How to Tell a Real Track Record From a Backtest

What a backtest actually is

A backtest applies a set of rules to historical price data and reports what would have happened. It is a legitimate research tool. It becomes a marketing problem when the results are presented with the same visual language as live performance — the same equity curve, the same percentages — without making clear that no money was at risk.

The core weakness is that the rules were chosen by someone who already knew what the market did. Every parameter — which indicator, what threshold, which period — could be adjusted until the curve looked good. That is curve-fitting, and it is nearly invisible in the output.

The disclosure language that tells you

Regulators in several jurisdictions require hypothetical performance to be labelled. In U.S. commodity trading, CFTC Rule 4.41 is the usual example: phrases along the lines of results being hypothetical, simulated, or not representing actual trading, and a note that they were prepared with the benefit of hindsight.

That last idea is the substantive one. If a disclosure anywhere on the page says results were achieved with hindsight or that no actual trading occurred, the numbers above it are a backtest regardless of how the chart is captioned. Read the smallest text on the page first — it is where the accurate description usually lives.

Signals that results are live

Live records have texture that simulations rarely reproduce. Look for: trades timestamped before publication rather than reconstructed after; a broker or third-party verification statement rather than a self-published spreadsheet; costs actually deducted — commissions, spread, slippage, financing; and drawdowns that are reported rather than smoothed away.

Gaps are a good sign, not a bad one. Real records have periods of inactivity, flat stretches, and losing runs. A curve that rises with unusual consistency across every market regime is describing a model, not an account.

Questions that settle it quickly

Ask directly: was real money at risk in this period, and if so, whose? Are these results net of all costs? Who verified them, and can I see that verification? What was the largest peak-to-trough decline, and when?

A seller with a genuine live record answers these in a sentence each. Deflection, or an answer that shifts to testimonials, is itself the answer. None of this tells you a strategy will work in future — a real track record is evidence about the past, and even a verified one carries no obligation to repeat.

The hybrid records that sit between the two

The clean division between backtest and live record describes the easy cases. A large share of what you will encounter is neither. A common pattern is a chart that runs for several years before the strategy was published and then continues past that date — the early portion is simulated and the later portion is live, spliced into one curve with a single caption. Another is paper trading, sometimes called forward testing, where trades are recorded in real time but no money is committed. That is better evidence than a backtest because the rules were fixed before the outcome was known, and weaker than a live record because it never paid a spread, never got a partial fill, and never had to be held through a real loss.

So ask where on the chart the live portion starts, and judge only that portion. A strategy with nine impressive simulated years and eight mediocre live months has produced eight months of evidence. If the seller cannot or will not identify the boundary date, treat the whole record as hypothetical. CFTC Rule 4.41 is written for commodity-trading results specifically, but the labelling idea is the one that matters here: if the record is simulated or prepared with hindsight, the numbers above it are not a live account.

A short checklist

For any performance chart: find the smallest text on the page and read it first; identify the date live trading began and ignore everything before it; confirm whether costs were deducted and which ones; ask who verified the results and request that verification; locate the largest peak-to-trough decline and when it happened; and check whether the record covers a period containing at least one significant market decline.

That last item does most of the work over time. A record built entirely inside a rising market has not yet been tested against the condition that decides whether a strategy is survivable, and a seller who cannot point to how it behaved in a bad stretch is offering an untested one regardless of how the figures look.

It also helps to write down what would change your mind before you commit anything. A live record gives you a distribution of past outcomes, so decide in advance what size of loss or how long a flat period would tell you the strategy has stopped working for you. Doing that first is the difference between an exit based on evidence and one based on how a bad month feels, and it is the single step most often skipped by people who did all the verification work correctly.