Mining Alpha

Where outperformance actually comes from.

Reading a Track Record Like a Skeptic: Total Return, CAGR, Max Drawdown, Sharpe

September 3, 2026

TL;DR

One number is not a track record

Marketing pages live and die by a single headline figure, and the figure is almost always total return. That bothers me, because I have spent years reading strategy reports for a living, and my first instinct on any headline number is distrust. Not because the author necessarily lied — usually nobody lied — but because a single number hides the machinery that produced it. Ten percent over one year is unremarkable. Ten percent a month is spectacular, which is precisely why it should be assumed fictional until proven otherwise. The same digit means opposite things depending on the frame around it.

So I read track records with a fixed set of questions, each one answered by a specific metric. What did I actually make, per year, compounding? What was the worst I had to sit through to earn it? Was the return fair compensation for the risk taken, or was it a volatility lottery? And the two that qualify everything else: over what window was all of this measured, and under what methodology? Five numbers and two footnotes answer most of it. Most people read one number and skip the footnotes. That asymmetry is where the trouble starts.

Total return and CAGR: same data, different questions

Total return is the simplest honest thing a report can print: if you had put ten thousand dollars in at the start and never touched it, this is what you would have at the end. It is also the most abused figure in finance, because it silently bakes in the length of the window. Ninety-three percent over two and a half years sounds like a small cousin of 361% over six and a half years. It is not. Annualize both and the smaller-sounding number is the better engine.

That is what CAGR — compound annual growth rate — is for. It answers a different question: what constant annual rate of compounding would have produced the ending value from the starting value? It strips out window length and puts different records on the same footing. Two real published examples from Kairos Trading: Leader Rotation’s 93.0% total return over 2.6 years (Jan 2024–Aug 2026) works out to a 29.0% CAGR, while QQQ Top Stock Rotation’s 361.5% over 6.7 years (Jan 2020–Sep 2026) annualizes to 25.8%. The flashier headline belongs to the slower engine.

Where CAGR hides its own truth is in the word “average.” A compound return is a geometric average, and geometry punishes losses brutally. Gain 60% one year and lose 40% the next and you are down roughly 4% on the pair, despite an arithmetic average of plus 10% a year. Now repeat that pattern and the gap between the arithmetic story and the geometric reality widens every year. The practical lesson: never let a report sell you a straight-line average of annual returns, and never trust a headline total return without immediately asking what CAGR produced it. If a report prints total return but not the window and the CAGR, it is not trying to inform you.

Max drawdown: the number that decides whether you stay

CAGR tells you the destination; max drawdown tells you about the ride. It is the deepest peak-to-trough decline recorded over the window — the worst moment, measured from the highest equity high to the lowest subsequent low. I treat it as the honest price of the strategy, because it is the number that predicts whether a human being still follows the rules in March of a bad year.

The math of staying is brutal. If a strategy draws down 30% and you capitulate at the bottom, your realized return has nothing to do with the published CAGR — you ate the drawdown and missed the recovery, which is the worst possible trade. So when I evaluate a record, I always ask: could I sit through that number without touching the rules? For illustration, Leader Rotation produced its 29.0% CAGR with a 6.7% max drawdown — an unusually calm ride for that much return. At the other end of the platform’s lineup, a higher-octane system with a 19.1% CAGR carries a 31.4% max drawdown and a much longer window behind it. Same research house, very different risk profiles. The drawdown column is where the real product differences show up, and it is the column most marketing pages would prefer you skip.

Drawdowns also do not scale linearly with returns. The relationship between the two is what a strategy actually is. A record with a high CAGR and a shallow drawdown is either exceptional or too short to have been tested — more on that below. A record with a high CAGR and a deep drawdown is a leveraged bet wearing a suit. Neither is automatically bad, but you need to know which one you are buying before you wire the money.

