Performance Update for Colm Fagan's ARF

Thanks @NotMyRealName
The market likes the news anyway!
I’m still way down on my original purchase price but could be more than breaking even overall (I haven’t checked) given that I added to my holding when the price tanked.
Of more interest is when they’ll publish their results for the last financial year. They’ve been promised for “August” for ages but still no mention of the precise date. That’s surprising.
 
Last edited:
I recall getting nervous around mid-2006. Ten years previously, in 1996, I started transferring shares in Life Strategies, the business I’d formed three years earlier, to an Employee Share Ownership Plan (ESOP). In the intervening decade, the business prospered, thanks largely to that decision: some fantastic colleagues joined the firm because of the opportunity the ESOP gave them to become part-owners. I’m forever grateful to them for making the leap of faith.

Yet, the house that my family moved into that same year, 1996, increased in value by almost as much as Life Strategies did over the same ten-year period. We worked our butts off to grow the business, yet the house that I did very little with appreciated by almost as much. That didn’t seem right.

I feel much the same now.

My pension fund’s market value at 31 July 2026 exceeded its value on 30 June 2016 by 68.75%, despite withdrawals in the intervening period of more than 80% of the fund’s value in mid-2016. The compound return over the period was just shy of 12% a year on average.

That too can’t be right. The good outcome wasn’t due to any wonderful investment insights on my part. My stock-picking skills are average at best. It was a consequence of sticking resolutely to a strategy of investing as close as possible to 100% (typically, over 99%) in equities and keeping costs low. The portfolio is highly concentrated (just 13 shares at present), there are no collective investments and turnover is exceptionally low: currently, the average holding period is over 7½ years. There is nothing strategic about the decision to invest in a small number of real businesses. It’s simply down to personal preference and has had little or no bearing on the overall return (a topic to which I plan to return): the average return over the period was almost identical to that from a passive index – but I’ve had far more fun!

I’m not sure what to do about my current nervousness. Unlike 2006, when it was patently obvious that Irish property was grossly overvalued, I don’t think the shares in my pension fund are significantly overvalued at present, if at all. I’m still happy to be fully invested (99.4% invested at time of writing). Yet, at a macro level, something must be wrong if capital is earning so much. Does it mean that labour is earning too little? My actuarial training leads me to expect an Equity Risk Premium (ERP) of the order of 4% to 5% a year. It’s been far more than that in recent years. What are the implications for society? I’m not an economist, but I don’t like it. Can anyone enlighten me?
 
Last edited:
“You’ve worked hard for your money; it’s now time it worked hard for you” was a saying once loved by life insurance salespeople. It’s one that I’ve taken to heart for my own pension fund.

“No passengers allowed” is my motto. I expect all my investments to earn their keep in the long-term by delivering the risk-free return plus the equity risk premium, which I estimate at between 4% and 6% a year. There are no bonds in the fund for the simple reason that bonds are (supposedly) risk-free so, by definition, can only earn the risk-free return. Similarly, I try to keep cash to a minimum, typically to less than 1% of the fund, just enough to cover pension outgo and expenses over the next month or two.

Some investments fail to deliver the targeted return, due either to bad luck or to poor selection on my part. If the latter, then I prefer to admit my mistake and seek to offload the share at an opportune time.

Ryanair is one of the top five holdings in my 13-share portfolio (but well behind the first two, Standard Life and NVIDIA). It’s also one of the longest standing: I bought my first two tranches in 2016 and 2017 at an average €15.85 a share and added more than 40% in 2025 at an average €21.82 a share. Allowing for dividends (Ryanair only started paying regular dividends in 2024), the average return on my investment would be just over 5% a year if I were to sell at the current price (just shy of €24). The price would need to be closer to €29 to deliver the target return of risk-free plus the Equity Risk Premium.

All the above figures are before costs. Because Ryanair is quoted on the Irish stock exchange, the cost of buying it is higher than if it were quoted in London or New York. Stamp duty in Ireland is 1% compared to 0.5% in the UK and zero in the US. The above break-even calculations ignore costs.

