an undergraduate – famous for its music – and he now composes classical music in his retirement.
His whole approach reflects that: structured, organised, thorough, with an analytical process-driven system. Over the top of this sits an intellectual curiosity and questioning, which enables him to take a less conventional approach and take calculated risks to achieve a superior outcome.
He has a checklist-based approach, which I like because it improves the odds of a good outcome and removes judgement and "gut feel" from the places where it is less helpful.
For example
Do I have an investment thesis that I can easily explain? (From Peter Lynch – I like people who listen to others and credit them.)
Does the company have a clear, strong business franchise? Is the business model robust?
Does the company have a weak balance sheet? Does it pass the Altman Z score? If not, it's usually best avoided.
Always follow the cash, and make sure the cash flow backs up the story.
Does he trust the management? "Someone who has let down or disappointed investors once is more likely to do it again."
In all these cases he is not completely black and white, but he specifically says in some cases that he would have had a better outcome if he had been.
Question 3: What are your favourite Bolton investments? Why?
Cairn Energy: here the strengths were good management in an area where trust is vital. Layered on top was a good business model identified by that management of balancing low-risk cash flow-generating assets with big stakes in higher-risk exploration interests, which, if they succeed, would make shareholders a lot of money. The cash flow limits the downside risk and pays for the bets. A 7M investment yielded a £3bn outcome and pushed Cairn into the FTSE 100, making it a 100 bagger for its original investors. A great example of most of his principles in action.
Question 4: What are your least favourite Bolton investments? Why?
Autonomy. He bought this because of a tip from a colleague. He does not say so, but I don't think he understood what the company did. I say this because I never met anyone who could tell me convincingly how they made money or what they did. Words like "mining", "organising unstructured company proprietary data", and "helping drive insights" were bandied about along with astonishing growth, but details of how they did this? No one ever seemed to know, but the word "proprietary" was used a lot, I remember. He does admit that the only smart thing he did was to sell before the bubble fully burst. Rather depressingly, it was one of his best performers. Sometimes it's more important to be lucky than good, though you might have to put a value investor on a rack to get them to admit that...
Question 5: What are 1 or 2 stocks that Anthony Bolton might find attractive in the current environment?
I have picked Sage Group PLC.
Sage is an accounting and payroll software company based in the UK, but its business is international, with the UK only making up 20% of its market. It offers a series of packages, from a small company using Sage Line 50, with upgrades to multinational, multicurrency support for global companies. Like all software companies, it has been marked down by the indiscriminate sell-off due to the perceived AI threat. It is down from £13 per share to £8.40 in the last year.
I think this share would have appealed to Anthony Bolton. He used a range of valuation methods, and one he specifically mentions is historical PE. Currently it has a PE of 20. The last year you could buy it cheaper than this was 2018, then back to 2014 for the next opportunity, when it traded around a PE of 18. If it grows in line with broker forecasts, it has an FPE of 16. Its metrics are excellent; ROCE, CROCE and EBIT margin are all over 20%, all of which he mentions as hallmarks of a good company. It grows around 9% a year, and it's forecast to do that for the next 3 years. It has paid a dividend for 34 years.
He would have known how deeply entrenched it was in its customers and how many cheap alternatives have tried to displace it over the years, and how skilled it has been in defending itself. I think he would have thought the danger of AI modest because:
Companies don't choose accounting software on cost; they choose it for appropriateness, support, longevity, ecosystem and scalability. No one ever changes their accounting system without very good reason; it's expensive, risky, and rarely produces any benefit. It is usually done to avoid the pain of continuing to live with a system that the company has outgrown or is no longer adequately supported.
Every accountancy practice is familiar with Sage, so support is easily available, and there is no learning curve with your accountant to become familiar with an unknown system.
The packages are scalable, so moving up the ladder as the company evolves is relatively painless (no accounting change is without pain).
