I fooled myself three times

For two years the question I asked the simulator was "will this work?" In the last month it has changed to "which version?" I stop work in April 2029, and the plan has stopped being a spreadsheet exercise. It is now a set of real decisions: how much to take as tax-free cash, whether to buy an annuity, how much to move out of the stock market before I stop, and what order to draw everything in.

Every one of those is a comparison. So I spent a Sunday running my plan one way, then another, and putting the results side by side. By the end of the week I had misread the answer three times.

None of the three was the calculator getting the sums wrong. Each time it was me reading a difference that was not the one I thought I was looking at. They are worth writing down, because I suspect anyone comparing plans, in any tool, will walk into the same three.

One: two plans £175,000 apart before anything happened

I ran two versions of my plan that were identical except for one thing: in one, half my pot buys an annuity at 62. In the other, it stays in drawdown. I expected the two to track each other exactly until 62, then split.

They did not. At 59, three years before any annuity, the drawdown-only plan showed about £175,000 more in the pension. Nothing had happened yet that could explain it.

What had happened was this. A Monte Carlo run is thousands of possible futures, and the figures you look at year by year come from one of them: the one that lands in the middle once they are all sorted by how well they turned out. Change the plan and the sorting changes, so a different future can land in the middle. My two "middle" futures had different markets. In the first year the drawdown plan's middle future earned 12.3%; the annuity plan's earned 6.2%. I was comparing two different stock markets, not two decisions.

Turning off the randomness settled it. With growth and inflation set to fixed, both plans held exactly the same pension, to the pound, every year until 62, and split only when the annuity was bought. That is the comparison I thought I was making in the first place.

The fix, if you want to compare two strategies: set growth and inflation to fixed on the Markets & inflation tab. There is then only one future, so there is nothing to pick differently. The "use the same futures every time" switch on the About you tab keeps runs repeatable, and it is the right tool for smaller changes, but it cannot stop a big change of strategy putting a different future in the middle.

Two: the income that never grew

My plan targets £60,000 a year after tax. Comparing the two plans year by year, both showed exactly £60,000 at 65, at 70, at 75. Suspiciously exactly. It looked as if nobody had applied any inflation.

Someone had. The comparison was showing today's money, and my plan is set to keep my spending power level. £60,000 in today's money is the whole point of the setting. The pounds actually paid at 65 were £71,821. The flat line was the plan working, not the plan broken.

Both views are useful, and they answer different questions. Today's money tells you whether you can afford the same life. Nominal tells you what will actually land in the bank account. The side-by-side comparison now shows either at the flick of a switch.

Three: £70,000 out, £45,000 less

This one is personal. I want to buy an overland vehicle when I stop, and the obvious way to pay for it is tax-free cash. So I compared taking £60,000 of tax-free cash with taking £130,000 and spending £70,000 of it on the vehicle.

The pension at the end of the first year was only £45,000 lower, not £70,000. So where did £25,000 go?

Some of it was real. The £70,000 does leave the pension the moment you take it. But the 25% tax-free allowance is for life, not for each withdrawal, so taking more of it up front leaves less for everything drawn afterwards. That same year, only £8,922 of my ordinary withdrawal came out tax-free with the bigger lump sum, against £13,311 with the smaller one, and I paid £1,756 more tax. Most of the rest was the first mistake again: two runs, two different middle futures.

The part worth knowing is the first. Tax-free cash is not free money you simply take or leave. Taking more of it now makes later withdrawals more expensive. On my plan it is still the right way to pay for the vehicle, because the alternative is drawing the same money as taxable income. But it is not costless, and it will not show up where you expect it to.

The one that saved real money

Not every comparison was a mistake. My first year of retirement showed £20,280 of income taxed at 40%. I assumed more tax-free cash would fix that. It made it worse, for the reason above.

What fixed it was changing the order the money is drawn in, from savings first to filling the 20% band first. Same plan, same take-home, first-year tax down from £15,652 to £10,615, and the slice taxed at 40% down from £20,280 to £7,687. About £5,000 in one year, from one setting.

Two ways to compare, and why there are two

There were already one-change comparisons on the site: your plan beside the same plan retiring later or earlier, spending more or less, or securing the essentials with guaranteed income, all run on the same futures. That is the right tool for tinkering with one lever.

What I was actually doing that Sunday was different. I had whole plans, made on different days, each with several choices in it, saved as the spreadsheets the year-by-year page downloads. I wanted them side by side. So over the weekend I designed a page for exactly that: drop in two to five of them and you get how many futures each one works in, what is left at your target age, the tax paid getting there, charts of wealth, take-home and tax by age, one year across every plan, and a list of only the settings that differ between them.

That last list is the one I would not do without. Two of my three mistakes started with a difference I had not noticed I had made.

Where my own plan has landed

For the first time I think I can see the shape of it. On my figures, and mine only:

• About three years of spending moved out of the stock market before I stop, into gilts or a fixed-term guaranteed product, so a crash in the first years does not force me to sell.
• A little more tax-free cash than I had planned, for the vehicle, knowing it costs something later.
• The rest left invested, drawn to fill the 20% band first.
• The annuity question still open, and now being asked properly, on a single fixed path rather than two different random ones.

None of that is a recommendation. It is where one person's numbers point, and it will move again.

If you are comparing plans

1. Change one thing, then check that you only changed one thing.
2. For a change of strategy, compare on fixed growth and inflation first. Add the randomness back once you know what the decision itself does.
3. Know whether you are looking at today's money or the actual pounds.
4. Look at the first year of the year-by-year. If it differs before anything has happened, you are looking at luck.

If you have compared retirement plans, in this tool or any other, I would like to know what caught you out. Corrections go in with credit, as always.

Figures from the simulator at versions 1.67 to 1.74, on my own plan. Nothing here is a recommendation, and I am not a financial adviser.