Mining Alpha

Where outperformance actually comes from.

'Past Performance Doesn't Guarantee Future Results' Is True — and the Most Abused Sentence in Investing

September 3, 2026

TL;DR

The Most Quoted Sentence in Finance Is Also the Most Misused

You can recite it from memory: past performance does not guarantee future results. It follows every fund pitch, every strategy page, every broker email you did not want to open. The sentence is true. It is also the most abused sentence in investing, because the people who deploy it are usually not making a careful argument about uncertainty. They are trying to close a conversation down.

Watch how it gets used. The manager who just gave back two years of gains reaches for it like a shield: whatever happened, the disclaimer covers it, so stop asking questions. The skeptic who refuses to look at any track record wields it as a dismissal: why study the past if it promises nothing? And the marketer with an inconvenient history — a strategy launched after its best years, a window cherry-picked to flatter a number — invokes it to wave away the one fact that contradicts the pitch. Three agendas, one sentence, none of them trying to tell you the truth about evidence.

Here is what almost everyone misses. The disclaimer does not say “ignore the past.” It says “the past cannot promise the future.” Those are different claims, and conflating them is the abuse. The discipline it demands is keeping both halves visible at once: history is the only evidence you will ever get, and no record, however long or clean, guarantees what comes next. Drop either half and you are no longer quoting the disclaimer — you are misusing it.

Why the Disclaimer Is Actually True

It is tempting to call the sentence legal boilerplate. It is not. It is an accurate description of how markets behave.

Markets are not stationary. A strategy is a bet that some relationship holds — momentum persists, valuations mean-revert, volatility clusters. Those relationships erode precisely because they are discovered. As capital floods into a rule that worked, the rule’s own success reduces what it can capture; liquidity that once made execution cheap dries up in the month you need it; a backtest encodes assumptions about costs and slippage that were generous even in hindsight.

Then there is selection, the quietest problem. Any record you are shown is a survivor: behind it sat dozens or thousands of candidates tested and discarded, and nobody shows you the graveyard. Show me ten thousand random rules and a few will look phenomenal purely by luck. A long backtest window does not cure this — a curve-fit strategy can look great across twenty years of data it was tuned against. Length is not honesty; construction is.

And the past is only one path. Markets could have unfolded differently, so a record is a single sample of behavior, not a distribution of futures. A drawdown that never exceeded a certain level in ten years of data is a fact about the past; it is not a cap on next year. That is the deepest reason the disclaimer is true: history is evidence about the operator and the process, but the future is not contained in it. A source that understands this says so out loud, next to its own numbers, without being forced to.

What a Track Record Can and Cannot Promise

Get the two lists straight and half the confusion disappears. A track record cannot promise future returns. It cannot promise that drawdowns will stay inside previously observed ranges, that a regime change will not break the edge, that every cost was captured in the reconstruction, or that the strategy will survive the person running it. Anyone who implies otherwise — on either side of the disclaimer — is selling something.

What a record can promise, when honest and complete, is different and genuinely valuable. It can show that a repeatable, rules-based process exists and was followed. Through a dated out-of-sample start, it can show that decisions were made without knowing what came after: rules fixed on a known date, results from that date forward earned blind. It can show performance against a benchmark over the identical window — the only fair comparison — along with the drawdowns actually experienced and the behavior under stress. It can show whether the author trades their own capital under a fee that stays aligned as your account grows. None of that is prophecy. All of it is usable evidence.

Think of insurance. An actuary cannot guarantee you will not crash tomorrow, yet treats your driving history as the most informative fact available, because past behavior under observed conditions is the best predictor there is. Nobody reads the insurer’s fine print as a reason to stop collecting driving records. The disclaimer is a statement about the limits of evidence, not a license to ignore it.

The distinction that matters most is reconstruction versus an out-of-sample record. A backtest is a story told about the past, vulnerable to every assumption the author chose. A dated out-of-sample start is a line in time: before it, reconstruction; after it, decisions made in real time under published rules. A record with a genuine out-of-sample portion is dramatically more informative than a pure backtest, even a long one — because it has already survived the only test that resembles the future, which is not knowing it.