There is an ecosystem of support software agents that can build interfaces to other proprietary software and provide tailored software support.
An AI-based system would have to replicate all of this, and even then many companies would only switch when they had to, even if the new system was virtually free, because accounting software is a small cost to most businesses. The pain of transitioning systems is huge, the benefits are zero (accounts are accounts), and accounting disasters are the quickest way to bankruptcy of all. I remember a very successful listed consultancy company in the early 2000s that very nearly went bust when its accounting system was unable to raise invoices for more than a month due to software issues. Its share price did the most spectacular V-shaped drop and recovery, and lots of the company employees bought shares on the dip because they knew it was a sure thing.
Lastly, Sage has proved adept at buying promising competitors and either developing their products or killing them. Never underestimate what a good company with plenty of cash can do to stamp on competitors when they look threatening but are still small. Where it can't do that, it copies them: Xero has been an excellent cloud-based competitor for 20 years, and Sage responded by moving to the cloud, improving its offering and making it easy for small businesses to transition from Xero to Sage as their business grows and needs more features and integration. I think that Anthony Bolton would have appreciated all these kinds of institutional details which make general scares great buying opportunities if the scare does not really apply to the company under consideration.
Question 6: Come up with an AI prompt based on Bolton’s approach.
"As a financial analyst, make a list of companies in the UK FTSE 250 with strong fundamentals that have been heavily sold recently. List them in order of their likely recovery with reasons for whether they might recover or not."
Generated an interesting list.
On a scale of 1 (lowest) to 10 (highest), what are the degrees of difficulty in answering the question of how AI will affect Software companies in general and Sage Group in particular?
As usual, this is a good question.
Like so many questions about the future, this is hard and uncertain, so let's give it an 8. But as investors we are used to this. We know that we have no idea where the stock market will be in a year's time or what interest rates will be in 2 years' time. I'd give both of those an 8 score as well. So can we say that we know the outcome? Definitely not. But is there anything useful we can say about it based on experience? Here, as in so many other areas of the market, I think that the answer is a yes, but always with the humility of bitter experience, I could easily be wrong.
I'm going to break it down into some separate questions to try and give a sensible response.
How good is the current type of AI going to get in the near future? Will it be able to give a step change in current performance rather than a steady evolution of its current capabilities?
I'm going to give a cautious no on this. The approach is a continuing increase in computing power and post-processing modules and already shows signs of diminishing returns. This approach of an LLM is always going to weigh towards correlation rather than a true model of the real world, because it is built on correlation, which is by its nature error-prone.
Counterfactual. Already the models are focusing on checking and rechecking their own work for accuracy, and new modules with more bespoke abilities in areas where LLMs are weak, like arithmetic, are being built onto the front of LLMs. Might progress here fully address the weaknesses and allow much more accurate "thinking" to emerge?
Is there evidence that a new and stronger, more general intelligence AI is likely to emerge from the current models which would change the game?
Not that I can see.
I have neither seen nor read nor had any experience of any evidence of a breakthrough in this area, and I have little theoretical reason to think that the current approach is showing any convergence with human intelligence. And if it does, it won't just be software stocks that will have to rethink everything; pretty much the entire planet will need to, including every company, organisation, government and individual.
Counterfactual. Apart from the previous counterfactual, none.
How effective will AI, which is undoubtedly improved but still based on current technology, be in helping programme new platforms? Undoubtedly it will help speed programming and help less skilled programmers do more ambitious things.
Counterfactual. There are questions around whether programming complex systems will actually be speeded up if the code written is of poor quality and the skilled programmer has to spend significant time correcting it, and the unskilled simply misses the problem and writes code which either fails under different circumstances or is very difficult to follow and debug – similar to the so-called spaghetti code written in the early days when software writing programmes were much more freeform. However, I think this is unlikely to be a zero-sum game, and the result will still be quicker and therefore cheaper.
Is AI proprietary and a "winner takes all" technology?