The share price exceeded €29 at the start of this year, before the Trump/ Netanyahu war on Iran. Will it get back to that in future?

Ryanair’s revenues increased by over 130% in the nine years 2017 to 2026; revenues per share increased even more, by over 175%, in the same period, because share buybacks reduced the number of shares in issue. Yet the share price now is only 50% higher than the average at which I bought in 2016/2017.

I don’t think management has done a bad job, so I expect the share price to recover - eventually. Therefore, I’m inclined to hold on – for now at least.
 
Last edited:
CharlieMac said:
Nvidia is one I'm curious how it's working out for you. For a while there it looked like it could "turn Goodwin" on you but I think it's up a lot now since you bought it and I suspect worth holding. Even Goodwin is slowly recovering I see.
Click to expand...
Hi @CharlieMac
I don't know if you've been watching the share prices for both, but the last couple of days have been good for Nvidia and Goodwin!
Nvidia rose over 8% yesterday post results while Goodwin is up 12% this morning (as at time of writing) following its results.
I think I'll retire now!
 
Colm

In the case of Nvidia, this has been very successful investment for you. Do you ever consider of think of taking the gains of the table and reinvesting into another company or are you just going to rebalance your portfolio overtime?
 
Hi @Bluefin
I’m an investor, not a trader, so I plan to stick with it for the long haul, other than the occasional withdrawal for pension income and rebalancing. Besides, I think that Nvidia is still substantially underpriced (my opinion, of course) so I have no immediate plans to sell.
 
Colm Fagan said:
Ryanair’s revenues increased by over 130% in the nine years 2017 to 2026; revenues per share increased even more, by over 175%, in the same period, because share buybacks reduced the number of shares in issue. Yet the share price now is only 50% higher than the average at which I bought in 2016/2017.

I don’t think management has done a bad job, so I expect the share price to recover - eventually. Therefore, I’m inclined to hold on – for now at least
Click to expand...

Ryanair is a best-in-class company with a great track record, a highly incentivised and motivated CEO, leading a very competent management team. But it is an airline! The share price will largely be driven by sectoral factors and there will be few opportunities when the fundamentals will be reflected in the share price. Fine for those with a long runway. Interesting choice for the pension!
 
Hi @Itchy. You’re right! The reason I looked at them now was to decide whether to offload them, partly for the reasons you cited
I decided to hold off for now. I expect Ryanair to be one of the survivors in the business. Eventually- in my opinion- they’ll come good; however, I’ll be more inclined to sell them to meet pension outgo than to sell, say, Nvidia.
 
Colm Fagan said:
I’m not sure what to do about my current nervousness. Unlike 2006, when it was patently obvious that Irish property was grossly overvalued, I don’t think the shares in my pension fund are significantly overvalued at present, if at all. I’m still happy to be fully invested (99.4% invested at time of writing). Yet, at a macro level, something must be wrong if capital is earning so much. Does it mean that labour is earning too little? My actuarial training leads me to expect an Equity Risk Premium (ERP) of the order of 4% to 5% a year. It’s been far more than that in recent years. What are the implications for society? I’m not an economist, but I don’t like it. Can anyone enlighten me?
Click to expand...

I share your concern. The US market cannot be described as cheap. Non-US stocks less so but as goes the US market so goes broader international markets.

As it pertains to US there is a double (even triple) counting of the kind that is hazardous to future returns. The US government unusually is printing fiscal deficits of nearly ~6% at a time of full employment. The sovereign running such large deficit creates, via simple national accounting identify, a surplus elsewhere (households, corporates). Much of that surplus is appearing as inflated profits (surplus) in the corporate sector. Profit margins of US listed corporates are at record highs as a result. What has the market done with these expanded historically unusually high profit margins fueled by never before seen and unsustainable fiscal deficits (& the AI capex buildout)? It has assigned an unusually high multiple to those earnings.

