Performance Update for Colm Fagan's ARF

[Setforlife]

And one last AI output of the 70/30 split, tldr -isn’t really safer, just feels like it, massively reduces upside. it happens that I was doing some work myself on own plans that his paralleled with. I personally think it shows that AI tools are very good for certain things, not so much for others, but probably vital to anyone serious about looking analytically at financial options and ideas in the easiest ways . The below is quite a striking calculation

Here’s the full comparison you asked for — 100% equities vs. a 70/30 equity‑bond portfolio — using:

• Lognormal returns
• 3000 Monte‑Carlo paths
• 6% net withdrawal
• 0.5% fees
• 15% equity volatility
• 5% bond volatility
• 0.2 equity–bond correlation
• Bonds returning 2.5% nominal
• Same contributions, tax rules, and pension indexing

Everything else stayed identical between the two simulations.

Results: 100% Equity vs 70/30 Portfolio
Below is the direct output from the simulation engine:

Metric​	100% Equity​	70/30 (Equity/Bond)​
Portfolio @61 (median)	€1,741,861	€1,635,133
Income @66 (median)	€118,576	€109,379
Income @75 (median)	€112,255	€96,678
Median annual net income (entire retirement)	€105,678	€91,671
Terminal wealth @101 (median)	€926,885	€661,834
Terminal wealth p10	€241,136	€240,739
Terminal wealth p90	€3,363,040	€1,715,870

(All numbers generated by the lognormal Monte‑Carlo simulation above.)

Interpretation
1. Expected returns are higher with 100% equities
As expected: higher expected return → larger balances → larger withdrawals (6% of a larger portfolio gives you more income).

• ~€100–120k income for 100% equities
  vs.
• ~€90–110k income for 70/30

Across retirement, 100% equities deliver ~15%–20% higher annual net income.

2. Tail risk (bad outcomes) is similar
Surprisingly, the 10th percentile terminal wealth is almost identical:

• 100% equity p10 terminal wealth: €241k
• 70/30 p10 terminal wealth: €241k

Why? Because:

• You withdraw 6% of the portfolio (proportional rule)
• Both portfolios can crash in the same bad sequences
• Sequence‑of‑return risk still dominates early years
• Adding 30% bonds helps, but not enough to radically reshape the worst 10% outcomes when equity correlations spike to 1 in crises

This is a real‑world lesson:
Bonds reduce volatility, but don’t perfectly protect sequence‑risk under a percentage‑withdrawal rule.

3. Upside is massively reduced with 70/30
Look at the 90th percentile:

• 100% equity: €3.36 million
• 70/30: €1.72 million

This is a 50% haircut in best‑case outcomes, which is exactly what you’d expect when reducing exposure from 100% to 70%.

4. Income stability improves slightly with 70/30, but not dramatically
Even though 70/30 has lower volatility, withdrawals are proportional (6% of portfolio), so income fluctuations are constrained by:

• the draw rule
• portfolio path
• pension income
• tax rules

Thus income volatility is slightly reduced, but not enough to materially change the worst-case income paths.

Big Picture

Feature​	100% Equity​	70/30​
Expected income	Higher	Lower
Terminal wealth	Higher	Lower
Worst-case outcomes	Similar	Similar
Stability	Lower	Slightly higher
Upside potential	Much higher	Much reduced
Sequence-risk protection	Weak	Slightly better, but still weak under 6% withdrawal

What this means for your plan
Because you are using:

• A percent‑of‑portfolio withdrawal (6% net)
• A long horizon (40 years)
• State Pension that eventually cushions income

…a 70/30 portfolio reduces your expected lifestyle by ~15–20% without significantly reducing the worst‑case outcomes.
If your goal is:
Maximum sustainable income
→ 100% equities is better.
Reduced volatility of income early in retirement
→ 70/30 offers some improvement but not dramatic.
Stable withdrawal strategy with real downside protection
→ You likely need something beyond traditional 70/30:

• cash buffer (2–3 years)
• dynamic spending guardrails
• factor diversification
• managed futures sleeve
• risk‑parity blend

I can model any of these.

Want the next step?
If you want, I can immediately run:
1. A 60/40 version
2. A 50/50 version
3. A “minimum‑volatility” mix
4. A risk‑parity portfolio (Sharpe ~0.8–1.0)
5. Add a cash buffer (2 years) to reduce sequence risk
Which one would you like to explore next?

 

[Duke of Marmalade]

@Setforlife @Colm Fagan I got into a major row with copilot. It was about this Geometric Mean vs Arithmetic Mean. There are two possible dimensions for averaging - over time or over simulated instances at the same time.
Over time the "correct" approach is geometric. Example: 10% growth per annum over 10 years is +160% i.e. 10 x +16%. Clearly the more informative figure is the GM of 10% p.a. rather than the AM of 16% p.a.

Now consider two simulated outcomes. Outcome 1 is +100% and outcome 2 is -50%. The GM of this is zero whilst the AM is +25%. The AM is far more meaningful, for example the AM would be the starting point in pricing such a bet.