This is, to me, a critical point. ChatGPT, Anthropic and Google are all producing models with very similar abilities, and the lead is passing from one to the other rapidly. There is no frontrunner, and therefore no "winner takes all".
Unless this changes, AI models could easily become commoditised.
Counterfactual. This is linked to whether there is a breakthrough in model performance. Maybe one firm might achieve this and the others not, but it's hard to see this as a likely outcome. There is too much money available to poach the talent and find out what has been done. China has produced very capable models with far fewer computing resources, showing what can be done.
Given that programming becomes easier, does this cost advantage alone mean a displacement of current software companies by cheaper competitors?
I'm going to give a reasonably certain no to this. There have been plenty of technological breakthroughs, and plenty of companies that have grown large and powerful on the back of them, but the shape of this is more driven by the business model than the technology. Meta, Google and similar platforms built businesses by giving free stuff that people wanted to consumers and sold the targeted data to advertisers. This squeezed the whole remaining advertising market and damaged the incumbents there indirectly, rather than by head-to-head competition. The incumbent software companies sell expensive stuff to other corporations with complex needs and embedded systems. Different business model, and probably a different outcome. Most software companies have a very in-depth understanding of their customers' needs and are in a prime position to develop AI solutions for existing customers while still supporting their complex legacy needs. This particularly holds if one AI model is much the same as another in terms of power and reliability.
I see the most likely outcome is that the strong and capable software companies will use AI both to reduce their costs and sell new services based on AI to their customers, leveraging their knowledge of their customers' needs and trust in their solutions, and outsiders will have a long battle to win market share. The worst-case scenario is that their pricing is compromised. The weaker ones, with products that are easy to reproduce and customers not so tied in, are much more vulnerable. Consumer-facing software is the most vulnerable of all, where it is much easier for individuals to switch.
Do I think these arguments apply to Sage? Yes. It will become ever easier to build good accountancy "add-on" packages to existing software, which is something that has happened for many years, so it is not a new problem, but it will grow. It will also be easier for startups to program new accounting AI native software, whose benefits might easily be an ability to do simple accounts without the help of a bookkeeper, which would give a nice sales point to new small traders. More pressure and more competition, but more opportunities as well for acquisition, upselling of services, penetration into smaller markets using lower-cost programming and more rapid development and integration of existing products.
Q1, a contrarian - against the crown investing, owned limited (max 50) issues - most funds own many more securities
based the investment in a company on a thesis - which if violated, would give a reason to re-evaluate or sell
always looked at portfolio as if he was building for the first time (from-scratch) as opposed to holding on just because you own it
used the importance of valuing management as opposed to just relying on numbers to evaluated an investment, he needed to separate the spin as opposed to facts from management
would reduce outsized positions in bear markets
he "dug deeper" , searched widely and kept an open mind regarding the investments he made,
he needed the company to have a competitive advantage or moat
a company also needed to be a cash generator
he looked at under owned sectors and companies
he also looked for underfollowed companies
he used technical analysis to evaluate the results he got from his fundamental analysis
he incorporated market sentiment into his evaluations
Q2. He graduated with a degree in engineering. This gave his a different perspective than those who had graduated from business schools in finance. His first jobs were in investment management as opposed to the more popular investment banking field. He was able to see the operations of the investment profession from a first hand basis a watch it evolve over his career.
Q3. Cellnet and Security Services as they were his largest holdings over a number of years and were undervalued.
Q4. Parkfield and Sportingbet - He was referred Parkfield by a broker prior to meeting the CEO. Financial difficulties were later disclosed, probably a result of financing acquisitions. Sports betting - a bad investment.
Q5. Landstar - a slightly undervalued mid-sized transportation company.
Q6. Evaluate both a European and South Asian indices for mid-size companies using Bolton's criteria for evaluating companies, management, shares, sentiment, risks, financials, valuations, takeovers and special situations.