There's a seductive simplicity to a price chart. Numbers go up, numbers go down, and somewhere in between, the story of a market unfolds. But a price feed, on its own, tells you almost nothing about whether a strategy has a repeatable edge — or whether what looks like an edge is just noise wearing a convincing costume.

This is the problem SDB is designed to solve.

The Gap Between Data and Intelligence

Most traders who approach backtesting for the first time make the same mistake: they treat output as confirmation. A positive net P&L at the end of a test run feels like validation. It isn't. A strategy that made money in a bull market on one symbol during one particular regime has demonstrated exactly one thing — that it made money in that specific context. Without the derived context around that result, you're holding a number with no frame — and no way to know whether deploying that strategy live will compound your account or steadily drain it.

SDB's architecture is built around that frame. Every backtest produces not just performance figures, but a layered set of derived metrics that describe how a strategy behaved, not just whether it produced a positive number.

What Derived Metrics Actually Tell You

Take the relationship between win rate and profit factor — two of the first metrics visible in any SDB summary header.

Win rate alone is nearly useless. A strategy can win 80% of its trades and still lose money if the losing trades are catastrophically larger than the winners. Conversely, a strategy with a 35% win rate can be exceptionally profitable if its winners run long and its losers are cut short. The signal isn't in either number independently — it's in the tension between them.

Profit factor (gross profit divided by gross loss) captures that tension in a single ratio. A profit factor above 1.5 across a meaningful sample of trades suggests the strategy is doing something structurally sound: it's either sizing its winners well relative to losers, or its signal timing is generating asymmetric outcomes. Below 1.0, the conversation is over regardless of win rate — every dollar the strategy generates in gross profit is being outpaced by losses, meaning live deployment would be a controlled liquidation of capital.

SDB surfaces these relationships explicitly so you're never evaluating a metric in isolation.

The Equity Curve as a Behavioral Map

The equity curve in a backtest report is often treated as a visual flourish — a line to glance at before reading the real numbers. That's a mistake. The equity curve is arguably the most information-dense output in the entire report.

A steady upward slope tells you the strategy's edge was reasonably consistent across different market conditions within the test period. A strategy that only trends upward during specific phases and then flatlines — or drops off a ledge — is revealing a regime dependency. It works somewhere, but not everywhere. Deploy it live in the wrong regime, and the historical performance is irrelevant — you will be funding losses with real capital while the conditions the strategy was implicitly tuned to have not yet returned, and may not.

Cliff drops deserve particular attention. A sudden, steep decline in the equity curve usually signals correlated failure: the strategy is breaking down in the same type of conditions every time it breaks down. That's not random variance — it's structural. SDB's drawdown chart complements the equity curve specifically to make this visible, showing not just how deep losses went, but how long the strategy stayed underwater before recovering.

Duration matters as much as depth. A strategy that loses 15% of equity and recovers in two weeks is a very different instrument than one that loses 15% and takes six months to recover. The max drawdown figure is identical. The behavioral reality is not. In that six-month scenario, a live trader faces compounding pressure: reduced position sizing capacity, mounting cost of every session opened in the red, and the real possibility of abandoning the strategy at the bottom of its drawdown — locking in the loss precisely before the recovery occurs.

The Composite Score: Ranking Without Context Collapse

One of the harder problems in strategy research is cross-comparison. If you've run a dozen configurations on the same symbol, how do you rank them without drowning in a spreadsheet of competing metrics?

SDB addresses this with a composite score — a single number that integrates return, consistency, and risk-adjusted performance. The score is designed for relative ranking within a controlled context: same symbol, same interval, different strategy configurations. It gives you a meaningful hierarchy without collapsing the nuance that the full report preserves.

Importantly, composite scores are intentionally not comparable across different symbols. A score on a high-volatility altcoin pair and a score on a major pair are measuring against different volatility environments. Conflating them produces misleading rankings — a configuration could appear superior purely because it operated in a more favorable volatility environment, not because it has a more durable edge, leading a researcher to allocate toward the wrong strategy. SDB makes this explicit, which is the kind of methodological transparency that separates a research tool from a marketing dashboard.

Determinism as a Feature

One aspect of SDB's design that deserves more attention than it typically gets: the backtester is fully deterministic. The same strategy, symbol, interval, and historical dataset will produce identical output every time, without exception.

This might sound obvious, but it isn't universal. Systems that introduce any stochastic element — randomized fill assumptions, jittered entry timing, sampled data — make reproducibility difficult and comparison unreliable. When two configurations produce different results, you need to know the difference came from the strategy, not from the testing infrastructure. If you cannot isolate that variable, you cannot trust that optimizing a parameter is improving anything — you may simply be chasing the favorable end of a random distribution.

Determinism also matters for collaborative work. When a result can be reproduced exactly, it can be shared, scrutinized, and built upon. The output becomes a stable artifact rather than a probabilistic estimate.

The Trade Log as a Forensic Tool

Every SDB backtest includes a full trade log — a row-by-row record of every entry and exit with timestamps, holding periods, and per-trade P&L. This is where pattern recognition shifts from statistical to forensic.

Aggregate metrics describe the strategy's behavior in bulk. The trade log lets you examine when things went wrong. A cluster of consecutive losses concentrated in a specific two-week window isn't statistical noise — it's evidence of a regime dependency. The strategy failed because market conditions changed in a way its rules weren't equipped to handle.

This kind of temporal clustering analysis is something aggregate metrics will never surface. A strategy with 200 trades can absorb a 10-trade losing streak without moving its win rate dramatically. But that streak, examined in the trade log, might map precisely onto a period of macroeconomic turbulence, a liquidity event, or a volatility regime shift. Knowing that changes how you think about deploying the strategy — and when you'd expect it to struggle again. More critically, it gives you a basis for deciding whether to pause or reduce exposure during conditions that resemble that window, rather than holding full position size through a known structural weakness.

Dashboard Signals vs. Raw Output

The distinction between a signal-driven dashboard and a raw performance feed comes down to whether the system is doing interpretive work for you. SDB's derived metrics — the equity curve, drawdown chart, composite score, profit factor relationship, and trade log — aren't decorative. Each one is answering a specific question that raw P&L cannot answer on its own.

The goal isn't to make backtesting feel simpler. It's to make the complexity navigable. A researcher who understands what each derived output is actually measuring can move faster, avoid more mistakes, and build more durable strategies than one who is staring at a net return figure trying to decide if it means something. The cost of that ambiguity isn't abstract — it's capital allocated to strategies that haven't been understood, and risk taken on without knowing what conditions will cause it to materialize.

That's the practical value of signal intelligence over raw data. Not abstraction — precision.