Multi-asset retesting

Why one symbol lies

Any optimizer, given enough freedom, will find parameters that fit the quirks of a single price series: the exact depth of one pullback, the timing of one breakout. Those settings are not wrong on that chart, they are just not about anything repeatable. Running the same settings on other symbols asks the only question that matters: did you find a market behaviour, or did you memorise one chart?

The retest workflow

TradingTune's multi-asset retest is built for exactly this. Once you have applied your best settings, retest them across other symbols without re-optimizing, so every chart sees the identical parameters.

1. 1

   Optimize on your primary symbol

   Run a normal optimization on the chart you care about and apply the best combination so the strategy inputs hold those values.

2. 2

   Pick a representative basket

   Choose several symbols from the same asset class and a few from adjacent ones, mixing leaders and laggards so the basket is not all one kind of market.

3. 3

   Retest without re-optimizing

   Run the applied settings across the basket unchanged. The point is to see how the fixed parameters travel, not to find new ones per symbol.

4. 4

   Read the consistency, not the best case

   Look at how the strategy behaves across the whole basket. Broad, if unspectacular, profitability beats one stellar chart surrounded by losers.

Picking the basket

A good basket is diverse enough to be a real test but related enough that the strategy should plausibly work. For an equity momentum strategy, that might mean several large caps across sectors plus an index. For a crypto strategy, a mix of majors and a couple of higher-beta alts. Avoid stacking the basket with near-identical instruments, which just tests the same thing repeatedly.

basket:
  core:        [NVDA, AAPL, MSFT]   # large-cap leaders
  rotation:    [XOM, JPM, WMT]      # different sectors
  benchmark:   [SPY]                # the market itself
pass_rule: profitable_on >= 5 of 7  # require breadth, not one hero chart

When to reject a strategy

If the settings only work on the symbol they were tuned on and lose elsewhere, reject them, even if that one chart looks spectacular. A strategy that is mildly profitable across most of a diverse basket is far more believable than one that triples on a single name and bleeds on the rest. Multi-asset retesting is a robustness check, and like out-of-sample testing and walk-forward analysis, its whole value is in being willing to act on a bad result.