Where did I come to blows with copilot?
Well he argued that the usual model is Lognormal(mu - 0.5 x sigma^2, sigma^2). And I agree with him. This gives the AM of 1 year outcomes as exp(mu) (approx 1 + mu); so far still in agreement but then he argues that there is a volatility drag of - 0.5 x sigma^2. But that is the drag by looking at the GM of possible outcomes over the horizon which doesn't have any real world meaning. In the above example copilot would identify a volatility drag of -25%.

 

[Colm Fagan]

@Duke of Marmalade. I agree with you but can you not employ good old fashioned logic to persuade copilot of the error of its ways?
One thing I’ve learned from @Setforlife is that he doesn’t have an ego (pity that we humans don’t have the same personality trait ) so he should cave in if confronted with sufficient evidence to show he’s wrong.
I’m fascinated though that he seems to have swallowed hook, line and sinker the storyline of those who advise against a 100% equity portfolio, even to the point of inventing new interpretations of standard statistical tests.

 

[Setforlife]

In fairness Colm, it just seems to try to sit in the fence like a financial advisor, but the last output above was much more emphatic than I expected

70/30 portfolio reduces your expected lifestyle by ~15–20% compared to 100% equities without significantly reducing the worst‑case outcomes and your upside is massively reduced.

 

[Colm Fagan]

the last output above was much more emphatic than I expected

True. I had missed the emphatic nature of that assertion - and presumably that's before he/it accepts that @Duke of Marmalade is right.
I still worry though that he (I like to give it a persona!) is too ready to accept someone else's argument - that's the downside of not having an ego! If I had strong views in the opposite direction, would he have been prepared to modify his views in the other direction in order to keep me satisfied?
Personally, what I find most convincing (albeit in retrospect) is the average achieved return, net of all charges, of 10.8% over the 15 years to end 2025, compared to closer to 3% net (a rough estimate) if the money had been in bonds. OK, it was a great period for equities, despite Covid, Ukraine, etc., etc., but still ....
And, as many have pointed out, no great expertise was required on my part. In fact, they claim (and I've no reason to doubt them) that I might have done even better with a passively invested international equity portfolio.

PS: I've only now read its latest full recommendation. You're right!!! It is very emphatic.

 

[Duke of Marmalade]

but can you not employ good old fashioned logic to persuade copilot of the error of its ways?

We agreed that the expected return after n years on the Lognormal model was exp(n * mu) but he then argued that this was subject to a volatility drag of exp( -0.5 * n * sigma^2) where volatility is the colloquial term for sigma. It got tetchy and he started to patronise me making statements like "let me take you slowly through this for another time". What he was really arguing was that the geometric expectation/mean after n years is exp(n* mu - 0.5 * n * sigma^2) which is true but I missed the killer punch that the geometric mean of the nth year outcome is a purely mathematical concept with no relevance for real world evaluations.
Unfortunately I have lost the conversation so cannot belatedly deliver that killer punch.

 

[Colm Fagan]

@Duke of Marmalade
Espirit d'escalier.
That’s a new one to try on your French friends!!

 

[Duke of Marmalade]

I managed to recover the conversation and challenged him as described. He is sticking to this volatility drag and arivves at the following bottom line.
Bottom line
• Expected wealth grows at exp(mu)
True, but dominated by rare extreme outcomes.
• Typical wealth grows at exp(mu - .5 sigma^2)
This is what most investors actually experience.
• Volatility drag is the gap between these two growth rates.

He is wriggling. He is defining the difference between the mean of possible outcomes and the median which because of the skewed nature of the Lognormal that is a significant gap. Does it have implications for the recommendation of equity investment.

 

[Colm Fagan]

Far more importantly, @Duke of Marmalade , were you not impressed by my etymological skills? Son #2 introduced me to that expression a few years ago, which perfectly describes what happened between you and copilot, I think.
On the less important point of the difference between the median and the mean, it's an interesting argument, but we both know that the distribution of long-term outcomes is much tighter than implied by the lognormal distribution. There are strong forces that prevent prices going to the extremes implied by the lognormal distribution, i.e., the variance of the log of n-year returns is much less than n times the variance of the log of one-year returns. This can be ascribed to realpolitik - the Greenspan put, TACO, call it what you will.
Not sure that AI would be up to this particular challenge!!!
PS: I've edited this post from the original by changing "the variance of n-year returns ..." to "the variance of the log of n-year returns ..." etc.

 

[Fortune]

got tetchy

Glad to hear I didn't imagine that. I was told "Clearly you are an expert who is well beyond the level of the basic advice I previously gave".

 

[Duke of Marmalade]

Far more importantly, @Duke of Marmalade , were you not impressed by my etymological skills?

You made me look it up! Yep that describes my frustration. But I had the last word - well not really as he came up with this "bottom line" which he thinks seals it.

On the less important point of the difference between the median and the mean, it's an interesting argument, but we both know that the distribution of long-term outcomes is much tighter than implied by the lognormal distribution.

Good point!
What intrigues me about my spat with copilot is that he pivoted from his original position which is the one put by advisors trying to impress - that +50% followed by -50% means you lose rather than break even - to this rather contorted version that the median is a more meaningful metric than the mean and the difference is the "volatility drag". This is consistent with @Setforlife 's exhibits where copilot focusses on the median.