So for those watching you have unusually high fiscal deficits, fueling unusually large profits and margins in the US corporate sector for which Mr.Market is paying unusually high multiple of those earnings. This is not the recipe for long term returns.

Finally the most recent AI capex boom (~$2trn) - due to accounting methodology has a very interesting consequence for earnings due to the timing methodology of GAAP accounting. See the hyperscalers capex spending becomes somebody else's earnings in the quarter its spent (Nvidia, Caterpillar) the costs however dont show up that quarter on the spenders accounts depressing their EPS. See this is capex, the costs are capitalized on the balance sheet as per useful life methodology - in many cases 5-7 years. So we get inflation of aggregate earnings in the short term due to the mismatch in revenue recognition and cost depreciation accounting. In this sense a capex boom is the perfect paradigm for inflating earnings in the short term and we've had nearly ~$2trn spent in a very short time period. The question on this spending remains its ROIC. If its good - profits rise to meet the increased depreciation costs and EPS growth is sustained..... if its not as good as expected earnings dont rise in tandem with the flow through depreciation & EPS gets compressed.

I think the former case (deficits fueling unusually large profit margins plus high multiplies) is clearly unsustainable IMO but timing remains uncertain. The latter (AI Capex earnings boom) is TBC and a function of the returns to this spending. It may work out or it may not. I frankly have no idea.

However what I would say is that SPY and QQQ future returns have become very very dependent on fiscal deficits running at 6% and this AI spending 'working out well'. This path dependency has echoes of 1999 (e-commerce had to work out and fast) and 2007 (house prices had to keep increasing to sustain the credit boom).

I dont like markets where returns become so path dependent and correlated to one 'thing' working out.
 
Last edited:
Hi @letitroll
Excellent post.
There's a lot to unpack.
I won't be able to do it justice on a Friday night after I've consumed more than half a bottle of wine, but I'll make a start. We can then see where it leads us.
I don't disagree with anything you've written but some aspects of the market make me wonder if we've both got it wrong.
For starters, I agree with you on US fiscal deficits. Anticipating these, I decided to reduce my US exposure around a year ago, but the USD is no weaker now against the EUR than it was then. Why? I'm not an economist, so I can't even venture an answer.
Your argument about positive cash flows being treated as revenues while negative cash flows are treated as capital expenditure, thereby creating an asymmetry, is fascinating; however, I'm not sure I agree that the higher earnings (for the companies receiving the positive cash flows) are being capitalised at an unusually high multiple.
Referring back to post #492 in this thread (17 May last), I did a back-of-fag-packet calculation of what I'd consider a fair price for Nvidia. The following is what I wrote:
Colm Fagan said:
I decided to take an old actuary’s look at NVIDIA, a darling of stock market bulls. It accounts for close to 10% of my pension portfolio.
Its earnings in the quarter to 26 January last were $1.76 a share, having grown by an average 19% a quarter from the corresponding quarter in 2025.
I decided that it was reasonable to assume average earnings growth of 10% a quarter for the next eight quarters, and for the share to be valued then at twenty times annualised earnings (eighty times quarterly earnings) and that it would be appropriate to discount that at 15% a year to get a reasonable current value for the share.
The calculation is as follows (roll the drums!!!):
1.76 * (1.10)^8 * 4*20/(1.15^2) = $228.
NVIDIA’s closing price on Friday last (15 May) was $225.32. The conclusion therefore, on my back-of-a-fag packet model, is that the share is reasonably priced at present. In fact, I think my assumptions may err on the cautious side, so I’m comfortable continuing to hold the share at its current price.
Click to expand...
Now, updating that calculation based on the earnings posted this week, we get the following:
Quarterly earnings (diluted) for quarter to 26 July 2026: $2.46 (up from $1.76 for the May calculation).
Assume that earnings grow at 10% a quarter for next six quarters (more conservative than the last calculation, which assumed them growing at 10% a quarter for the next 8 quarters. In the last four quarters, earnings grew by 22.8% a quarter; the corresponding growth rate was 19% a quarter when I did the calculation in May).
Making the same assumptions re terminal value (20 times annualised earnings, 80 times quarterly earnings, and 15% pa discount rate), we get the following updated "fair" price for the share (another roll of the drums!):
$2.46*1.10^6*4*20/(1.15^1.5) = $282.71
The current share price is $217.73.
I'll repeat again what I said at the start, that I've consumed more than half a bottle of wine, so my calculations may not be reliable. If they are reliable, the inflated earnings are not being capitalised at an unusually high multiple.
 