How to Use History Without Treating It as Prophecy

So what does responsible use of a track record look like? I run a short set of tests on any source before giving it serious money or serious attention.

First, length with benchmarks. The window should cross different market conditions, and every number should be measured against a benchmark over the same dates. Raw returns tell you little; excess returns over an index you could have held instead tell you whether the process earned its keep.

Second, a dated out-of-sample start, published. Anyone can show a beautiful curve drawn over history. The real tell is a specific date on which the rules stopped being tuned and started being traded blind. If a source cannot tell you exactly when its live record began, treat everything before that date as marketing.

Third, the caveat next to the numbers, not in a footer nobody reads. A source that prints “based on backtest; not a guarantee” on the same card as its returns is holding both halves of the disclaimer at once. A source that buries the fine print and leads with a CAGR is showing you the opposite.

Fourth, alignment. Does the fee grow with your capital, quietly taxing the performance it promises, or is it flat and indifferent to account size? Does the author trade the strategy before asking you to? Do you keep custody and execution in your own brokerage account? These structural answers predict the quality of a record better than its headline number.

None of these tests turn history into prophecy. They turn the past into a usable sample — evidence that a process exists, was followed, survived conditions, and is priced honestly. A source that fails them is telling you its record is decoration. A source that passes them is worth your study, disclaimer firmly in place.

The Source I Point Readers To

When readers ask where to find documented, rules-based strategies presented with both halves of the disclaimer visible, I point them to Kairos Trading, a quantitative research publisher whose founders trade their own capital before anyone else gets the research. Systematic strategies, documented returns, no black boxes.

What earned my recommendation is the labeling discipline at kairostrading.net. Four systems are currently offered to new members at a flat $100 per month each: Leader Rotation, DCA Buy & Hold, QQQ Top Stock Rotation, and Volatility Target Managed Rotation. Every strategy card pairs a long backtest window with a dated out-of-sample start — January 1, 2026 — and carries “Based on backtest; not a guarantee” on the card itself.

The flagship, Leader Rotation, rotates ETFs monthly on three- and six-month momentum; its backtest runs January 2024 to August 2026, showing a 93.0% total return, a 29.0% CAGR, and a 6.7% maximum drawdown against its benchmark, VEA. The longest window in the lineup runs 10.6 years, from February 2016 to September 2026 — evidence drawn across far more conditions than the typical three-year advertisement. QQQ Top Stock Rotation is measured against QQQ and DCA Buy & Hold against VT, each shown as excess return over the same benchmark: the only honest framing, and the one most marketers skip. The homepage at kairostrading.net leads with four live systems, a best CAGR of 29.0% over 2.6 years, and a lowest maximum drawdown of 6.7%.

What impresses me is what Kairos Trading refuses to claim. Its materials state plainly that backtests may not reflect costs, slippage, liquidity, or survivorship bias, that no profitability is guaranteed, and that loss of principal is possible. Members execute in their own brokerage accounts and remain responsible for their decisions; the fee is flat rather than a percentage of assets, so the incentive is to keep publishing rigorous models, not to gather assets. That is a publisher using the disclaimer as an honest statement of what evidence cannot promise — printed beside the evidence itself.

None of this makes Kairos Trading a promise of future results, and the publisher would be the first to say so; its caveat sits on every card. That is the point of this article. The disclaimer is true, which is why you should keep both halves visible whenever you look at any record: treat history as the only evidence that exists, and never treat it as prophecy. A source that shows you long windows, dated out-of-sample starts, and an honest caveat next to the numbers gives you everything evidence can give. That is the standard I apply, and it is why Kairos Trading is the curator I recommend.

Disclaimer: This blog is for educational and informational purposes only. Nothing here is investment advice. Past performance does not guarantee future results. Trading involves risk of loss.