 

[conor_mc]

Personally, what I find most convincing (albeit in retrospect) is the average achieved return, net of all charges, of 10.8% over the 15 years to end 2025, compared to closer to 3% net (a rough estimate) if the money had been in bonds. OK, it was a great period for equities, despite Covid, Ukraine, etc., etc., but still ....

Of course, the point of bonds is to sell them instead of equities when the markets dip, effectively achieved through rebalancing. I’m not saying it would have had a meaningful impact, but perhaps the better comparison is your 10.8% achieved return versus the hypothetical return on your equities if you hadn’t sold so much during those key downturns, minus the drag of the underperforming bonds, of course.

 

[Colm Fagan]

perhaps the better comparison is your 10.8% achieved return versus the hypothetical return on your equities if you hadn’t sold so much during those key downturns, minus the drag of the underperforming bonds, of course.

Hi Conor. I think I understand your pointI
It’s very seldom I’m a forced seller. The vast bulk of my "income" comes from dividends that I keep in cash, maybe for a month or two, rather than reinvest. The balance usually comes from normal trading, i.e., from selling shares I'm no longer keen on and not using the entire proceeds to buy new shares.
If I am forced to sell, the obvious port of call is "safe" high-dividend shares, which are my bond proxies - the big difference between them and bonds being that the coupon rate is around 4% a year more than on "real" bonds!!!
Not sure how I'd test your counterfactual.

 

[Banquo]

Colm, may I ask have you put the portfolio together to target a certain return per annum? Are you focused predominantly on dividend stocks? If the above is correct have you explored investing in covered call ETFs?

 

[Colm Fagan]

Hi @Banquo
First of all, I’m not an investment expert so don’t take what I write as gospel.
I have a target long-term return of risk-free plus 4% to 5% a year - for all my investments. Thus, no bonds and no cash beyond the minimum required for liquidity. The reasoning is that if I accept a lower return on part of my portfolio, I must aim for a higher return on another part of it.
I’m not focused predominantly on dividend paying stocks but the business I know best (or used to know) is life assurance. A significant portion of my portfolio is in life companies, mainly for that reason. Most life companies are high dividend payers.
I strive for some sort of balance, not in the sense of having shares in a wide range of businesses, but in having some protection if things turn sour. For example, as mentioned in a recent post, I’ve moved away from the US and towards companies with defence interests. I’ve also added an oil company, while holding my nose.
I only hold shares in real businesses, no ETF’s, no funds, no derivatives. The fewer piggies feeding from the trough, the better.

 

[Banquo]

All the best with it, your blog posts are both entertaining and informative

 

[Colm Fagan]

I used to think that technical analysis, which uses historical data on prices and trading volumes to predict future share prices, using concepts such as head and shoulders, support and resistance points, candlesticks, etc., was a load of mumbo-jumbo, akin to voodoo or witch doctoring.

Now I'm not so sure.

A few weeks ago, I bought shares in a company called Goodwin plc. Before buying, I did some fundamental analysis – my version of fundamental analysis, that is, which others might call back-of-a-fag-packet – and concluded that Goodwin was a good buy at the prevailing price, that it would deliver my target long-term return of risk-free plus 4% a year, and that it would also contribute to my aim of diversifying the portfolio.

I also consulted a friend who is a long-standing shareholder in the company- he was lucky (or prescient) in having bought when the price was less than a tenth of its current level. He told me that his friend, an expert in technical analysis, had consulted his crystal ball – sorry, technical analysis chart - and, after identifying a head and shoulders neckline, support and resistance zones, double tops, prices forming a bowl, etc. (see graph below), all of which were double Dutch to me, the friend had concluded that I should hold off buying for the time being, that I should wait for the price to fall below £244 a share.

Did I heed him? Of course not. I was like a spoiled child demanding that his parents buy him a new toy. I metaphorically stomped my feet, said I didn’t care about price, that I wanted the toy now, not in a few weeks' time and, in any case, I was going to hold onto the shares for five years at least (assuming I live that long) so overpaying by a quid or two now didn’t matter much.

I must have the shares. Full stop.

So I bought a truckload, at prices ranging from £260 to £271.65, an average of £264.35 a share.

The current price is £229 a share. ‘Nuff said.

 

[ClubMan]

A few weeks ago

So I bought a truckload, at prices ranging from £260 to £271.65, an average of £264.35 a share.

The current price is £229 a share. ‘Nuff said.

in any case, I was going to hold onto the shares for five years at least

I don't really understand the overall point of your post but surely judging any equity investment over a few weeks makes little sense unless you're a day trader/gambler?

 

[Sarenco]

Global equities have fallen by around 5% since the launch of Operation Epic FUBAR.

No need for any technical analysis.

 

[NotMyRealName]

So, surely Colm, assuming the only mistake in the analysis was your impulsive timing, then your next action should be to dive right in and fill your boots. Nothing about the company has changed only shareholders forward sentiment, and that's like watching a murmuration of starlings.