What the AI portfolio has actually done

How has the Invest Board AI portfolio performed?

The value book started on 30 June 2025 at 10,000 and stood at 13,267 on 29 August 2026, a total return of +32.7%. The S&P 500 price index returned +20.0% over the same window, so the book is ahead by 12.7 percentage points, or 27.5% annualised against 16.9%. It got there with more risk: 15.9% volatility against the index at 10.2%, and a worst peak-to-trough fall of 17.5% against 9.1%. Every number on this page is dated 29 August 2026 and comes from the same performance API the app itself reads.

Those five numbers belong together, and most published track records show you one of them. Return without volatility is a half-truth. Return without the drawdown is a story about a ride nobody actually took. Return without the benchmark is not a claim at all, because a rising market pays everyone. The table below is the whole set, as of 29 August 2026.

Read the last two rows together and the character of the book shows up. It is up on more days than the index and it falls harder when it falls. Its best day in the window is +3.8% on 9 April 2026 and its worst is -4.1% on 6 June 2026. A five-stock portfolio does not behave like an index and should not be judged as though it does.

How is the AI portfolio constructed?

It holds the top five names from the value screen, equal weighted at 20% each, with the remainder in cash. Ranking is the composite score: moat, risk, valuation, optionality, management and business model, the same score the company pages show. The book rebalances monthly. When a name drops out of the top five it is sold in full and the new name is bought at 20%. There is no discretion in the loop, no position sizing by conviction, and no override when the machine picks something uncomfortable.

Equal weighting is doing something specific here. It removes the easiest place for a backtest to flatter itself, which is putting the biggest weight on the name that happened to work. Every position enters at the same size, so the return is the score's return, not a sizing decision dressed up as one. The rule that a name leaves the book the moment it leaves the top five is the same kind of constraint: it forces exits that a human would talk themselves out of.

The composite that does the ranking is documented in full on the scoring methodology page: which categories exist, what each one weighs, which data feeds it, and what a score cannot see. The machines page covers the layer underneath, the analysis agents that read filings and write the category reasoning before any number is assigned. If you want to disagree with the book, those two pages are where the argument actually is.

The book, position by position

Five names are held. Twelve have been closed. Publishing the closed ones is the part that matters, because a track record that shows only current holdings is a survivorship bias machine: sell the mistakes, forget them, and the remaining portfolio looks like genius.

Of the twelve closed positions, six finished up and six finished down. The best was Microsoft at +30.6% over roughly fourteen months. The worst was Novo Nordisk at -22.7%, held for four weeks in late 2025 and cut. A 50% hit rate that still produces outperformance is the ordinary arithmetic of investing: the winners ran longer than the losers were allowed to.

Among the five names still held, Nvidia is up 22.3% since June 2025 and carries the book. Alphabet is down 3.0%, Taiwan Semiconductor is flat at +0.2%, Progressive is up 4.3% since late July 2026 and Blackstone is down 0.7% since late August 2026. Two of the five are less than two months old. The concentration in that first name is not a design choice, it is what happens when one position compounds and the equal weights are only reset monthly.

Did the AI portfolio beat the S&P 500 on a risk-adjusted basis?

No. On raw return it is 12.7 points ahead. On Sharpe ratio the index wins: 1.20 against 1.01, because the book earned its extra return with roughly 55% more volatility and a drawdown nearly twice as deep. That is the honest read of 14 months. A concentrated five-stock book should be more volatile than a 500-stock index, and this one is. If you would not sit through a 17.5% decline without selling, the raw return number is not the one that matters to you.

We publish the Sharpe ratio knowing it makes the book look worse, because the alternative is publishing a return and letting the reader assume the risk was free. Volatility is a poor definition of risk for a long-term owner. Permanent loss of capital is the real one, and no ratio measures it. But volatility is the honest proxy for the question most people actually have, which is whether they could have held this thing.

What happened in the drawdown?

The book peaked on 29 January 2026 and bottomed on 31 March 2026, down 17.5%. The index fell about 9% over roughly the same stretch. By the end of March the book was behind the index for the year to date, and it stayed behind until the spring. It recovered because two positions did the work, not because the process changed. The worst single day was 6 June 2026 at -4.1%; the best was 9 April 2026 at +3.8%.

Look at the chart at the top of this page and the shape of the year is obvious. For seven months the book and the index are the same line. Then a two-month decline that hurts more than the index does, then a step up in April that opens the entire gap. Almost all of the outperformance in this window arrives in a handful of weeks. That is how concentrated books work, and it is exactly why a 14-month sample cannot tell you whether the process is sound.

How do the growth and short-term books compare?

Both started on 18 January 2026, so they get judged against a shorter window. Since that date the value book is +15.0%, the short-term book +17.9%, the growth book +3.3%, and the S&P 500 price index +7.3%. The growth book is behind the index and carries 20.9% volatility with a 21.4% drawdown. The short-term book is ahead of the index but is the wildest of the three at 28.6% volatility. Publishing the laggard alongside the leaders is the point: one book beating the index while another trails is what a small sample looks like.

The growth book is the useful one to stare at. Same infrastructure, same data, same discipline, different weighting of the same categories, and it is behind the index with the second-deepest drawdown of the four lines. Nothing about running an analysis on a machine guarantees the weighting is right. Volatility and drawdown figures for the value book cover its full history from June 2025, which is a longer window than the return column beside them.

Is 14 months long enough to judge an investment process?

No, and it is not close. The standard result in the literature is that separating skill from luck takes decades, not quarters: for a manager running roughly market-level volatility, a track record long enough to make a 5% annual edge statistically significant runs into the tens of years. Ten years of outperformance can still be noise. So treat 14 months as evidence that the process runs, that it exits losers, and that it does not quietly hide them, rather than as evidence that it works. The value of publishing early is the audit trail, not the number at the top.

The arithmetic is unforgiving. If a strategy's true information ratio is 1, you need roughly four years of live results before two standard deviations of statistical confidence, and information ratios of 1 are rare. Research on manager evaluation puts three-year outcomes at overwhelmingly luck-driven, and finds that even a decade of outperformance often fails to clear the bar. A book with five positions makes the sample problem worse, not better, because a handful of large bets carries a large standard error.

So the claim on this page is deliberately small. The book exists, it runs by rule, its exits are dated, its losers are listed, and after 14 months it is ahead of the index on return and behind on risk-adjusted return. That is all the data supports. Ask again in five years.

What do these performance numbers exclude?

Trading commissions, bid-ask spread, slippage and taxes are not modelled. Dividends are not credited, on either side: the benchmark is the S&P 500 price index, so the comparison is like for like but both figures understate a real total return. Positions are marked at the daily close, which is the same end-of-day basis the rest of the app uses. It is a model book, not a brokerage statement, and it is published as one.

One more limitation worth naming: the book trades on end-of-day closes, and the screen it draws from refreshes weekly, so it cannot claim any intraday skill and does not try to. The point of the book was never to be a product. It is a control experiment on the scoring, run in public, with the exits visible. If the score is worth anything, it should show up here eventually, and if it is not, this page is where that becomes obvious first.

The same data drives the panel on the discover page inside the app, so the numbers in this essay and the numbers a subscriber sees are the same numbers, read from the same API on the same day. General investment analysis, not personalised advice.