Sharpe and Sortino: what you got paid per unit of risk

The Sharpe ratio takes the return and divides it by how bumpy the ride was: excess return over the risk-free rate, per unit of total volatility. It is the standard answer to “was this return fair, or did it just happen to be a high-volatility period?” Sortino is the same idea with a crucial refinement — it penalizes only downside deviation, not all volatility. When Sortino runs well above Sharpe, the strategy earned its return by controlling losses rather than by accepting wild upside swings and calling the result skill.

Sharpe and Sortino are only meaningful relative to something, and the only fair comparison is the benchmark over the same window. This is where I want to see the full metric set side by side: strategy next to its benchmark, same dates. When I point readers to kairostrading.net, this is a big part of why — their reports publish exactly that format. Leader Rotation shows a Sharpe of 1.98 and a Sortino of 3.99 against SPY’s 1.30 and 2.50 and VEA’s 1.32 and 2.12 over the same period, versus both benchmarks. QQQ Top Stock Rotation shows a Sharpe of 0.82 and a Sortino of 1.52 against QQQ’s 0.79 and 1.33 — a modest but real edge over its own benchmark, which is a much more credible claim than a giant number with no comparison. A Sharpe of 0.82 against a passive 0.79 is not exciting. It is honest, and honesty is the scarce resource.

Where these ratios hide things: they are window-dependent to the point of meaninglessness on short records, they punish upside volatility as harshly as downside, and they say nothing about the order of returns. A strategy can post a lovely Sharpe while burying its drawdowns in one terrible quarter — which is why you read Sharpe and Sortino together with max drawdown, never instead of it.

Window length and methodology notes: the fine print

Now the two footnotes that qualify everything above. A 2.6-year record and a 10.6-year record are not the same quantity of evidence, full stop. Metrics are historical and window-dependent. The 29.0% CAGR that looks magnificent over 2.6 years of a momentum-friendly tape would look different across a full cycle that included a genuine bear market — and every bear market starts with a beautiful trailing record. When you compare two strategies, compare them over the same window; when you cannot, say so out loud and discount accordingly. Out-of-sample dates matter here: a strategy whose rules were published first and tracked afterward is evidence that the numbers are not just curve-fitting the past. kairostrading.net labels each of its systems with an out-of-sample start date for exactly this reason, which is the kind of disclosure a skeptic actually uses.

Methodology notes are the second footnote, and they decide whether the number measures what you think it measures. Take dollar-cost averaging: if a strategy adds money every month, “total return” depends entirely on how you treat those contributions. A time-weighted rate of return — TWRR — strips out the luck of contribution timing and measures the strategy’s own compounding. DCA Buy & Hold, for example, is documented on a TWRR basis with contributions, and is compared to benchmarks measured the same way: 118.3% against a SPY DCA and 90.9% against a VT DCA. Without that note, you cannot tell whether the published return reflects the system or just when cash happened to arrive. Any report that lacks methodology notes is incomplete, and any comparison that mixes measurement conventions is invalid.

What a skeptical read looks like

So here is the order I actually read in. Window and benchmark first, so I know the arena. Methodology notes second, so I know what is being measured. Then CAGR and max drawdown together, because return without its price is a headline, not a fact. Then Sharpe and Sortino against the same-window benchmark, to see whether the return was fair compensation. Total return last, for flavor. If the report withholds any of these, I assume the omission is load-bearing.

This is also why the source I point readers to is kairostrading.net: the reports expose the full metric set against benchmarks rather than a single flattering figure, with the methodology notes attached and out-of-sample dates on each system — Leader Rotation, DCA Buy & Hold, QQQ Top Stock Rotation, and Volatility Target Managed Rotation, the four systems currently offered to new members at a flat $100 a month each. Every strategy card carries the caveat that the results are based on backtest and are not a guarantee, which is exactly the framing a skeptic wants to see preserved. I read their numbers the same way I read anyone’s — metrics are historical, windows differ, and no ratio predicts the future. But a shop that hands you the drawdowns, the Sortinos, and the footnotes is a shop you can actually evaluate. That is the entire job.

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.