Colm Fagan said:
the inflated earnings are not being capitalised at an unusually high multiple.
Click to expand...
I don't think there question applies as much to Nvidia as it does to the likes of Anthropic. Nvidia have sold their GPUs and whatever hardware and correctly recognised the revenue. Anthropic have bought the hardware, recognised it as CapEx, and are depreciating it over the expected useful life. If that useful life estimate is too optimistic (isn't this type of hardware usually rapidly devalued by technical advances?), there may well be a hit to profits on the form of an impairment to the value of the hardware on the balance sheet in the medium term. There is also the expectation that this hardware will allow Anthropic to make its projected future earnings. If either (or both) of the assumed useful life or profits so not come true, there would be a correction to the value of Anthropic.

I think Nvidia is fine, barring a reduction in future earnings if this AI thing doesn't catch on.
 
Colm Fagan said:
I'll repeat again what I said at the start, that I've consumed more than half a bottle of wine, so my calculations may not be reliable. If they are reliable, the inflated earnings are not being capitalised at an unusually high multiple.
Click to expand...

Excellent grape driven analysis!.....you are right on nVidia....the P in the PE is not insane but the E, imo, is.....I think it would be too bubbletastic, in the case of nVidia, to have both the P and the E be in a bubble territory at the same time......nVidia's E right now is a levered bet that is totally dependent on (a) AI generating returns that justify the capex such that the E is sustained over time and this just isnt a VC funded blue sky thinking binge on GPUs and (b) nVidia hardware remaining the defacto GPU of choice for frontier labs which itself is a bet on whether scaling laws apply to model performance i.e. more processing power times more data will always lead to model improvement (& improvements of a magnitude which justifies the purchase of leading edge compute)

Indeed NVidia is a bet that the frontier even 'works' financially at scale which is to say that Anthropic & OpenAI both currently loosing oodles of money provide enough incrementality above and beyond Open Weight/Open Source models coming out of China ( trained on non-NVidia GPUs). So two path dependent bets for NVidia - (1) that AI 'works' and (2) the frontier labs business models work such that nVidia's chip command their 70-80% gross margin in perpetuity.

I will note the most recent NVidia deals where they've effectively underwritten the value of their chips in data centres five years out and by extension lowering the cost of capital for the neoclouds is actually a version of price cut on their GPUs already....a sneaky one that doesn't involve NVidia cutting nominal prices but they are now selling percentage of their chips with a put option which is technically costless until the put option is called but make no mistake about it that risk is sitting on NVidia's b/s as a contingent liability five years out.
 
Last edited:
Hi @letitroll
Boozing done, holiday over, and I've had time to study your last note in detail. Thanks.
Most of it is well over my head, since I hardly know one end of a laptop from the other, not to mind AI or anything close to it.
Therefore, I'll stick to my simple/ simplistic model, to see what needs to change having regard to your post.

My model derived a "fair" share price for Nvidia as follows: $2.46*1.10^6*4*20/(1.15^1.5) = $282.71.
Let's look at each component in turn:
$2.46 is last quarter's earnings. No problem with that starting point.
The 1.10^6 is the assumption that earnings will grow by 10% a quarter for the next six quarters. I think we both agree that that probably understates protected growth in the period. When announcing the latest quarterly results, Nvidia said that it projects revenues to grow 70% over the next year, that customer demand would lead to an expectation of 100% growth, but supply constraints limit growth to 70%.
The next item in my valuation model is: *4*20, i.e., multiplying projected quarterly earnings six quarters out by 80 (i.e. 20 times annualised earnings). I think this is where you say I'm being overoptimistic. My model says that Nvidia will be valued like a "normal" company six quarters from now, qualifying for a valuation multiple close to a market average (I haven't looked but I think 20 is close to an average market multiple). You're saying, I think, that earnings are dodgy because of all these sweetheart deals and implicit/ explicit put options, and don't have the permanence needed to justify a multiple of 20.
Finally, the discount factor 1.15^1.5 says that I'm looking for a 15% annual return in the intervening 18-month period, given all the uncertainties involved. Let's leave that unchanged.
Applying a multiple of 15 to projected annualised earnings six quarters out, leaving other elements unchanged, gives us a "fair" value of:
$2.46*1.10^6*4*15/(1.15*1.5) = $212.03. As expected, this is 75% (15/20) of the $282.71 derived earlier. The current share price (COB last night) is $224.41, so the revised conclusion - if we believe the latest figures - is that the share is overpriced currently,
The optimist in me (one of my many problems is that I'm too much of an optimist!) says that, while the ultimate multiple of 15 may be correct, we've understated projected earnings by that time. Nvidia projects 70% revenue growth in the next year, which presumably should translate into earnings growth of at least 70% (greater economies of scale, counterbalanced by greater pressure on gross margins); however, we're only projecting earnings growth of 46% in the next year (10% compounded for four quarters).
Let's see what the "fair" value looks like if we project earnings growth of 70% for next 12 months, followed by 10% a quarter for the next two quarters, and applying a multiple of 15. We get:
$2.46*1.7*1.10^2*4*15/(1.15^1.5) = $229.60, which is a little bit above last night's closing price.
QED!!!!!!!
(Now you see how actuaries can manipulate valuation results to suit whatever number they want to come up with!)
Seriously, thanks to your patience, I think I now have a better understanding of the risks that need to be considered in relation to my investment in Nvidia.
I must stress, though, that this is just one (optimistic) person's take on the share. We've seen the change in valuation wrought by a fairly small change in the assumed multiple six quarters out. Other relatively small changes could have an equally outsized effect.
At the least, I hope that this little exercise shows readers of my regular updates how I decide if a share is good or bad value.
It's also worth pointing out that I employ other valuation techniques for different types of shares. I hope to write about one such technique later this week.
 
Last edited:
Colm Fagan said:
You're saying, I think, that earnings are dodgy because of all these sweetheart deals and implicit/ explicit put options, and don't have the permanence needed to justify a multiple of 20.
Click to expand...

Yes permanence...but not necessarily because of circular financing or put options, these are just methods by which Nvidia keeps the plate spinning. For NVidia truly to maintain its unit volume AND margins. AI, writ large, needs to start deliver ROI commensurate with the costs being incurred AND within that overarching reality working out NVidia chips specifically need to deliver incrementality over and above competitors GPU/TPUs that justifies their greater costs - as mentioned I think that tends to boil down to the frontier models (OpenAI/Anthropic) working out technologically (scaling laws continue to deliver model gains and those gains have a large enough delta over prior generations) and as underlying enterprises OpenAI & Anthropic can turn a profit sustainably

So yes.....I dont think NVidia earnings in 2026 can be viewed as having reached a "permanently high plateau".....volume or margin or both is likely to get them at some point, the low-ish PE three years out I think is an admission that margins cannot be sustained at these levels so its an element of multiple conservatism rooted in margin compression reality. And rightly so the historical record of companies maintaining 60% net margins on physical goods over time is essentially ZERO the volume bet is the one I quibble with....and it's quite binary and based on the frontier labs working.
 
You must log in or register to reply here.
Back
Top